diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,31325 @@ +{ + "best_metric": 2.7908167839050293, + "best_model_checkpoint": "runs/gen_Maestro_TSD/TSD_bpe20000/checkpoint-100000", + "epoch": 200.80321285140562, + "global_step": 100000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.04, + "learning_rate": 6.666666666666667e-08, + "loss": 10.007, + "step": 20 + }, + { + "epoch": 0.08, + "learning_rate": 1.3333333333333334e-07, + "loss": 10.0035, + "step": 40 + }, + { + "epoch": 0.12, + "learning_rate": 2.0000000000000002e-07, + "loss": 9.9936, + "step": 60 + }, + { + "epoch": 0.16, + "learning_rate": 2.6666666666666667e-07, + "loss": 9.9802, + "step": 80 + }, + { + "epoch": 0.2, + "learning_rate": 3.3333333333333335e-07, + "loss": 9.9627, + "step": 100 + }, + { + "epoch": 0.24, + "learning_rate": 4.0000000000000003e-07, + "loss": 9.943, + "step": 120 + }, + { + "epoch": 0.28, + "learning_rate": 4.666666666666667e-07, + "loss": 9.9213, + "step": 140 + }, + { + "epoch": 0.32, + "learning_rate": 5.333333333333333e-07, + "loss": 9.9013, + "step": 160 + }, + { + "epoch": 0.36, + "learning_rate": 6.000000000000001e-07, + "loss": 9.8861, + "step": 180 + }, + { + "epoch": 0.4, + "learning_rate": 6.666666666666667e-07, + "loss": 9.8696, + "step": 200 + }, + { + "epoch": 0.44, + "learning_rate": 7.333333333333333e-07, + "loss": 9.8561, + "step": 220 + }, + { + "epoch": 0.48, + "learning_rate": 8.000000000000001e-07, + "loss": 9.8427, + "step": 240 + }, + { + "epoch": 0.52, + "learning_rate": 8.666666666666667e-07, + "loss": 9.8315, + "step": 260 + }, + { + "epoch": 0.56, + "learning_rate": 9.333333333333334e-07, + "loss": 9.8192, + "step": 280 + }, + { + "epoch": 0.6, + "learning_rate": 1.0000000000000002e-06, + "loss": 9.8082, + "step": 300 + }, + { + "epoch": 0.64, + "learning_rate": 1.0666666666666667e-06, + "loss": 9.797, + "step": 320 + }, + { + "epoch": 0.68, + "learning_rate": 1.1333333333333334e-06, + "loss": 9.7852, + "step": 340 + }, + { + "epoch": 0.72, + "learning_rate": 1.2000000000000002e-06, + "loss": 9.7757, + "step": 360 + }, + { + "epoch": 0.76, + "learning_rate": 1.2666666666666667e-06, + "loss": 9.7622, + "step": 380 + }, + { + "epoch": 0.8, + "learning_rate": 1.3333333333333334e-06, + "loss": 9.7501, + "step": 400 + }, + { + "epoch": 0.84, + "learning_rate": 1.4000000000000001e-06, + "loss": 9.7389, + "step": 420 + }, + { + "epoch": 0.88, + "learning_rate": 1.4666666666666667e-06, + "loss": 9.7278, + "step": 440 + }, + { + "epoch": 0.92, + "learning_rate": 1.5333333333333334e-06, + "loss": 9.7139, + "step": 460 + }, + { + "epoch": 0.96, + "learning_rate": 1.6000000000000001e-06, + "loss": 9.6992, + "step": 480 + }, + { + "epoch": 1.0, + "learning_rate": 1.6666666666666667e-06, + "loss": 9.6859, + "step": 500 + }, + { + "epoch": 1.04, + "learning_rate": 1.7333333333333334e-06, + "loss": 9.6706, + "step": 520 + }, + { + "epoch": 1.08, + "learning_rate": 1.8e-06, + "loss": 9.6588, + "step": 540 + }, + { + "epoch": 1.12, + "learning_rate": 1.8666666666666669e-06, + "loss": 9.6432, + "step": 560 + }, + { + "epoch": 1.16, + "learning_rate": 1.9333333333333336e-06, + "loss": 9.625, + "step": 580 + }, + { + "epoch": 1.2, + "learning_rate": 2.0000000000000003e-06, + "loss": 9.6122, + "step": 600 + }, + { + "epoch": 1.24, + "learning_rate": 2.0666666666666666e-06, + "loss": 9.5954, + "step": 620 + }, + { + "epoch": 1.29, + "learning_rate": 2.1333333333333334e-06, + "loss": 9.5791, + "step": 640 + }, + { + "epoch": 1.33, + "learning_rate": 2.2e-06, + "loss": 9.5627, + "step": 660 + }, + { + "epoch": 1.37, + "learning_rate": 2.266666666666667e-06, + "loss": 9.5473, + "step": 680 + }, + { + "epoch": 1.41, + "learning_rate": 2.3333333333333336e-06, + "loss": 9.5294, + "step": 700 + }, + { + "epoch": 1.45, + "learning_rate": 2.4000000000000003e-06, + "loss": 9.5125, + "step": 720 + }, + { + "epoch": 1.49, + "learning_rate": 2.4666666666666666e-06, + "loss": 9.4947, + "step": 740 + }, + { + "epoch": 1.53, + "learning_rate": 2.5333333333333334e-06, + "loss": 9.4789, + "step": 760 + }, + { + "epoch": 1.57, + "learning_rate": 2.6e-06, + "loss": 9.4608, + "step": 780 + }, + { + "epoch": 1.61, + "learning_rate": 2.666666666666667e-06, + "loss": 9.4433, + "step": 800 + }, + { + "epoch": 1.65, + "learning_rate": 2.7333333333333336e-06, + "loss": 9.4253, + "step": 820 + }, + { + "epoch": 1.69, + "learning_rate": 2.8000000000000003e-06, + "loss": 9.4124, + "step": 840 + }, + { + "epoch": 1.73, + "learning_rate": 2.8666666666666666e-06, + "loss": 9.391, + "step": 860 + }, + { + "epoch": 1.77, + "learning_rate": 2.9333333333333333e-06, + "loss": 9.3767, + "step": 880 + }, + { + "epoch": 1.81, + "learning_rate": 3e-06, + "loss": 9.3581, + "step": 900 + }, + { + "epoch": 1.85, + "learning_rate": 3.066666666666667e-06, + "loss": 9.343, + "step": 920 + }, + { + "epoch": 1.89, + "learning_rate": 3.133333333333333e-06, + "loss": 9.3265, + "step": 940 + }, + { + "epoch": 1.93, + "learning_rate": 3.2000000000000003e-06, + "loss": 9.305, + "step": 960 + }, + { + "epoch": 1.97, + "learning_rate": 3.2666666666666666e-06, + "loss": 9.2954, + "step": 980 + }, + { + "epoch": 2.01, + "learning_rate": 3.3333333333333333e-06, + "loss": 9.2791, + "step": 1000 + }, + { + "epoch": 2.01, + "eval_loss": 9.271135330200195, + "eval_runtime": 34.3655, + "eval_samples_per_second": 22.29, + "eval_steps_per_second": 0.116, + "eval_tse_ndup": 2.135672762616435e-05, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.33273534353650863, + "step": 1000 + }, + { + "epoch": 2.05, + "learning_rate": 3.4000000000000005e-06, + "loss": 9.2572, + "step": 1020 + }, + { + "epoch": 2.09, + "learning_rate": 3.466666666666667e-06, + "loss": 9.246, + "step": 1040 + }, + { + "epoch": 2.13, + "learning_rate": 3.5333333333333335e-06, + "loss": 9.2311, + "step": 1060 + }, + { + "epoch": 2.17, + "learning_rate": 3.6e-06, + "loss": 9.2169, + "step": 1080 + }, + { + "epoch": 2.21, + "learning_rate": 3.666666666666667e-06, + "loss": 9.199, + "step": 1100 + }, + { + "epoch": 2.25, + "learning_rate": 3.7333333333333337e-06, + "loss": 9.1862, + "step": 1120 + }, + { + "epoch": 2.29, + "learning_rate": 3.8e-06, + "loss": 9.176, + "step": 1140 + }, + { + "epoch": 2.33, + "learning_rate": 3.866666666666667e-06, + "loss": 9.1608, + "step": 1160 + }, + { + "epoch": 2.37, + "learning_rate": 3.9333333333333335e-06, + "loss": 9.1461, + "step": 1180 + }, + { + "epoch": 2.41, + "learning_rate": 4.000000000000001e-06, + "loss": 9.1415, + "step": 1200 + }, + { + "epoch": 2.45, + "learning_rate": 4.066666666666666e-06, + "loss": 9.1302, + "step": 1220 + }, + { + "epoch": 2.49, + "learning_rate": 4.133333333333333e-06, + "loss": 9.1164, + "step": 1240 + }, + { + "epoch": 2.53, + "learning_rate": 4.2000000000000004e-06, + "loss": 9.1044, + "step": 1260 + }, + { + "epoch": 2.57, + "learning_rate": 4.266666666666667e-06, + "loss": 9.0976, + "step": 1280 + }, + { + "epoch": 2.61, + "learning_rate": 4.333333333333334e-06, + "loss": 9.087, + "step": 1300 + }, + { + "epoch": 2.65, + "learning_rate": 4.4e-06, + "loss": 9.07, + "step": 1320 + }, + { + "epoch": 2.69, + "learning_rate": 4.4666666666666665e-06, + "loss": 9.068, + "step": 1340 + }, + { + "epoch": 2.73, + "learning_rate": 4.533333333333334e-06, + "loss": 9.0556, + "step": 1360 + }, + { + "epoch": 2.77, + "learning_rate": 4.6e-06, + "loss": 9.0503, + "step": 1380 + }, + { + "epoch": 2.81, + "learning_rate": 4.666666666666667e-06, + "loss": 9.0445, + "step": 1400 + }, + { + "epoch": 2.85, + "learning_rate": 4.7333333333333335e-06, + "loss": 9.0324, + "step": 1420 + }, + { + "epoch": 2.89, + "learning_rate": 4.800000000000001e-06, + "loss": 9.0265, + "step": 1440 + }, + { + "epoch": 2.93, + "learning_rate": 4.866666666666667e-06, + "loss": 9.0242, + "step": 1460 + }, + { + "epoch": 2.97, + "learning_rate": 4.933333333333333e-06, + "loss": 9.0193, + "step": 1480 + }, + { + "epoch": 3.01, + "learning_rate": 5e-06, + "loss": 9.0072, + "step": 1500 + }, + { + "epoch": 3.05, + "learning_rate": 5.066666666666667e-06, + "loss": 9.0046, + "step": 1520 + }, + { + "epoch": 3.09, + "learning_rate": 5.133333333333334e-06, + "loss": 8.9979, + "step": 1540 + }, + { + "epoch": 3.13, + "learning_rate": 5.2e-06, + "loss": 8.9953, + "step": 1560 + }, + { + "epoch": 3.17, + "learning_rate": 5.266666666666667e-06, + "loss": 8.9791, + "step": 1580 + }, + { + "epoch": 3.21, + "learning_rate": 5.333333333333334e-06, + "loss": 8.984, + "step": 1600 + }, + { + "epoch": 3.25, + "learning_rate": 5.4e-06, + "loss": 8.9794, + "step": 1620 + }, + { + "epoch": 3.29, + "learning_rate": 5.466666666666667e-06, + "loss": 8.9672, + "step": 1640 + }, + { + "epoch": 3.33, + "learning_rate": 5.5333333333333334e-06, + "loss": 8.9667, + "step": 1660 + }, + { + "epoch": 3.37, + "learning_rate": 5.600000000000001e-06, + "loss": 8.972, + "step": 1680 + }, + { + "epoch": 3.41, + "learning_rate": 5.666666666666667e-06, + "loss": 8.9615, + "step": 1700 + }, + { + "epoch": 3.45, + "learning_rate": 5.733333333333333e-06, + "loss": 8.9488, + "step": 1720 + }, + { + "epoch": 3.49, + "learning_rate": 5.8e-06, + "loss": 8.9495, + "step": 1740 + }, + { + "epoch": 3.53, + "learning_rate": 5.866666666666667e-06, + "loss": 8.949, + "step": 1760 + }, + { + "epoch": 3.57, + "learning_rate": 5.933333333333334e-06, + "loss": 8.9348, + "step": 1780 + }, + { + "epoch": 3.61, + "learning_rate": 6e-06, + "loss": 8.9392, + "step": 1800 + }, + { + "epoch": 3.65, + "learning_rate": 6.066666666666667e-06, + "loss": 8.9174, + "step": 1820 + }, + { + "epoch": 3.69, + "learning_rate": 6.133333333333334e-06, + "loss": 8.9186, + "step": 1840 + }, + { + "epoch": 3.73, + "learning_rate": 6.2e-06, + "loss": 8.9141, + "step": 1860 + }, + { + "epoch": 3.78, + "learning_rate": 6.266666666666666e-06, + "loss": 8.9029, + "step": 1880 + }, + { + "epoch": 3.82, + "learning_rate": 6.333333333333334e-06, + "loss": 8.907, + "step": 1900 + }, + { + "epoch": 3.86, + "learning_rate": 6.4000000000000006e-06, + "loss": 8.8997, + "step": 1920 + }, + { + "epoch": 3.9, + "learning_rate": 6.466666666666667e-06, + "loss": 8.8802, + "step": 1940 + }, + { + "epoch": 3.94, + "learning_rate": 6.533333333333333e-06, + "loss": 8.8711, + "step": 1960 + }, + { + "epoch": 3.98, + "learning_rate": 6.6e-06, + "loss": 8.8621, + "step": 1980 + }, + { + "epoch": 4.02, + "learning_rate": 6.666666666666667e-06, + "loss": 8.8511, + "step": 2000 + }, + { + "epoch": 4.02, + "eval_loss": 8.858022689819336, + "eval_runtime": 32.4666, + "eval_samples_per_second": 23.594, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.055588010043100315, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.2675267970495905, + "step": 2000 + }, + { + "epoch": 4.06, + "learning_rate": 6.733333333333333e-06, + "loss": 8.8408, + "step": 2020 + }, + { + "epoch": 4.1, + "learning_rate": 6.800000000000001e-06, + "loss": 8.8268, + "step": 2040 + }, + { + "epoch": 4.14, + "learning_rate": 6.866666666666667e-06, + "loss": 8.8118, + "step": 2060 + }, + { + "epoch": 4.18, + "learning_rate": 6.933333333333334e-06, + "loss": 8.8077, + "step": 2080 + }, + { + "epoch": 4.22, + "learning_rate": 7.000000000000001e-06, + "loss": 8.784, + "step": 2100 + }, + { + "epoch": 4.26, + "learning_rate": 7.066666666666667e-06, + "loss": 8.7738, + "step": 2120 + }, + { + "epoch": 4.3, + "learning_rate": 7.133333333333333e-06, + "loss": 8.7553, + "step": 2140 + }, + { + "epoch": 4.34, + "learning_rate": 7.2e-06, + "loss": 8.7351, + "step": 2160 + }, + { + "epoch": 4.38, + "learning_rate": 7.266666666666668e-06, + "loss": 8.7188, + "step": 2180 + }, + { + "epoch": 4.42, + "learning_rate": 7.333333333333334e-06, + "loss": 8.7008, + "step": 2200 + }, + { + "epoch": 4.46, + "learning_rate": 7.4e-06, + "loss": 8.6814, + "step": 2220 + }, + { + "epoch": 4.5, + "learning_rate": 7.4666666666666675e-06, + "loss": 8.6593, + "step": 2240 + }, + { + "epoch": 4.54, + "learning_rate": 7.533333333333334e-06, + "loss": 8.6451, + "step": 2260 + }, + { + "epoch": 4.58, + "learning_rate": 7.6e-06, + "loss": 8.6293, + "step": 2280 + }, + { + "epoch": 4.62, + "learning_rate": 7.666666666666667e-06, + "loss": 8.5929, + "step": 2300 + }, + { + "epoch": 4.66, + "learning_rate": 7.733333333333334e-06, + "loss": 8.5899, + "step": 2320 + }, + { + "epoch": 4.7, + "learning_rate": 7.8e-06, + "loss": 8.5626, + "step": 2340 + }, + { + "epoch": 4.74, + "learning_rate": 7.866666666666667e-06, + "loss": 8.5402, + "step": 2360 + }, + { + "epoch": 4.78, + "learning_rate": 7.933333333333334e-06, + "loss": 8.5191, + "step": 2380 + }, + { + "epoch": 4.82, + "learning_rate": 8.000000000000001e-06, + "loss": 8.4982, + "step": 2400 + }, + { + "epoch": 4.86, + "learning_rate": 8.066666666666667e-06, + "loss": 8.4877, + "step": 2420 + }, + { + "epoch": 4.9, + "learning_rate": 8.133333333333332e-06, + "loss": 8.459, + "step": 2440 + }, + { + "epoch": 4.94, + "learning_rate": 8.200000000000001e-06, + "loss": 8.444, + "step": 2460 + }, + { + "epoch": 4.98, + "learning_rate": 8.266666666666667e-06, + "loss": 8.4144, + "step": 2480 + }, + { + "epoch": 5.02, + "learning_rate": 8.333333333333334e-06, + "loss": 8.3985, + "step": 2500 + }, + { + "epoch": 5.06, + "learning_rate": 8.400000000000001e-06, + "loss": 8.3825, + "step": 2520 + }, + { + "epoch": 5.1, + "learning_rate": 8.466666666666666e-06, + "loss": 8.3478, + "step": 2540 + }, + { + "epoch": 5.14, + "learning_rate": 8.533333333333334e-06, + "loss": 8.3196, + "step": 2560 + }, + { + "epoch": 5.18, + "learning_rate": 8.599999999999999e-06, + "loss": 8.3193, + "step": 2580 + }, + { + "epoch": 5.22, + "learning_rate": 8.666666666666668e-06, + "loss": 8.2958, + "step": 2600 + }, + { + "epoch": 5.26, + "learning_rate": 8.733333333333333e-06, + "loss": 8.2651, + "step": 2620 + }, + { + "epoch": 5.3, + "learning_rate": 8.8e-06, + "loss": 8.264, + "step": 2640 + }, + { + "epoch": 5.34, + "learning_rate": 8.866666666666668e-06, + "loss": 8.2275, + "step": 2660 + }, + { + "epoch": 5.38, + "learning_rate": 8.933333333333333e-06, + "loss": 8.2075, + "step": 2680 + }, + { + "epoch": 5.42, + "learning_rate": 9e-06, + "loss": 8.1864, + "step": 2700 + }, + { + "epoch": 5.46, + "learning_rate": 9.066666666666667e-06, + "loss": 8.1827, + "step": 2720 + }, + { + "epoch": 5.5, + "learning_rate": 9.133333333333335e-06, + "loss": 8.1518, + "step": 2740 + }, + { + "epoch": 5.54, + "learning_rate": 9.2e-06, + "loss": 8.1312, + "step": 2760 + }, + { + "epoch": 5.58, + "learning_rate": 9.266666666666667e-06, + "loss": 8.1162, + "step": 2780 + }, + { + "epoch": 5.62, + "learning_rate": 9.333333333333334e-06, + "loss": 8.0976, + "step": 2800 + }, + { + "epoch": 5.66, + "learning_rate": 9.4e-06, + "loss": 8.0745, + "step": 2820 + }, + { + "epoch": 5.7, + "learning_rate": 9.466666666666667e-06, + "loss": 8.0655, + "step": 2840 + }, + { + "epoch": 5.74, + "learning_rate": 9.533333333333334e-06, + "loss": 8.0412, + "step": 2860 + }, + { + "epoch": 5.78, + "learning_rate": 9.600000000000001e-06, + "loss": 8.022, + "step": 2880 + }, + { + "epoch": 5.82, + "learning_rate": 9.666666666666667e-06, + "loss": 8.0064, + "step": 2900 + }, + { + "epoch": 5.86, + "learning_rate": 9.733333333333334e-06, + "loss": 7.9822, + "step": 2920 + }, + { + "epoch": 5.9, + "learning_rate": 9.800000000000001e-06, + "loss": 7.9796, + "step": 2940 + }, + { + "epoch": 5.94, + "learning_rate": 9.866666666666667e-06, + "loss": 7.9593, + "step": 2960 + }, + { + "epoch": 5.98, + "learning_rate": 9.933333333333334e-06, + "loss": 7.9523, + "step": 2980 + }, + { + "epoch": 6.02, + "learning_rate": 1e-05, + "loss": 7.9207, + "step": 3000 + }, + { + "epoch": 6.02, + "eval_loss": 7.949766159057617, + "eval_runtime": 34.1198, + "eval_samples_per_second": 22.45, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.11260286289788565, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0028498073301391476, + "step": 3000 + }, + { + "epoch": 6.06, + "learning_rate": 1.0066666666666668e-05, + "loss": 7.9043, + "step": 3020 + }, + { + "epoch": 6.1, + "learning_rate": 1.0133333333333333e-05, + "loss": 7.896, + "step": 3040 + }, + { + "epoch": 6.14, + "learning_rate": 1.02e-05, + "loss": 7.8749, + "step": 3060 + }, + { + "epoch": 6.18, + "learning_rate": 1.0266666666666668e-05, + "loss": 7.8751, + "step": 3080 + }, + { + "epoch": 6.22, + "learning_rate": 1.0333333333333333e-05, + "loss": 7.8486, + "step": 3100 + }, + { + "epoch": 6.27, + "learning_rate": 1.04e-05, + "loss": 7.8353, + "step": 3120 + }, + { + "epoch": 6.31, + "learning_rate": 1.0466666666666668e-05, + "loss": 7.8187, + "step": 3140 + }, + { + "epoch": 6.35, + "learning_rate": 1.0533333333333335e-05, + "loss": 7.8069, + "step": 3160 + }, + { + "epoch": 6.39, + "learning_rate": 1.06e-05, + "loss": 7.7952, + "step": 3180 + }, + { + "epoch": 6.43, + "learning_rate": 1.0666666666666667e-05, + "loss": 7.7851, + "step": 3200 + }, + { + "epoch": 6.47, + "learning_rate": 1.0733333333333334e-05, + "loss": 7.7618, + "step": 3220 + }, + { + "epoch": 6.51, + "learning_rate": 1.08e-05, + "loss": 7.7374, + "step": 3240 + }, + { + "epoch": 6.55, + "learning_rate": 1.0866666666666667e-05, + "loss": 7.7399, + "step": 3260 + }, + { + "epoch": 6.59, + "learning_rate": 1.0933333333333334e-05, + "loss": 7.7154, + "step": 3280 + }, + { + "epoch": 6.63, + "learning_rate": 1.1000000000000001e-05, + "loss": 7.6959, + "step": 3300 + }, + { + "epoch": 6.67, + "learning_rate": 1.1066666666666667e-05, + "loss": 7.6788, + "step": 3320 + }, + { + "epoch": 6.71, + "learning_rate": 1.1133333333333334e-05, + "loss": 7.6717, + "step": 3340 + }, + { + "epoch": 6.75, + "learning_rate": 1.1200000000000001e-05, + "loss": 7.6695, + "step": 3360 + }, + { + "epoch": 6.79, + "learning_rate": 1.1266666666666667e-05, + "loss": 7.6427, + "step": 3380 + }, + { + "epoch": 6.83, + "learning_rate": 1.1333333333333334e-05, + "loss": 7.6214, + "step": 3400 + }, + { + "epoch": 6.87, + "learning_rate": 1.1400000000000001e-05, + "loss": 7.6203, + "step": 3420 + }, + { + "epoch": 6.91, + "learning_rate": 1.1466666666666666e-05, + "loss": 7.5949, + "step": 3440 + }, + { + "epoch": 6.95, + "learning_rate": 1.1533333333333334e-05, + "loss": 7.5903, + "step": 3460 + }, + { + "epoch": 6.99, + "learning_rate": 1.16e-05, + "loss": 7.576, + "step": 3480 + }, + { + "epoch": 7.03, + "learning_rate": 1.1666666666666668e-05, + "loss": 7.5624, + "step": 3500 + }, + { + "epoch": 7.07, + "learning_rate": 1.1733333333333333e-05, + "loss": 7.5486, + "step": 3520 + }, + { + "epoch": 7.11, + "learning_rate": 1.18e-05, + "loss": 7.5387, + "step": 3540 + }, + { + "epoch": 7.15, + "learning_rate": 1.1866666666666668e-05, + "loss": 7.5188, + "step": 3560 + }, + { + "epoch": 7.19, + "learning_rate": 1.1933333333333333e-05, + "loss": 7.5127, + "step": 3580 + }, + { + "epoch": 7.23, + "learning_rate": 1.2e-05, + "loss": 7.5077, + "step": 3600 + }, + { + "epoch": 7.27, + "learning_rate": 1.2066666666666667e-05, + "loss": 7.4777, + "step": 3620 + }, + { + "epoch": 7.31, + "learning_rate": 1.2133333333333335e-05, + "loss": 7.4592, + "step": 3640 + }, + { + "epoch": 7.35, + "learning_rate": 1.22e-05, + "loss": 7.4621, + "step": 3660 + }, + { + "epoch": 7.39, + "learning_rate": 1.2266666666666667e-05, + "loss": 7.4544, + "step": 3680 + }, + { + "epoch": 7.43, + "learning_rate": 1.2333333333333334e-05, + "loss": 7.4348, + "step": 3700 + }, + { + "epoch": 7.47, + "learning_rate": 1.24e-05, + "loss": 7.4206, + "step": 3720 + }, + { + "epoch": 7.51, + "learning_rate": 1.2466666666666667e-05, + "loss": 7.4139, + "step": 3740 + }, + { + "epoch": 7.55, + "learning_rate": 1.2533333333333332e-05, + "loss": 7.3757, + "step": 3760 + }, + { + "epoch": 7.59, + "learning_rate": 1.2600000000000001e-05, + "loss": 7.3812, + "step": 3780 + }, + { + "epoch": 7.63, + "learning_rate": 1.2666666666666668e-05, + "loss": 7.3661, + "step": 3800 + }, + { + "epoch": 7.67, + "learning_rate": 1.2733333333333334e-05, + "loss": 7.3556, + "step": 3820 + }, + { + "epoch": 7.71, + "learning_rate": 1.2800000000000001e-05, + "loss": 7.3463, + "step": 3840 + }, + { + "epoch": 7.75, + "learning_rate": 1.2866666666666668e-05, + "loss": 7.3223, + "step": 3860 + }, + { + "epoch": 7.79, + "learning_rate": 1.2933333333333334e-05, + "loss": 7.3261, + "step": 3880 + }, + { + "epoch": 7.83, + "learning_rate": 1.3000000000000001e-05, + "loss": 7.2937, + "step": 3900 + }, + { + "epoch": 7.87, + "learning_rate": 1.3066666666666666e-05, + "loss": 7.2999, + "step": 3920 + }, + { + "epoch": 7.91, + "learning_rate": 1.3133333333333334e-05, + "loss": 7.2829, + "step": 3940 + }, + { + "epoch": 7.95, + "learning_rate": 1.32e-05, + "loss": 7.2797, + "step": 3960 + }, + { + "epoch": 7.99, + "learning_rate": 1.3266666666666666e-05, + "loss": 7.2448, + "step": 3980 + }, + { + "epoch": 8.03, + "learning_rate": 1.3333333333333333e-05, + "loss": 7.2514, + "step": 4000 + }, + { + "epoch": 8.03, + "eval_loss": 7.271445274353027, + "eval_runtime": 32.3137, + "eval_samples_per_second": 23.705, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.05492504538485124, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00013423585095570596, + "step": 4000 + }, + { + "epoch": 8.07, + "learning_rate": 1.3400000000000002e-05, + "loss": 7.2469, + "step": 4020 + }, + { + "epoch": 8.11, + "learning_rate": 1.3466666666666666e-05, + "loss": 7.2316, + "step": 4040 + }, + { + "epoch": 8.15, + "learning_rate": 1.3533333333333335e-05, + "loss": 7.2147, + "step": 4060 + }, + { + "epoch": 8.19, + "learning_rate": 1.3600000000000002e-05, + "loss": 7.2097, + "step": 4080 + }, + { + "epoch": 8.23, + "learning_rate": 1.3666666666666666e-05, + "loss": 7.199, + "step": 4100 + }, + { + "epoch": 8.27, + "learning_rate": 1.3733333333333335e-05, + "loss": 7.1869, + "step": 4120 + }, + { + "epoch": 8.31, + "learning_rate": 1.3800000000000002e-05, + "loss": 7.1685, + "step": 4140 + }, + { + "epoch": 8.35, + "learning_rate": 1.3866666666666667e-05, + "loss": 7.1544, + "step": 4160 + }, + { + "epoch": 8.39, + "learning_rate": 1.3933333333333334e-05, + "loss": 7.1471, + "step": 4180 + }, + { + "epoch": 8.43, + "learning_rate": 1.4000000000000001e-05, + "loss": 7.1336, + "step": 4200 + }, + { + "epoch": 8.47, + "learning_rate": 1.4066666666666667e-05, + "loss": 7.1346, + "step": 4220 + }, + { + "epoch": 8.51, + "learning_rate": 1.4133333333333334e-05, + "loss": 7.1317, + "step": 4240 + }, + { + "epoch": 8.55, + "learning_rate": 1.42e-05, + "loss": 7.1226, + "step": 4260 + }, + { + "epoch": 8.59, + "learning_rate": 1.4266666666666667e-05, + "loss": 7.1089, + "step": 4280 + }, + { + "epoch": 8.63, + "learning_rate": 1.4333333333333334e-05, + "loss": 7.0861, + "step": 4300 + }, + { + "epoch": 8.67, + "learning_rate": 1.44e-05, + "loss": 7.0828, + "step": 4320 + }, + { + "epoch": 8.71, + "learning_rate": 1.4466666666666667e-05, + "loss": 7.0804, + "step": 4340 + }, + { + "epoch": 8.76, + "learning_rate": 1.4533333333333335e-05, + "loss": 7.0738, + "step": 4360 + }, + { + "epoch": 8.8, + "learning_rate": 1.4599999999999999e-05, + "loss": 7.05, + "step": 4380 + }, + { + "epoch": 8.84, + "learning_rate": 1.4666666666666668e-05, + "loss": 7.0413, + "step": 4400 + }, + { + "epoch": 8.88, + "learning_rate": 1.4733333333333335e-05, + "loss": 7.0399, + "step": 4420 + }, + { + "epoch": 8.92, + "learning_rate": 1.48e-05, + "loss": 7.0155, + "step": 4440 + }, + { + "epoch": 8.96, + "learning_rate": 1.4866666666666668e-05, + "loss": 7.0122, + "step": 4460 + }, + { + "epoch": 9.0, + "learning_rate": 1.4933333333333335e-05, + "loss": 7.005, + "step": 4480 + }, + { + "epoch": 9.04, + "learning_rate": 1.5e-05, + "loss": 6.9909, + "step": 4500 + }, + { + "epoch": 9.08, + "learning_rate": 1.5066666666666668e-05, + "loss": 6.9896, + "step": 4520 + }, + { + "epoch": 9.12, + "learning_rate": 1.5133333333333333e-05, + "loss": 6.9681, + "step": 4540 + }, + { + "epoch": 9.16, + "learning_rate": 1.52e-05, + "loss": 6.9632, + "step": 4560 + }, + { + "epoch": 9.2, + "learning_rate": 1.5266666666666667e-05, + "loss": 6.9717, + "step": 4580 + }, + { + "epoch": 9.24, + "learning_rate": 1.5333333333333334e-05, + "loss": 6.9525, + "step": 4600 + }, + { + "epoch": 9.28, + "learning_rate": 1.54e-05, + "loss": 6.9502, + "step": 4620 + }, + { + "epoch": 9.32, + "learning_rate": 1.546666666666667e-05, + "loss": 6.9311, + "step": 4640 + }, + { + "epoch": 9.36, + "learning_rate": 1.5533333333333333e-05, + "loss": 6.9127, + "step": 4660 + }, + { + "epoch": 9.4, + "learning_rate": 1.56e-05, + "loss": 6.911, + "step": 4680 + }, + { + "epoch": 9.44, + "learning_rate": 1.5666666666666667e-05, + "loss": 6.9102, + "step": 4700 + }, + { + "epoch": 9.48, + "learning_rate": 1.5733333333333334e-05, + "loss": 6.8915, + "step": 4720 + }, + { + "epoch": 9.52, + "learning_rate": 1.58e-05, + "loss": 6.8868, + "step": 4740 + }, + { + "epoch": 9.56, + "learning_rate": 1.586666666666667e-05, + "loss": 6.8721, + "step": 4760 + }, + { + "epoch": 9.6, + "learning_rate": 1.5933333333333332e-05, + "loss": 6.8767, + "step": 4780 + }, + { + "epoch": 9.64, + "learning_rate": 1.6000000000000003e-05, + "loss": 6.8667, + "step": 4800 + }, + { + "epoch": 9.68, + "learning_rate": 1.606666666666667e-05, + "loss": 6.8612, + "step": 4820 + }, + { + "epoch": 9.72, + "learning_rate": 1.6133333333333334e-05, + "loss": 6.8461, + "step": 4840 + }, + { + "epoch": 9.76, + "learning_rate": 1.62e-05, + "loss": 6.8405, + "step": 4860 + }, + { + "epoch": 9.8, + "learning_rate": 1.6266666666666665e-05, + "loss": 6.8375, + "step": 4880 + }, + { + "epoch": 9.84, + "learning_rate": 1.6333333333333335e-05, + "loss": 6.8332, + "step": 4900 + }, + { + "epoch": 9.88, + "learning_rate": 1.6400000000000002e-05, + "loss": 6.8248, + "step": 4920 + }, + { + "epoch": 9.92, + "learning_rate": 1.6466666666666666e-05, + "loss": 6.8068, + "step": 4940 + }, + { + "epoch": 9.96, + "learning_rate": 1.6533333333333333e-05, + "loss": 6.7985, + "step": 4960 + }, + { + "epoch": 10.0, + "learning_rate": 1.66e-05, + "loss": 6.7957, + "step": 4980 + }, + { + "epoch": 10.04, + "learning_rate": 1.6666666666666667e-05, + "loss": 6.7711, + "step": 5000 + }, + { + "epoch": 10.04, + "eval_loss": 6.791956424713135, + "eval_runtime": 32.5787, + "eval_samples_per_second": 23.512, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.023346051818536433, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.17856106759501e-05, + "step": 5000 + }, + { + "epoch": 10.08, + "learning_rate": 1.6733333333333335e-05, + "loss": 6.7715, + "step": 5020 + }, + { + "epoch": 10.12, + "learning_rate": 1.6800000000000002e-05, + "loss": 6.7576, + "step": 5040 + }, + { + "epoch": 10.16, + "learning_rate": 1.6866666666666666e-05, + "loss": 6.7582, + "step": 5060 + }, + { + "epoch": 10.2, + "learning_rate": 1.6933333333333333e-05, + "loss": 6.7386, + "step": 5080 + }, + { + "epoch": 10.24, + "learning_rate": 1.7000000000000003e-05, + "loss": 6.7476, + "step": 5100 + }, + { + "epoch": 10.28, + "learning_rate": 1.7066666666666667e-05, + "loss": 6.7374, + "step": 5120 + }, + { + "epoch": 10.32, + "learning_rate": 1.7133333333333334e-05, + "loss": 6.7335, + "step": 5140 + }, + { + "epoch": 10.36, + "learning_rate": 1.7199999999999998e-05, + "loss": 6.719, + "step": 5160 + }, + { + "epoch": 10.4, + "learning_rate": 1.726666666666667e-05, + "loss": 6.7151, + "step": 5180 + }, + { + "epoch": 10.44, + "learning_rate": 1.7333333333333336e-05, + "loss": 6.7125, + "step": 5200 + }, + { + "epoch": 10.48, + "learning_rate": 1.74e-05, + "loss": 6.6928, + "step": 5220 + }, + { + "epoch": 10.52, + "learning_rate": 1.7466666666666667e-05, + "loss": 6.6889, + "step": 5240 + }, + { + "epoch": 10.56, + "learning_rate": 1.7533333333333334e-05, + "loss": 6.6804, + "step": 5260 + }, + { + "epoch": 10.6, + "learning_rate": 1.76e-05, + "loss": 6.6739, + "step": 5280 + }, + { + "epoch": 10.64, + "learning_rate": 1.7666666666666668e-05, + "loss": 6.6544, + "step": 5300 + }, + { + "epoch": 10.68, + "learning_rate": 1.7733333333333335e-05, + "loss": 6.656, + "step": 5320 + }, + { + "epoch": 10.72, + "learning_rate": 1.78e-05, + "loss": 6.6644, + "step": 5340 + }, + { + "epoch": 10.76, + "learning_rate": 1.7866666666666666e-05, + "loss": 6.6394, + "step": 5360 + }, + { + "epoch": 10.8, + "learning_rate": 1.7933333333333337e-05, + "loss": 6.6474, + "step": 5380 + }, + { + "epoch": 10.84, + "learning_rate": 1.8e-05, + "loss": 6.6374, + "step": 5400 + }, + { + "epoch": 10.88, + "learning_rate": 1.8066666666666668e-05, + "loss": 6.6198, + "step": 5420 + }, + { + "epoch": 10.92, + "learning_rate": 1.8133333333333335e-05, + "loss": 6.6106, + "step": 5440 + }, + { + "epoch": 10.96, + "learning_rate": 1.8200000000000002e-05, + "loss": 6.6075, + "step": 5460 + }, + { + "epoch": 11.0, + "learning_rate": 1.826666666666667e-05, + "loss": 6.605, + "step": 5480 + }, + { + "epoch": 11.04, + "learning_rate": 1.8333333333333333e-05, + "loss": 6.5842, + "step": 5500 + }, + { + "epoch": 11.08, + "learning_rate": 1.84e-05, + "loss": 6.5814, + "step": 5520 + }, + { + "epoch": 11.12, + "learning_rate": 1.8466666666666667e-05, + "loss": 6.5772, + "step": 5540 + }, + { + "epoch": 11.16, + "learning_rate": 1.8533333333333334e-05, + "loss": 6.5732, + "step": 5560 + }, + { + "epoch": 11.2, + "learning_rate": 1.86e-05, + "loss": 6.5577, + "step": 5580 + }, + { + "epoch": 11.24, + "learning_rate": 1.866666666666667e-05, + "loss": 6.5648, + "step": 5600 + }, + { + "epoch": 11.29, + "learning_rate": 1.8733333333333332e-05, + "loss": 6.5333, + "step": 5620 + }, + { + "epoch": 11.33, + "learning_rate": 1.88e-05, + "loss": 6.5348, + "step": 5640 + }, + { + "epoch": 11.37, + "learning_rate": 1.886666666666667e-05, + "loss": 6.5452, + "step": 5660 + }, + { + "epoch": 11.41, + "learning_rate": 1.8933333333333334e-05, + "loss": 6.5267, + "step": 5680 + }, + { + "epoch": 11.45, + "learning_rate": 1.9e-05, + "loss": 6.5187, + "step": 5700 + }, + { + "epoch": 11.49, + "learning_rate": 1.9066666666666668e-05, + "loss": 6.5159, + "step": 5720 + }, + { + "epoch": 11.53, + "learning_rate": 1.9133333333333332e-05, + "loss": 6.4981, + "step": 5740 + }, + { + "epoch": 11.57, + "learning_rate": 1.9200000000000003e-05, + "loss": 6.4817, + "step": 5760 + }, + { + "epoch": 11.61, + "learning_rate": 1.926666666666667e-05, + "loss": 6.4986, + "step": 5780 + }, + { + "epoch": 11.65, + "learning_rate": 1.9333333333333333e-05, + "loss": 6.4889, + "step": 5800 + }, + { + "epoch": 11.69, + "learning_rate": 1.94e-05, + "loss": 6.4897, + "step": 5820 + }, + { + "epoch": 11.73, + "learning_rate": 1.9466666666666668e-05, + "loss": 6.4664, + "step": 5840 + }, + { + "epoch": 11.77, + "learning_rate": 1.9533333333333335e-05, + "loss": 6.4897, + "step": 5860 + }, + { + "epoch": 11.81, + "learning_rate": 1.9600000000000002e-05, + "loss": 6.4551, + "step": 5880 + }, + { + "epoch": 11.85, + "learning_rate": 1.9666666666666666e-05, + "loss": 6.4572, + "step": 5900 + }, + { + "epoch": 11.89, + "learning_rate": 1.9733333333333333e-05, + "loss": 6.4442, + "step": 5920 + }, + { + "epoch": 11.93, + "learning_rate": 1.9800000000000004e-05, + "loss": 6.447, + "step": 5940 + }, + { + "epoch": 11.97, + "learning_rate": 1.9866666666666667e-05, + "loss": 6.4437, + "step": 5960 + }, + { + "epoch": 12.01, + "learning_rate": 1.9933333333333334e-05, + "loss": 6.4317, + "step": 5980 + }, + { + "epoch": 12.05, + "learning_rate": 2e-05, + "loss": 6.4122, + "step": 6000 + }, + { + "epoch": 12.05, + "eval_loss": 6.401874542236328, + "eval_runtime": 34.1711, + "eval_samples_per_second": 22.417, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.014522831829404562, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.2664635915288658e-06, + "step": 6000 + }, + { + "epoch": 12.09, + "learning_rate": 2.0063333333333335e-05, + "loss": 6.4102, + "step": 6020 + }, + { + "epoch": 12.13, + "learning_rate": 2.0130000000000002e-05, + "loss": 6.3864, + "step": 6040 + }, + { + "epoch": 12.17, + "learning_rate": 2.0196666666666666e-05, + "loss": 6.3975, + "step": 6060 + }, + { + "epoch": 12.21, + "learning_rate": 2.0263333333333333e-05, + "loss": 6.3919, + "step": 6080 + }, + { + "epoch": 12.25, + "learning_rate": 2.033e-05, + "loss": 6.37, + "step": 6100 + }, + { + "epoch": 12.29, + "learning_rate": 2.0396666666666667e-05, + "loss": 6.3731, + "step": 6120 + }, + { + "epoch": 12.33, + "learning_rate": 2.0463333333333334e-05, + "loss": 6.3698, + "step": 6140 + }, + { + "epoch": 12.37, + "learning_rate": 2.053e-05, + "loss": 6.3602, + "step": 6160 + }, + { + "epoch": 12.41, + "learning_rate": 2.0596666666666665e-05, + "loss": 6.3681, + "step": 6180 + }, + { + "epoch": 12.45, + "learning_rate": 2.0663333333333336e-05, + "loss": 6.3438, + "step": 6200 + }, + { + "epoch": 12.49, + "learning_rate": 2.0730000000000003e-05, + "loss": 6.3508, + "step": 6220 + }, + { + "epoch": 12.53, + "learning_rate": 2.0796666666666667e-05, + "loss": 6.3494, + "step": 6240 + }, + { + "epoch": 12.57, + "learning_rate": 2.0863333333333334e-05, + "loss": 6.3304, + "step": 6260 + }, + { + "epoch": 12.61, + "learning_rate": 2.093e-05, + "loss": 6.3397, + "step": 6280 + }, + { + "epoch": 12.65, + "learning_rate": 2.099666666666667e-05, + "loss": 6.3086, + "step": 6300 + }, + { + "epoch": 12.69, + "learning_rate": 2.1063333333333336e-05, + "loss": 6.3319, + "step": 6320 + }, + { + "epoch": 12.73, + "learning_rate": 2.113e-05, + "loss": 6.3216, + "step": 6340 + }, + { + "epoch": 12.77, + "learning_rate": 2.1196666666666666e-05, + "loss": 6.3067, + "step": 6360 + }, + { + "epoch": 12.81, + "learning_rate": 2.1263333333333334e-05, + "loss": 6.3092, + "step": 6380 + }, + { + "epoch": 12.85, + "learning_rate": 2.133e-05, + "loss": 6.3004, + "step": 6400 + }, + { + "epoch": 12.89, + "learning_rate": 2.1396666666666668e-05, + "loss": 6.2775, + "step": 6420 + }, + { + "epoch": 12.93, + "learning_rate": 2.1463333333333335e-05, + "loss": 6.2868, + "step": 6440 + }, + { + "epoch": 12.97, + "learning_rate": 2.153e-05, + "loss": 6.2839, + "step": 6460 + }, + { + "epoch": 13.01, + "learning_rate": 2.159666666666667e-05, + "loss": 6.2552, + "step": 6480 + }, + { + "epoch": 13.05, + "learning_rate": 2.1663333333333337e-05, + "loss": 6.2578, + "step": 6500 + }, + { + "epoch": 13.09, + "learning_rate": 2.173e-05, + "loss": 6.2433, + "step": 6520 + }, + { + "epoch": 13.13, + "learning_rate": 2.1796666666666667e-05, + "loss": 6.2402, + "step": 6540 + }, + { + "epoch": 13.17, + "learning_rate": 2.1863333333333335e-05, + "loss": 6.2387, + "step": 6560 + }, + { + "epoch": 13.21, + "learning_rate": 2.1930000000000002e-05, + "loss": 6.2512, + "step": 6580 + }, + { + "epoch": 13.25, + "learning_rate": 2.199666666666667e-05, + "loss": 6.2283, + "step": 6600 + }, + { + "epoch": 13.29, + "learning_rate": 2.2063333333333333e-05, + "loss": 6.2113, + "step": 6620 + }, + { + "epoch": 13.33, + "learning_rate": 2.213e-05, + "loss": 6.2176, + "step": 6640 + }, + { + "epoch": 13.37, + "learning_rate": 2.2196666666666667e-05, + "loss": 6.2046, + "step": 6660 + }, + { + "epoch": 13.41, + "learning_rate": 2.2263333333333334e-05, + "loss": 6.214, + "step": 6680 + }, + { + "epoch": 13.45, + "learning_rate": 2.233e-05, + "loss": 6.1869, + "step": 6700 + }, + { + "epoch": 13.49, + "learning_rate": 2.239666666666667e-05, + "loss": 6.1905, + "step": 6720 + }, + { + "epoch": 13.53, + "learning_rate": 2.2463333333333332e-05, + "loss": 6.1955, + "step": 6740 + }, + { + "epoch": 13.57, + "learning_rate": 2.253e-05, + "loss": 6.1833, + "step": 6760 + }, + { + "epoch": 13.61, + "learning_rate": 2.259666666666667e-05, + "loss": 6.1788, + "step": 6780 + }, + { + "epoch": 13.65, + "learning_rate": 2.2663333333333334e-05, + "loss": 6.1745, + "step": 6800 + }, + { + "epoch": 13.69, + "learning_rate": 2.273e-05, + "loss": 6.1734, + "step": 6820 + }, + { + "epoch": 13.73, + "learning_rate": 2.2796666666666668e-05, + "loss": 6.1549, + "step": 6840 + }, + { + "epoch": 13.78, + "learning_rate": 2.2863333333333335e-05, + "loss": 6.1648, + "step": 6860 + }, + { + "epoch": 13.82, + "learning_rate": 2.2930000000000002e-05, + "loss": 6.1589, + "step": 6880 + }, + { + "epoch": 13.86, + "learning_rate": 2.299666666666667e-05, + "loss": 6.1347, + "step": 6900 + }, + { + "epoch": 13.9, + "learning_rate": 2.3063333333333333e-05, + "loss": 6.1399, + "step": 6920 + }, + { + "epoch": 13.94, + "learning_rate": 2.313e-05, + "loss": 6.1318, + "step": 6940 + }, + { + "epoch": 13.98, + "learning_rate": 2.3196666666666668e-05, + "loss": 6.1175, + "step": 6960 + }, + { + "epoch": 14.02, + "learning_rate": 2.3263333333333335e-05, + "loss": 6.1187, + "step": 6980 + }, + { + "epoch": 14.06, + "learning_rate": 2.3330000000000002e-05, + "loss": 6.1159, + "step": 7000 + }, + { + "epoch": 14.06, + "eval_loss": 6.070368766784668, + "eval_runtime": 32.3616, + "eval_samples_per_second": 23.67, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.017255196154812854, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.2664635915288658e-06, + "step": 7000 + }, + { + "epoch": 14.1, + "learning_rate": 2.3396666666666666e-05, + "loss": 6.1051, + "step": 7020 + }, + { + "epoch": 14.14, + "learning_rate": 2.3463333333333333e-05, + "loss": 6.0892, + "step": 7040 + }, + { + "epoch": 14.18, + "learning_rate": 2.3530000000000003e-05, + "loss": 6.0843, + "step": 7060 + }, + { + "epoch": 14.22, + "learning_rate": 2.3596666666666667e-05, + "loss": 6.0838, + "step": 7080 + }, + { + "epoch": 14.26, + "learning_rate": 2.3663333333333334e-05, + "loss": 6.0847, + "step": 7100 + }, + { + "epoch": 14.3, + "learning_rate": 2.373e-05, + "loss": 6.0705, + "step": 7120 + }, + { + "epoch": 14.34, + "learning_rate": 2.379666666666667e-05, + "loss": 6.07, + "step": 7140 + }, + { + "epoch": 14.38, + "learning_rate": 2.3863333333333336e-05, + "loss": 6.0657, + "step": 7160 + }, + { + "epoch": 14.42, + "learning_rate": 2.3930000000000003e-05, + "loss": 6.0749, + "step": 7180 + }, + { + "epoch": 14.46, + "learning_rate": 2.3996666666666667e-05, + "loss": 6.0646, + "step": 7200 + }, + { + "epoch": 14.5, + "learning_rate": 2.4063333333333334e-05, + "loss": 6.0441, + "step": 7220 + }, + { + "epoch": 14.54, + "learning_rate": 2.413e-05, + "loss": 6.0496, + "step": 7240 + }, + { + "epoch": 14.58, + "learning_rate": 2.4196666666666668e-05, + "loss": 6.0446, + "step": 7260 + }, + { + "epoch": 14.62, + "learning_rate": 2.4263333333333335e-05, + "loss": 6.0383, + "step": 7280 + }, + { + "epoch": 14.66, + "learning_rate": 2.433e-05, + "loss": 6.0443, + "step": 7300 + }, + { + "epoch": 14.7, + "learning_rate": 2.4396666666666666e-05, + "loss": 6.0334, + "step": 7320 + }, + { + "epoch": 14.74, + "learning_rate": 2.4463333333333337e-05, + "loss": 6.0363, + "step": 7340 + }, + { + "epoch": 14.78, + "learning_rate": 2.453e-05, + "loss": 6.0326, + "step": 7360 + }, + { + "epoch": 14.82, + "learning_rate": 2.4596666666666668e-05, + "loss": 6.0159, + "step": 7380 + }, + { + "epoch": 14.86, + "learning_rate": 2.4663333333333335e-05, + "loss": 6.0149, + "step": 7400 + }, + { + "epoch": 14.9, + "learning_rate": 2.473e-05, + "loss": 6.008, + "step": 7420 + }, + { + "epoch": 14.94, + "learning_rate": 2.479666666666667e-05, + "loss": 5.9926, + "step": 7440 + }, + { + "epoch": 14.98, + "learning_rate": 2.4863333333333336e-05, + "loss": 5.9917, + "step": 7460 + }, + { + "epoch": 15.02, + "learning_rate": 2.493e-05, + "loss": 5.9806, + "step": 7480 + }, + { + "epoch": 15.06, + "learning_rate": 2.4996666666666667e-05, + "loss": 5.9725, + "step": 7500 + }, + { + "epoch": 15.1, + "learning_rate": 2.5063333333333334e-05, + "loss": 5.9696, + "step": 7520 + }, + { + "epoch": 15.14, + "learning_rate": 2.5130000000000005e-05, + "loss": 5.9628, + "step": 7540 + }, + { + "epoch": 15.18, + "learning_rate": 2.519666666666667e-05, + "loss": 5.9704, + "step": 7560 + }, + { + "epoch": 15.22, + "learning_rate": 2.5263333333333333e-05, + "loss": 5.9589, + "step": 7580 + }, + { + "epoch": 15.26, + "learning_rate": 2.5330000000000003e-05, + "loss": 5.9433, + "step": 7600 + }, + { + "epoch": 15.3, + "learning_rate": 2.539666666666667e-05, + "loss": 5.9524, + "step": 7620 + }, + { + "epoch": 15.34, + "learning_rate": 2.5463333333333334e-05, + "loss": 5.941, + "step": 7640 + }, + { + "epoch": 15.38, + "learning_rate": 2.5530000000000005e-05, + "loss": 5.9431, + "step": 7660 + }, + { + "epoch": 15.42, + "learning_rate": 2.559666666666667e-05, + "loss": 5.93, + "step": 7680 + }, + { + "epoch": 15.46, + "learning_rate": 2.5663333333333332e-05, + "loss": 5.9357, + "step": 7700 + }, + { + "epoch": 15.5, + "learning_rate": 2.573e-05, + "loss": 5.9245, + "step": 7720 + }, + { + "epoch": 15.54, + "learning_rate": 2.579666666666667e-05, + "loss": 5.9173, + "step": 7740 + }, + { + "epoch": 15.58, + "learning_rate": 2.5863333333333334e-05, + "loss": 5.9338, + "step": 7760 + }, + { + "epoch": 15.62, + "learning_rate": 2.5929999999999997e-05, + "loss": 5.9171, + "step": 7780 + }, + { + "epoch": 15.66, + "learning_rate": 2.5996666666666668e-05, + "loss": 5.9147, + "step": 7800 + }, + { + "epoch": 15.7, + "learning_rate": 2.6063333333333335e-05, + "loss": 5.917, + "step": 7820 + }, + { + "epoch": 15.74, + "learning_rate": 2.613e-05, + "loss": 5.9025, + "step": 7840 + }, + { + "epoch": 15.78, + "learning_rate": 2.619666666666667e-05, + "loss": 5.9035, + "step": 7860 + }, + { + "epoch": 15.82, + "learning_rate": 2.6263333333333333e-05, + "loss": 5.8943, + "step": 7880 + }, + { + "epoch": 15.86, + "learning_rate": 2.633e-05, + "loss": 5.8812, + "step": 7900 + }, + { + "epoch": 15.9, + "learning_rate": 2.639666666666667e-05, + "loss": 5.8727, + "step": 7920 + }, + { + "epoch": 15.94, + "learning_rate": 2.6463333333333335e-05, + "loss": 5.8782, + "step": 7940 + }, + { + "epoch": 15.98, + "learning_rate": 2.653e-05, + "loss": 5.8661, + "step": 7960 + }, + { + "epoch": 16.02, + "learning_rate": 2.659666666666667e-05, + "loss": 5.8576, + "step": 7980 + }, + { + "epoch": 16.06, + "learning_rate": 2.6663333333333336e-05, + "loss": 5.8545, + "step": 8000 + }, + { + "epoch": 16.06, + "eval_loss": 5.796634197235107, + "eval_runtime": 34.0997, + "eval_samples_per_second": 22.464, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.008509840095610656, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0, + "step": 8000 + }, + { + "epoch": 16.1, + "learning_rate": 2.6726666666666666e-05, + "loss": 5.842, + "step": 8020 + }, + { + "epoch": 16.14, + "learning_rate": 2.6793333333333337e-05, + "loss": 5.8295, + "step": 8040 + }, + { + "epoch": 16.18, + "learning_rate": 2.686e-05, + "loss": 5.8437, + "step": 8060 + }, + { + "epoch": 16.22, + "learning_rate": 2.6926666666666667e-05, + "loss": 5.8318, + "step": 8080 + }, + { + "epoch": 16.27, + "learning_rate": 2.6993333333333338e-05, + "loss": 5.8407, + "step": 8100 + }, + { + "epoch": 16.31, + "learning_rate": 2.7060000000000002e-05, + "loss": 5.8377, + "step": 8120 + }, + { + "epoch": 16.35, + "learning_rate": 2.7126666666666666e-05, + "loss": 5.8297, + "step": 8140 + }, + { + "epoch": 16.39, + "learning_rate": 2.7193333333333336e-05, + "loss": 5.8202, + "step": 8160 + }, + { + "epoch": 16.43, + "learning_rate": 2.7260000000000003e-05, + "loss": 5.8245, + "step": 8180 + }, + { + "epoch": 16.47, + "learning_rate": 2.7326666666666667e-05, + "loss": 5.8402, + "step": 8200 + }, + { + "epoch": 16.51, + "learning_rate": 2.739333333333333e-05, + "loss": 5.818, + "step": 8220 + }, + { + "epoch": 16.55, + "learning_rate": 2.746e-05, + "loss": 5.8133, + "step": 8240 + }, + { + "epoch": 16.59, + "learning_rate": 2.752666666666667e-05, + "loss": 5.8034, + "step": 8260 + }, + { + "epoch": 16.63, + "learning_rate": 2.7593333333333332e-05, + "loss": 5.7979, + "step": 8280 + }, + { + "epoch": 16.67, + "learning_rate": 2.7660000000000003e-05, + "loss": 5.7991, + "step": 8300 + }, + { + "epoch": 16.71, + "learning_rate": 2.7726666666666667e-05, + "loss": 5.7757, + "step": 8320 + }, + { + "epoch": 16.75, + "learning_rate": 2.7793333333333334e-05, + "loss": 5.7913, + "step": 8340 + }, + { + "epoch": 16.79, + "learning_rate": 2.7860000000000004e-05, + "loss": 5.7618, + "step": 8360 + }, + { + "epoch": 16.83, + "learning_rate": 2.7926666666666668e-05, + "loss": 5.7816, + "step": 8380 + }, + { + "epoch": 16.87, + "learning_rate": 2.7993333333333332e-05, + "loss": 5.7695, + "step": 8400 + }, + { + "epoch": 16.91, + "learning_rate": 2.8060000000000002e-05, + "loss": 5.7718, + "step": 8420 + }, + { + "epoch": 16.95, + "learning_rate": 2.8126666666666666e-05, + "loss": 5.7682, + "step": 8440 + }, + { + "epoch": 16.99, + "learning_rate": 2.8193333333333333e-05, + "loss": 5.76, + "step": 8460 + }, + { + "epoch": 17.03, + "learning_rate": 2.8260000000000004e-05, + "loss": 5.7454, + "step": 8480 + }, + { + "epoch": 17.07, + "learning_rate": 2.8326666666666668e-05, + "loss": 5.7499, + "step": 8500 + }, + { + "epoch": 17.11, + "learning_rate": 2.839333333333333e-05, + "loss": 5.7426, + "step": 8520 + }, + { + "epoch": 17.15, + "learning_rate": 2.8460000000000002e-05, + "loss": 5.7262, + "step": 8540 + }, + { + "epoch": 17.19, + "learning_rate": 2.852666666666667e-05, + "loss": 5.7313, + "step": 8560 + }, + { + "epoch": 17.23, + "learning_rate": 2.8593333333333333e-05, + "loss": 5.7344, + "step": 8580 + }, + { + "epoch": 17.27, + "learning_rate": 2.8660000000000003e-05, + "loss": 5.7263, + "step": 8600 + }, + { + "epoch": 17.31, + "learning_rate": 2.8726666666666667e-05, + "loss": 5.7383, + "step": 8620 + }, + { + "epoch": 17.35, + "learning_rate": 2.8793333333333334e-05, + "loss": 5.7302, + "step": 8640 + }, + { + "epoch": 17.39, + "learning_rate": 2.8860000000000005e-05, + "loss": 5.7045, + "step": 8660 + }, + { + "epoch": 17.43, + "learning_rate": 2.892666666666667e-05, + "loss": 5.7214, + "step": 8680 + }, + { + "epoch": 17.47, + "learning_rate": 2.8993333333333332e-05, + "loss": 5.711, + "step": 8700 + }, + { + "epoch": 17.51, + "learning_rate": 2.9060000000000003e-05, + "loss": 5.7085, + "step": 8720 + }, + { + "epoch": 17.55, + "learning_rate": 2.912666666666667e-05, + "loss": 5.6965, + "step": 8740 + }, + { + "epoch": 17.59, + "learning_rate": 2.9193333333333334e-05, + "loss": 5.6932, + "step": 8760 + }, + { + "epoch": 17.63, + "learning_rate": 2.9260000000000004e-05, + "loss": 5.6818, + "step": 8780 + }, + { + "epoch": 17.67, + "learning_rate": 2.9326666666666668e-05, + "loss": 5.6689, + "step": 8800 + }, + { + "epoch": 17.71, + "learning_rate": 2.9393333333333335e-05, + "loss": 5.6761, + "step": 8820 + }, + { + "epoch": 17.75, + "learning_rate": 2.946e-05, + "loss": 5.6738, + "step": 8840 + }, + { + "epoch": 17.79, + "learning_rate": 2.952666666666667e-05, + "loss": 5.6768, + "step": 8860 + }, + { + "epoch": 17.83, + "learning_rate": 2.9593333333333333e-05, + "loss": 5.66, + "step": 8880 + }, + { + "epoch": 17.87, + "learning_rate": 2.9659999999999997e-05, + "loss": 5.6589, + "step": 8900 + }, + { + "epoch": 17.91, + "learning_rate": 2.9726666666666668e-05, + "loss": 5.6582, + "step": 8920 + }, + { + "epoch": 17.95, + "learning_rate": 2.9793333333333335e-05, + "loss": 5.6462, + "step": 8940 + }, + { + "epoch": 17.99, + "learning_rate": 2.986e-05, + "loss": 5.6659, + "step": 8960 + }, + { + "epoch": 18.03, + "learning_rate": 2.992666666666667e-05, + "loss": 5.6347, + "step": 8980 + }, + { + "epoch": 18.07, + "learning_rate": 2.9993333333333333e-05, + "loss": 5.6355, + "step": 9000 + }, + { + "epoch": 18.07, + "eval_loss": 5.554161071777344, + "eval_runtime": 34.6983, + "eval_samples_per_second": 22.076, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.00561897226930361, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0, + "step": 9000 + }, + { + "epoch": 18.11, + "learning_rate": 3.006e-05, + "loss": 5.6488, + "step": 9020 + }, + { + "epoch": 18.15, + "learning_rate": 3.012666666666667e-05, + "loss": 5.632, + "step": 9040 + }, + { + "epoch": 18.19, + "learning_rate": 3.0193333333333335e-05, + "loss": 5.6111, + "step": 9060 + }, + { + "epoch": 18.23, + "learning_rate": 3.0259999999999998e-05, + "loss": 5.6283, + "step": 9080 + }, + { + "epoch": 18.27, + "learning_rate": 3.032666666666667e-05, + "loss": 5.6158, + "step": 9100 + }, + { + "epoch": 18.31, + "learning_rate": 3.0393333333333336e-05, + "loss": 5.6065, + "step": 9120 + }, + { + "epoch": 18.35, + "learning_rate": 3.046e-05, + "loss": 5.6184, + "step": 9140 + }, + { + "epoch": 18.39, + "learning_rate": 3.052666666666667e-05, + "loss": 5.5967, + "step": 9160 + }, + { + "epoch": 18.43, + "learning_rate": 3.059333333333334e-05, + "loss": 5.6024, + "step": 9180 + }, + { + "epoch": 18.47, + "learning_rate": 3.066e-05, + "loss": 5.6062, + "step": 9200 + }, + { + "epoch": 18.51, + "learning_rate": 3.072666666666667e-05, + "loss": 5.5902, + "step": 9220 + }, + { + "epoch": 18.55, + "learning_rate": 3.0793333333333336e-05, + "loss": 5.6092, + "step": 9240 + }, + { + "epoch": 18.59, + "learning_rate": 3.086e-05, + "loss": 5.5985, + "step": 9260 + }, + { + "epoch": 18.63, + "learning_rate": 3.092666666666667e-05, + "loss": 5.5887, + "step": 9280 + }, + { + "epoch": 18.67, + "learning_rate": 3.0993333333333334e-05, + "loss": 5.5852, + "step": 9300 + }, + { + "epoch": 18.71, + "learning_rate": 3.106e-05, + "loss": 5.5804, + "step": 9320 + }, + { + "epoch": 18.76, + "learning_rate": 3.112666666666667e-05, + "loss": 5.5732, + "step": 9340 + }, + { + "epoch": 18.8, + "learning_rate": 3.119333333333334e-05, + "loss": 5.5819, + "step": 9360 + }, + { + "epoch": 18.84, + "learning_rate": 3.126e-05, + "loss": 5.5575, + "step": 9380 + }, + { + "epoch": 18.88, + "learning_rate": 3.132666666666667e-05, + "loss": 5.5665, + "step": 9400 + }, + { + "epoch": 18.92, + "learning_rate": 3.1393333333333337e-05, + "loss": 5.5558, + "step": 9420 + }, + { + "epoch": 18.96, + "learning_rate": 3.146e-05, + "loss": 5.5319, + "step": 9440 + }, + { + "epoch": 19.0, + "learning_rate": 3.1526666666666664e-05, + "loss": 5.5398, + "step": 9460 + }, + { + "epoch": 19.04, + "learning_rate": 3.1593333333333335e-05, + "loss": 5.5525, + "step": 9480 + }, + { + "epoch": 19.08, + "learning_rate": 3.166e-05, + "loss": 5.5249, + "step": 9500 + }, + { + "epoch": 19.12, + "learning_rate": 3.172666666666667e-05, + "loss": 5.5234, + "step": 9520 + }, + { + "epoch": 19.16, + "learning_rate": 3.179333333333333e-05, + "loss": 5.553, + "step": 9540 + }, + { + "epoch": 19.2, + "learning_rate": 3.186e-05, + "loss": 5.4998, + "step": 9560 + }, + { + "epoch": 19.24, + "learning_rate": 3.192666666666667e-05, + "loss": 5.5049, + "step": 9580 + }, + { + "epoch": 19.28, + "learning_rate": 3.199333333333334e-05, + "loss": 5.5019, + "step": 9600 + }, + { + "epoch": 19.32, + "learning_rate": 3.206e-05, + "loss": 5.5316, + "step": 9620 + }, + { + "epoch": 19.36, + "learning_rate": 3.2126666666666665e-05, + "loss": 5.5142, + "step": 9640 + }, + { + "epoch": 19.4, + "learning_rate": 3.2193333333333336e-05, + "loss": 5.4933, + "step": 9660 + }, + { + "epoch": 19.44, + "learning_rate": 3.226e-05, + "loss": 5.5044, + "step": 9680 + }, + { + "epoch": 19.48, + "learning_rate": 3.232666666666666e-05, + "loss": 5.5109, + "step": 9700 + }, + { + "epoch": 19.52, + "learning_rate": 3.2393333333333334e-05, + "loss": 5.5029, + "step": 9720 + }, + { + "epoch": 19.56, + "learning_rate": 3.2460000000000004e-05, + "loss": 5.4855, + "step": 9740 + }, + { + "epoch": 19.6, + "learning_rate": 3.252666666666667e-05, + "loss": 5.4862, + "step": 9760 + }, + { + "epoch": 19.64, + "learning_rate": 3.259333333333334e-05, + "loss": 5.5029, + "step": 9780 + }, + { + "epoch": 19.68, + "learning_rate": 3.266e-05, + "loss": 5.48, + "step": 9800 + }, + { + "epoch": 19.72, + "learning_rate": 3.2726666666666666e-05, + "loss": 5.4733, + "step": 9820 + }, + { + "epoch": 19.76, + "learning_rate": 3.279333333333334e-05, + "loss": 5.4774, + "step": 9840 + }, + { + "epoch": 19.8, + "learning_rate": 3.286e-05, + "loss": 5.4644, + "step": 9860 + }, + { + "epoch": 19.84, + "learning_rate": 3.2926666666666664e-05, + "loss": 5.4509, + "step": 9880 + }, + { + "epoch": 19.88, + "learning_rate": 3.2993333333333335e-05, + "loss": 5.4491, + "step": 9900 + }, + { + "epoch": 19.92, + "learning_rate": 3.3060000000000005e-05, + "loss": 5.4447, + "step": 9920 + }, + { + "epoch": 19.96, + "learning_rate": 3.312666666666667e-05, + "loss": 5.4363, + "step": 9940 + }, + { + "epoch": 20.0, + "learning_rate": 3.319333333333334e-05, + "loss": 5.4478, + "step": 9960 + }, + { + "epoch": 20.04, + "learning_rate": 3.3260000000000003e-05, + "loss": 5.4223, + "step": 9980 + }, + { + "epoch": 20.08, + "learning_rate": 3.332666666666667e-05, + "loss": 5.4409, + "step": 10000 + }, + { + "epoch": 20.08, + "eval_loss": 5.326024532318115, + "eval_runtime": 32.546, + "eval_samples_per_second": 23.536, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.003690495029122576, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.2664635915288658e-06, + "step": 10000 + }, + { + "epoch": 20.12, + "learning_rate": 3.339333333333334e-05, + "loss": 5.4198, + "step": 10020 + }, + { + "epoch": 20.16, + "learning_rate": 3.345666666666667e-05, + "loss": 5.4244, + "step": 10040 + }, + { + "epoch": 20.2, + "learning_rate": 3.352333333333333e-05, + "loss": 5.4185, + "step": 10060 + }, + { + "epoch": 20.24, + "learning_rate": 3.359e-05, + "loss": 5.4327, + "step": 10080 + }, + { + "epoch": 20.28, + "learning_rate": 3.365666666666667e-05, + "loss": 5.4165, + "step": 10100 + }, + { + "epoch": 20.32, + "learning_rate": 3.3723333333333336e-05, + "loss": 5.4048, + "step": 10120 + }, + { + "epoch": 20.36, + "learning_rate": 3.379e-05, + "loss": 5.4188, + "step": 10140 + }, + { + "epoch": 20.4, + "learning_rate": 3.385666666666667e-05, + "loss": 5.4002, + "step": 10160 + }, + { + "epoch": 20.44, + "learning_rate": 3.3923333333333334e-05, + "loss": 5.3976, + "step": 10180 + }, + { + "epoch": 20.48, + "learning_rate": 3.399e-05, + "loss": 5.3914, + "step": 10200 + }, + { + "epoch": 20.52, + "learning_rate": 3.405666666666667e-05, + "loss": 5.3779, + "step": 10220 + }, + { + "epoch": 20.56, + "learning_rate": 3.412333333333333e-05, + "loss": 5.4054, + "step": 10240 + }, + { + "epoch": 20.6, + "learning_rate": 3.419e-05, + "loss": 5.3863, + "step": 10260 + }, + { + "epoch": 20.64, + "learning_rate": 3.4256666666666674e-05, + "loss": 5.3914, + "step": 10280 + }, + { + "epoch": 20.68, + "learning_rate": 3.432333333333334e-05, + "loss": 5.3732, + "step": 10300 + }, + { + "epoch": 20.72, + "learning_rate": 3.439e-05, + "loss": 5.3825, + "step": 10320 + }, + { + "epoch": 20.76, + "learning_rate": 3.445666666666667e-05, + "loss": 5.3727, + "step": 10340 + }, + { + "epoch": 20.8, + "learning_rate": 3.4523333333333335e-05, + "loss": 5.3659, + "step": 10360 + }, + { + "epoch": 20.84, + "learning_rate": 3.459e-05, + "loss": 5.3491, + "step": 10380 + }, + { + "epoch": 20.88, + "learning_rate": 3.465666666666667e-05, + "loss": 5.3622, + "step": 10400 + }, + { + "epoch": 20.92, + "learning_rate": 3.4723333333333333e-05, + "loss": 5.3612, + "step": 10420 + }, + { + "epoch": 20.96, + "learning_rate": 3.479e-05, + "loss": 5.3485, + "step": 10440 + }, + { + "epoch": 21.0, + "learning_rate": 3.485666666666667e-05, + "loss": 5.351, + "step": 10460 + }, + { + "epoch": 21.04, + "learning_rate": 3.492333333333334e-05, + "loss": 5.3276, + "step": 10480 + }, + { + "epoch": 21.08, + "learning_rate": 3.499e-05, + "loss": 5.343, + "step": 10500 + }, + { + "epoch": 21.12, + "learning_rate": 3.505666666666667e-05, + "loss": 5.3302, + "step": 10520 + }, + { + "epoch": 21.16, + "learning_rate": 3.5123333333333336e-05, + "loss": 5.3236, + "step": 10540 + }, + { + "epoch": 21.2, + "learning_rate": 3.519e-05, + "loss": 5.3302, + "step": 10560 + }, + { + "epoch": 21.24, + "learning_rate": 3.5256666666666664e-05, + "loss": 5.3264, + "step": 10580 + }, + { + "epoch": 21.29, + "learning_rate": 3.5323333333333335e-05, + "loss": 5.3039, + "step": 10600 + }, + { + "epoch": 21.33, + "learning_rate": 3.539e-05, + "loss": 5.3203, + "step": 10620 + }, + { + "epoch": 21.37, + "learning_rate": 3.545666666666667e-05, + "loss": 5.2992, + "step": 10640 + }, + { + "epoch": 21.41, + "learning_rate": 3.552333333333334e-05, + "loss": 5.3107, + "step": 10660 + }, + { + "epoch": 21.45, + "learning_rate": 3.559e-05, + "loss": 5.3035, + "step": 10680 + }, + { + "epoch": 21.49, + "learning_rate": 3.565666666666667e-05, + "loss": 5.2946, + "step": 10700 + }, + { + "epoch": 21.53, + "learning_rate": 3.572333333333334e-05, + "loss": 5.2925, + "step": 10720 + }, + { + "epoch": 21.57, + "learning_rate": 3.579e-05, + "loss": 5.2945, + "step": 10740 + }, + { + "epoch": 21.61, + "learning_rate": 3.5856666666666665e-05, + "loss": 5.2776, + "step": 10760 + }, + { + "epoch": 21.65, + "learning_rate": 3.5923333333333336e-05, + "loss": 5.2989, + "step": 10780 + }, + { + "epoch": 21.69, + "learning_rate": 3.599e-05, + "loss": 5.3092, + "step": 10800 + }, + { + "epoch": 21.73, + "learning_rate": 3.605666666666666e-05, + "loss": 5.2933, + "step": 10820 + }, + { + "epoch": 21.77, + "learning_rate": 3.6123333333333334e-05, + "loss": 5.278, + "step": 10840 + }, + { + "epoch": 21.81, + "learning_rate": 3.6190000000000004e-05, + "loss": 5.2877, + "step": 10860 + }, + { + "epoch": 21.85, + "learning_rate": 3.625666666666667e-05, + "loss": 5.2807, + "step": 10880 + }, + { + "epoch": 21.89, + "learning_rate": 3.632333333333334e-05, + "loss": 5.2739, + "step": 10900 + }, + { + "epoch": 21.93, + "learning_rate": 3.639e-05, + "loss": 5.2706, + "step": 10920 + }, + { + "epoch": 21.97, + "learning_rate": 3.6456666666666666e-05, + "loss": 5.2739, + "step": 10940 + }, + { + "epoch": 22.01, + "learning_rate": 3.6523333333333337e-05, + "loss": 5.2487, + "step": 10960 + }, + { + "epoch": 22.05, + "learning_rate": 3.659e-05, + "loss": 5.2283, + "step": 10980 + }, + { + "epoch": 22.09, + "learning_rate": 3.6656666666666664e-05, + "loss": 5.2416, + "step": 11000 + }, + { + "epoch": 22.09, + "eval_loss": 5.129831314086914, + "eval_runtime": 32.4721, + "eval_samples_per_second": 23.589, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004821757073600221, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.799390774586597e-06, + "step": 11000 + }, + { + "epoch": 22.13, + "learning_rate": 3.6723333333333335e-05, + "loss": 5.2356, + "step": 11020 + }, + { + "epoch": 22.17, + "learning_rate": 3.6790000000000005e-05, + "loss": 5.2272, + "step": 11040 + }, + { + "epoch": 22.21, + "learning_rate": 3.685666666666667e-05, + "loss": 5.2205, + "step": 11060 + }, + { + "epoch": 22.25, + "learning_rate": 3.692333333333334e-05, + "loss": 5.2465, + "step": 11080 + }, + { + "epoch": 22.29, + "learning_rate": 3.699e-05, + "loss": 5.2224, + "step": 11100 + }, + { + "epoch": 22.33, + "learning_rate": 3.705666666666667e-05, + "loss": 5.2258, + "step": 11120 + }, + { + "epoch": 22.37, + "learning_rate": 3.712333333333334e-05, + "loss": 5.2227, + "step": 11140 + }, + { + "epoch": 22.41, + "learning_rate": 3.719e-05, + "loss": 5.2321, + "step": 11160 + }, + { + "epoch": 22.45, + "learning_rate": 3.7256666666666665e-05, + "loss": 5.245, + "step": 11180 + }, + { + "epoch": 22.49, + "learning_rate": 3.7323333333333336e-05, + "loss": 5.2223, + "step": 11200 + }, + { + "epoch": 22.53, + "learning_rate": 3.739e-05, + "loss": 5.2141, + "step": 11220 + }, + { + "epoch": 22.57, + "learning_rate": 3.745666666666667e-05, + "loss": 5.205, + "step": 11240 + }, + { + "epoch": 22.61, + "learning_rate": 3.7523333333333334e-05, + "loss": 5.2092, + "step": 11260 + }, + { + "epoch": 22.65, + "learning_rate": 3.7590000000000004e-05, + "loss": 5.2076, + "step": 11280 + }, + { + "epoch": 22.69, + "learning_rate": 3.765666666666667e-05, + "loss": 5.19, + "step": 11300 + }, + { + "epoch": 22.73, + "learning_rate": 3.772333333333333e-05, + "loss": 5.1858, + "step": 11320 + }, + { + "epoch": 22.77, + "learning_rate": 3.779e-05, + "loss": 5.1804, + "step": 11340 + }, + { + "epoch": 22.81, + "learning_rate": 3.7856666666666666e-05, + "loss": 5.1833, + "step": 11360 + }, + { + "epoch": 22.85, + "learning_rate": 3.792333333333333e-05, + "loss": 5.1785, + "step": 11380 + }, + { + "epoch": 22.89, + "learning_rate": 3.799e-05, + "loss": 5.1979, + "step": 11400 + }, + { + "epoch": 22.93, + "learning_rate": 3.805666666666667e-05, + "loss": 5.1958, + "step": 11420 + }, + { + "epoch": 22.97, + "learning_rate": 3.8123333333333335e-05, + "loss": 5.1732, + "step": 11440 + }, + { + "epoch": 23.01, + "learning_rate": 3.8190000000000005e-05, + "loss": 5.1609, + "step": 11460 + }, + { + "epoch": 23.05, + "learning_rate": 3.825666666666667e-05, + "loss": 5.1532, + "step": 11480 + }, + { + "epoch": 23.09, + "learning_rate": 3.832333333333333e-05, + "loss": 5.1539, + "step": 11500 + }, + { + "epoch": 23.13, + "learning_rate": 3.8390000000000003e-05, + "loss": 5.1266, + "step": 11520 + }, + { + "epoch": 23.17, + "learning_rate": 3.845666666666667e-05, + "loss": 5.1278, + "step": 11540 + }, + { + "epoch": 23.21, + "learning_rate": 3.852333333333333e-05, + "loss": 5.1264, + "step": 11560 + }, + { + "epoch": 23.25, + "learning_rate": 3.859e-05, + "loss": 5.1269, + "step": 11580 + }, + { + "epoch": 23.29, + "learning_rate": 3.865666666666667e-05, + "loss": 5.1526, + "step": 11600 + }, + { + "epoch": 23.33, + "learning_rate": 3.8723333333333336e-05, + "loss": 5.1362, + "step": 11620 + }, + { + "epoch": 23.37, + "learning_rate": 3.8790000000000006e-05, + "loss": 5.1418, + "step": 11640 + }, + { + "epoch": 23.41, + "learning_rate": 3.885666666666667e-05, + "loss": 5.1369, + "step": 11660 + }, + { + "epoch": 23.45, + "learning_rate": 3.8923333333333334e-05, + "loss": 5.1225, + "step": 11680 + }, + { + "epoch": 23.49, + "learning_rate": 3.8990000000000004e-05, + "loss": 5.1319, + "step": 11700 + }, + { + "epoch": 23.53, + "learning_rate": 3.905666666666667e-05, + "loss": 5.1311, + "step": 11720 + }, + { + "epoch": 23.57, + "learning_rate": 3.912333333333333e-05, + "loss": 5.1106, + "step": 11740 + }, + { + "epoch": 23.61, + "learning_rate": 3.919e-05, + "loss": 5.1322, + "step": 11760 + }, + { + "epoch": 23.65, + "learning_rate": 3.9256666666666666e-05, + "loss": 5.1095, + "step": 11780 + }, + { + "epoch": 23.69, + "learning_rate": 3.932333333333334e-05, + "loss": 5.1246, + "step": 11800 + }, + { + "epoch": 23.73, + "learning_rate": 3.939e-05, + "loss": 5.1112, + "step": 11820 + }, + { + "epoch": 23.78, + "learning_rate": 3.945666666666667e-05, + "loss": 5.0997, + "step": 11840 + }, + { + "epoch": 23.82, + "learning_rate": 3.9523333333333335e-05, + "loss": 5.101, + "step": 11860 + }, + { + "epoch": 23.86, + "learning_rate": 3.959e-05, + "loss": 5.0975, + "step": 11880 + }, + { + "epoch": 23.9, + "learning_rate": 3.965666666666667e-05, + "loss": 5.1098, + "step": 11900 + }, + { + "epoch": 23.94, + "learning_rate": 3.972333333333333e-05, + "loss": 5.0991, + "step": 11920 + }, + { + "epoch": 23.98, + "learning_rate": 3.979e-05, + "loss": 5.081, + "step": 11940 + }, + { + "epoch": 24.02, + "learning_rate": 3.985666666666667e-05, + "loss": 5.0732, + "step": 11960 + }, + { + "epoch": 24.06, + "learning_rate": 3.992333333333334e-05, + "loss": 5.0706, + "step": 11980 + }, + { + "epoch": 24.1, + "learning_rate": 3.999e-05, + "loss": 5.0677, + "step": 12000 + }, + { + "epoch": 24.1, + "eval_loss": 4.945012092590332, + "eval_runtime": 32.4873, + "eval_samples_per_second": 23.578, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0043539352403231, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.5865245140702057e-05, + "step": 12000 + }, + { + "epoch": 24.14, + "learning_rate": 4.005666666666667e-05, + "loss": 5.056, + "step": 12020 + }, + { + "epoch": 24.18, + "learning_rate": 4.012e-05, + "loss": 5.0681, + "step": 12040 + }, + { + "epoch": 24.22, + "learning_rate": 4.0186666666666666e-05, + "loss": 5.0763, + "step": 12060 + }, + { + "epoch": 24.26, + "learning_rate": 4.0253333333333336e-05, + "loss": 5.0536, + "step": 12080 + }, + { + "epoch": 24.3, + "learning_rate": 4.032e-05, + "loss": 5.0578, + "step": 12100 + }, + { + "epoch": 24.34, + "learning_rate": 4.0386666666666664e-05, + "loss": 5.0452, + "step": 12120 + }, + { + "epoch": 24.38, + "learning_rate": 4.0453333333333335e-05, + "loss": 5.0369, + "step": 12140 + }, + { + "epoch": 24.42, + "learning_rate": 4.0520000000000005e-05, + "loss": 5.0412, + "step": 12160 + }, + { + "epoch": 24.46, + "learning_rate": 4.058666666666667e-05, + "loss": 5.0408, + "step": 12180 + }, + { + "epoch": 24.5, + "learning_rate": 4.065333333333334e-05, + "loss": 5.0362, + "step": 12200 + }, + { + "epoch": 24.54, + "learning_rate": 4.072e-05, + "loss": 5.042, + "step": 12220 + }, + { + "epoch": 24.58, + "learning_rate": 4.078666666666667e-05, + "loss": 5.0357, + "step": 12240 + }, + { + "epoch": 24.62, + "learning_rate": 4.085333333333334e-05, + "loss": 5.0199, + "step": 12260 + }, + { + "epoch": 24.66, + "learning_rate": 4.092e-05, + "loss": 5.0299, + "step": 12280 + }, + { + "epoch": 24.7, + "learning_rate": 4.0986666666666665e-05, + "loss": 5.0365, + "step": 12300 + }, + { + "epoch": 24.74, + "learning_rate": 4.1053333333333336e-05, + "loss": 5.0363, + "step": 12320 + }, + { + "epoch": 24.78, + "learning_rate": 4.1120000000000006e-05, + "loss": 5.0159, + "step": 12340 + }, + { + "epoch": 24.82, + "learning_rate": 4.118666666666667e-05, + "loss": 5.0244, + "step": 12360 + }, + { + "epoch": 24.86, + "learning_rate": 4.1253333333333334e-05, + "loss": 5.0096, + "step": 12380 + }, + { + "epoch": 24.9, + "learning_rate": 4.1320000000000004e-05, + "loss": 5.0113, + "step": 12400 + }, + { + "epoch": 24.94, + "learning_rate": 4.138666666666667e-05, + "loss": 5.0135, + "step": 12420 + }, + { + "epoch": 24.98, + "learning_rate": 4.145333333333333e-05, + "loss": 5.0143, + "step": 12440 + }, + { + "epoch": 25.02, + "learning_rate": 4.152e-05, + "loss": 5.0039, + "step": 12460 + }, + { + "epoch": 25.06, + "learning_rate": 4.1586666666666666e-05, + "loss": 4.9923, + "step": 12480 + }, + { + "epoch": 25.1, + "learning_rate": 4.165333333333333e-05, + "loss": 4.9942, + "step": 12500 + }, + { + "epoch": 25.14, + "learning_rate": 4.172e-05, + "loss": 4.9871, + "step": 12520 + }, + { + "epoch": 25.18, + "learning_rate": 4.178666666666667e-05, + "loss": 4.9686, + "step": 12540 + }, + { + "epoch": 25.22, + "learning_rate": 4.1853333333333335e-05, + "loss": 4.9696, + "step": 12560 + }, + { + "epoch": 25.26, + "learning_rate": 4.1920000000000005e-05, + "loss": 4.9929, + "step": 12580 + }, + { + "epoch": 25.3, + "learning_rate": 4.198666666666667e-05, + "loss": 4.9664, + "step": 12600 + }, + { + "epoch": 25.34, + "learning_rate": 4.205333333333333e-05, + "loss": 4.9612, + "step": 12620 + }, + { + "epoch": 25.38, + "learning_rate": 4.212e-05, + "loss": 4.9862, + "step": 12640 + }, + { + "epoch": 25.42, + "learning_rate": 4.218666666666667e-05, + "loss": 4.9597, + "step": 12660 + }, + { + "epoch": 25.46, + "learning_rate": 4.225333333333333e-05, + "loss": 4.9712, + "step": 12680 + }, + { + "epoch": 25.5, + "learning_rate": 4.232e-05, + "loss": 4.9588, + "step": 12700 + }, + { + "epoch": 25.54, + "learning_rate": 4.238666666666667e-05, + "loss": 4.9601, + "step": 12720 + }, + { + "epoch": 25.58, + "learning_rate": 4.2453333333333336e-05, + "loss": 4.951, + "step": 12740 + }, + { + "epoch": 25.62, + "learning_rate": 4.2520000000000006e-05, + "loss": 4.9415, + "step": 12760 + }, + { + "epoch": 25.66, + "learning_rate": 4.258666666666667e-05, + "loss": 4.9285, + "step": 12780 + }, + { + "epoch": 25.7, + "learning_rate": 4.2653333333333334e-05, + "loss": 4.9603, + "step": 12800 + }, + { + "epoch": 25.74, + "learning_rate": 4.2720000000000004e-05, + "loss": 4.9717, + "step": 12820 + }, + { + "epoch": 25.78, + "learning_rate": 4.278666666666667e-05, + "loss": 4.942, + "step": 12840 + }, + { + "epoch": 25.82, + "learning_rate": 4.285333333333333e-05, + "loss": 4.9334, + "step": 12860 + }, + { + "epoch": 25.86, + "learning_rate": 4.292e-05, + "loss": 4.9322, + "step": 12880 + }, + { + "epoch": 25.9, + "learning_rate": 4.2986666666666666e-05, + "loss": 4.9396, + "step": 12900 + }, + { + "epoch": 25.94, + "learning_rate": 4.305333333333334e-05, + "loss": 4.9329, + "step": 12920 + }, + { + "epoch": 25.98, + "learning_rate": 4.312000000000001e-05, + "loss": 4.9282, + "step": 12940 + }, + { + "epoch": 26.02, + "learning_rate": 4.318666666666667e-05, + "loss": 4.9057, + "step": 12960 + }, + { + "epoch": 26.06, + "learning_rate": 4.3253333333333335e-05, + "loss": 4.9048, + "step": 12980 + }, + { + "epoch": 26.1, + "learning_rate": 4.332e-05, + "loss": 4.9185, + "step": 13000 + }, + { + "epoch": 26.1, + "eval_loss": 4.785652160644531, + "eval_runtime": 32.5295, + "eval_samples_per_second": 23.548, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0018280742779154151, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.1332317957644328e-05, + "step": 13000 + }, + { + "epoch": 26.14, + "learning_rate": 4.338666666666667e-05, + "loss": 4.9045, + "step": 13020 + }, + { + "epoch": 26.18, + "learning_rate": 4.345333333333333e-05, + "loss": 4.8889, + "step": 13040 + }, + { + "epoch": 26.22, + "learning_rate": 4.352e-05, + "loss": 4.8891, + "step": 13060 + }, + { + "epoch": 26.27, + "learning_rate": 4.358666666666667e-05, + "loss": 4.8793, + "step": 13080 + }, + { + "epoch": 26.31, + "learning_rate": 4.365333333333334e-05, + "loss": 4.8964, + "step": 13100 + }, + { + "epoch": 26.35, + "learning_rate": 4.372e-05, + "loss": 4.8764, + "step": 13120 + }, + { + "epoch": 26.39, + "learning_rate": 4.378666666666667e-05, + "loss": 4.9019, + "step": 13140 + }, + { + "epoch": 26.43, + "learning_rate": 4.3853333333333336e-05, + "loss": 4.8953, + "step": 13160 + }, + { + "epoch": 26.47, + "learning_rate": 4.392e-05, + "loss": 4.8754, + "step": 13180 + }, + { + "epoch": 26.51, + "learning_rate": 4.398666666666667e-05, + "loss": 4.8859, + "step": 13200 + }, + { + "epoch": 26.55, + "learning_rate": 4.4053333333333334e-05, + "loss": 4.8924, + "step": 13220 + }, + { + "epoch": 26.59, + "learning_rate": 4.412e-05, + "loss": 4.8695, + "step": 13240 + }, + { + "epoch": 26.63, + "learning_rate": 4.418666666666667e-05, + "loss": 4.8755, + "step": 13260 + }, + { + "epoch": 26.67, + "learning_rate": 4.425333333333334e-05, + "loss": 4.8643, + "step": 13280 + }, + { + "epoch": 26.71, + "learning_rate": 4.432e-05, + "loss": 4.8648, + "step": 13300 + }, + { + "epoch": 26.75, + "learning_rate": 4.438666666666667e-05, + "loss": 4.8657, + "step": 13320 + }, + { + "epoch": 26.79, + "learning_rate": 4.445333333333334e-05, + "loss": 4.886, + "step": 13340 + }, + { + "epoch": 26.83, + "learning_rate": 4.452e-05, + "loss": 4.8649, + "step": 13360 + }, + { + "epoch": 26.87, + "learning_rate": 4.458666666666667e-05, + "loss": 4.8716, + "step": 13380 + }, + { + "epoch": 26.91, + "learning_rate": 4.4653333333333335e-05, + "loss": 4.86, + "step": 13400 + }, + { + "epoch": 26.95, + "learning_rate": 4.472e-05, + "loss": 4.8625, + "step": 13420 + }, + { + "epoch": 26.99, + "learning_rate": 4.478666666666667e-05, + "loss": 4.8575, + "step": 13440 + }, + { + "epoch": 27.03, + "learning_rate": 4.485333333333333e-05, + "loss": 4.8128, + "step": 13460 + }, + { + "epoch": 27.07, + "learning_rate": 4.4920000000000004e-05, + "loss": 4.8168, + "step": 13480 + }, + { + "epoch": 27.11, + "learning_rate": 4.4986666666666674e-05, + "loss": 4.834, + "step": 13500 + }, + { + "epoch": 27.15, + "learning_rate": 4.505333333333334e-05, + "loss": 4.82, + "step": 13520 + }, + { + "epoch": 27.19, + "learning_rate": 4.512e-05, + "loss": 4.8175, + "step": 13540 + }, + { + "epoch": 27.23, + "learning_rate": 4.518666666666667e-05, + "loss": 4.8316, + "step": 13560 + }, + { + "epoch": 27.27, + "learning_rate": 4.5253333333333336e-05, + "loss": 4.8257, + "step": 13580 + }, + { + "epoch": 27.31, + "learning_rate": 4.532e-05, + "loss": 4.8295, + "step": 13600 + }, + { + "epoch": 27.35, + "learning_rate": 4.5386666666666664e-05, + "loss": 4.8178, + "step": 13620 + }, + { + "epoch": 27.39, + "learning_rate": 4.5453333333333334e-05, + "loss": 4.8073, + "step": 13640 + }, + { + "epoch": 27.43, + "learning_rate": 4.5520000000000005e-05, + "loss": 4.7964, + "step": 13660 + }, + { + "epoch": 27.47, + "learning_rate": 4.558666666666667e-05, + "loss": 4.8321, + "step": 13680 + }, + { + "epoch": 27.51, + "learning_rate": 4.565333333333334e-05, + "loss": 4.8036, + "step": 13700 + }, + { + "epoch": 27.55, + "learning_rate": 4.572e-05, + "loss": 4.8192, + "step": 13720 + }, + { + "epoch": 27.59, + "learning_rate": 4.5786666666666666e-05, + "loss": 4.819, + "step": 13740 + }, + { + "epoch": 27.63, + "learning_rate": 4.585333333333334e-05, + "loss": 4.8025, + "step": 13760 + }, + { + "epoch": 27.67, + "learning_rate": 4.592e-05, + "loss": 4.8091, + "step": 13780 + }, + { + "epoch": 27.71, + "learning_rate": 4.5986666666666665e-05, + "loss": 4.7976, + "step": 13800 + }, + { + "epoch": 27.75, + "learning_rate": 4.6053333333333335e-05, + "loss": 4.786, + "step": 13820 + }, + { + "epoch": 27.79, + "learning_rate": 4.612e-05, + "loss": 4.7907, + "step": 13840 + }, + { + "epoch": 27.83, + "learning_rate": 4.618666666666667e-05, + "loss": 4.7959, + "step": 13860 + }, + { + "epoch": 27.87, + "learning_rate": 4.625333333333334e-05, + "loss": 4.8017, + "step": 13880 + }, + { + "epoch": 27.91, + "learning_rate": 4.6320000000000004e-05, + "loss": 4.775, + "step": 13900 + }, + { + "epoch": 27.95, + "learning_rate": 4.638666666666667e-05, + "loss": 4.7663, + "step": 13920 + }, + { + "epoch": 27.99, + "learning_rate": 4.645333333333334e-05, + "loss": 4.7715, + "step": 13940 + }, + { + "epoch": 28.03, + "learning_rate": 4.652e-05, + "loss": 4.7521, + "step": 13960 + }, + { + "epoch": 28.07, + "learning_rate": 4.6586666666666666e-05, + "loss": 4.7393, + "step": 13980 + }, + { + "epoch": 28.11, + "learning_rate": 4.6653333333333336e-05, + "loss": 4.7474, + "step": 14000 + }, + { + "epoch": 28.11, + "eval_loss": 4.652778625488281, + "eval_runtime": 32.4407, + "eval_samples_per_second": 23.612, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0028533725000011645, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.2664635915288658e-06, + "step": 14000 + }, + { + "epoch": 28.15, + "learning_rate": 4.672e-05, + "loss": 4.7461, + "step": 14020 + }, + { + "epoch": 28.19, + "learning_rate": 4.678666666666667e-05, + "loss": 4.7262, + "step": 14040 + }, + { + "epoch": 28.23, + "learning_rate": 4.685000000000001e-05, + "loss": 4.73, + "step": 14060 + }, + { + "epoch": 28.27, + "learning_rate": 4.691666666666667e-05, + "loss": 4.7581, + "step": 14080 + }, + { + "epoch": 28.31, + "learning_rate": 4.6983333333333335e-05, + "loss": 4.7349, + "step": 14100 + }, + { + "epoch": 28.35, + "learning_rate": 4.705e-05, + "loss": 4.7237, + "step": 14120 + }, + { + "epoch": 28.39, + "learning_rate": 4.711666666666667e-05, + "loss": 4.7434, + "step": 14140 + }, + { + "epoch": 28.43, + "learning_rate": 4.718333333333333e-05, + "loss": 4.7503, + "step": 14160 + }, + { + "epoch": 28.47, + "learning_rate": 4.7249999999999997e-05, + "loss": 4.747, + "step": 14180 + }, + { + "epoch": 28.51, + "learning_rate": 4.731666666666667e-05, + "loss": 4.7362, + "step": 14200 + }, + { + "epoch": 28.55, + "learning_rate": 4.738333333333334e-05, + "loss": 4.7505, + "step": 14220 + }, + { + "epoch": 28.59, + "learning_rate": 4.745e-05, + "loss": 4.7154, + "step": 14240 + }, + { + "epoch": 28.63, + "learning_rate": 4.751666666666667e-05, + "loss": 4.7318, + "step": 14260 + }, + { + "epoch": 28.67, + "learning_rate": 4.7583333333333336e-05, + "loss": 4.7322, + "step": 14280 + }, + { + "epoch": 28.71, + "learning_rate": 4.765e-05, + "loss": 4.7444, + "step": 14300 + }, + { + "epoch": 28.76, + "learning_rate": 4.771666666666667e-05, + "loss": 4.7511, + "step": 14320 + }, + { + "epoch": 28.8, + "learning_rate": 4.7783333333333334e-05, + "loss": 4.7371, + "step": 14340 + }, + { + "epoch": 28.84, + "learning_rate": 4.785e-05, + "loss": 4.7268, + "step": 14360 + }, + { + "epoch": 28.88, + "learning_rate": 4.791666666666667e-05, + "loss": 4.7181, + "step": 14380 + }, + { + "epoch": 28.92, + "learning_rate": 4.798333333333334e-05, + "loss": 4.7247, + "step": 14400 + }, + { + "epoch": 28.96, + "learning_rate": 4.805e-05, + "loss": 4.7189, + "step": 14420 + }, + { + "epoch": 29.0, + "learning_rate": 4.811666666666667e-05, + "loss": 4.7216, + "step": 14440 + }, + { + "epoch": 29.04, + "learning_rate": 4.818333333333334e-05, + "loss": 4.6884, + "step": 14460 + }, + { + "epoch": 29.08, + "learning_rate": 4.825e-05, + "loss": 4.6795, + "step": 14480 + }, + { + "epoch": 29.12, + "learning_rate": 4.831666666666667e-05, + "loss": 4.6714, + "step": 14500 + }, + { + "epoch": 29.16, + "learning_rate": 4.8383333333333335e-05, + "loss": 4.6737, + "step": 14520 + }, + { + "epoch": 29.2, + "learning_rate": 4.845e-05, + "loss": 4.6851, + "step": 14540 + }, + { + "epoch": 29.24, + "learning_rate": 4.851666666666667e-05, + "loss": 4.684, + "step": 14560 + }, + { + "epoch": 29.28, + "learning_rate": 4.858333333333333e-05, + "loss": 4.6527, + "step": 14580 + }, + { + "epoch": 29.32, + "learning_rate": 4.8650000000000003e-05, + "loss": 4.6805, + "step": 14600 + }, + { + "epoch": 29.36, + "learning_rate": 4.8716666666666674e-05, + "loss": 4.6675, + "step": 14620 + }, + { + "epoch": 29.4, + "learning_rate": 4.878333333333334e-05, + "loss": 4.677, + "step": 14640 + }, + { + "epoch": 29.44, + "learning_rate": 4.885e-05, + "loss": 4.6793, + "step": 14660 + }, + { + "epoch": 29.48, + "learning_rate": 4.891666666666667e-05, + "loss": 4.6934, + "step": 14680 + }, + { + "epoch": 29.52, + "learning_rate": 4.8983333333333336e-05, + "loss": 4.6736, + "step": 14700 + }, + { + "epoch": 29.56, + "learning_rate": 4.905e-05, + "loss": 4.6622, + "step": 14720 + }, + { + "epoch": 29.6, + "learning_rate": 4.9116666666666663e-05, + "loss": 4.6714, + "step": 14740 + }, + { + "epoch": 29.64, + "learning_rate": 4.9183333333333334e-05, + "loss": 4.656, + "step": 14760 + }, + { + "epoch": 29.68, + "learning_rate": 4.9250000000000004e-05, + "loss": 4.6684, + "step": 14780 + }, + { + "epoch": 29.72, + "learning_rate": 4.931666666666667e-05, + "loss": 4.6335, + "step": 14800 + }, + { + "epoch": 29.76, + "learning_rate": 4.938333333333334e-05, + "loss": 4.6626, + "step": 14820 + }, + { + "epoch": 29.8, + "learning_rate": 4.945e-05, + "loss": 4.6513, + "step": 14840 + }, + { + "epoch": 29.84, + "learning_rate": 4.9516666666666666e-05, + "loss": 4.651, + "step": 14860 + }, + { + "epoch": 29.88, + "learning_rate": 4.958333333333334e-05, + "loss": 4.6675, + "step": 14880 + }, + { + "epoch": 29.92, + "learning_rate": 4.965e-05, + "loss": 4.6403, + "step": 14900 + }, + { + "epoch": 29.96, + "learning_rate": 4.9716666666666664e-05, + "loss": 4.6621, + "step": 14920 + }, + { + "epoch": 30.0, + "learning_rate": 4.9783333333333335e-05, + "loss": 4.6368, + "step": 14940 + }, + { + "epoch": 30.04, + "learning_rate": 4.9850000000000006e-05, + "loss": 4.6144, + "step": 14960 + }, + { + "epoch": 30.08, + "learning_rate": 4.991666666666667e-05, + "loss": 4.6032, + "step": 14980 + }, + { + "epoch": 30.12, + "learning_rate": 4.998333333333334e-05, + "loss": 4.6127, + "step": 15000 + }, + { + "epoch": 30.12, + "eval_loss": 4.523326396942139, + "eval_runtime": 34.4617, + "eval_samples_per_second": 22.228, + "eval_steps_per_second": 0.116, + "eval_tse_ndup": 0.002626506144970761, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.799390774586597e-06, + "step": 15000 + }, + { + "epoch": 30.16, + "learning_rate": 5.005e-05, + "loss": 4.6121, + "step": 15020 + }, + { + "epoch": 30.2, + "learning_rate": 5.011666666666667e-05, + "loss": 4.6079, + "step": 15040 + }, + { + "epoch": 30.24, + "learning_rate": 5.018333333333334e-05, + "loss": 4.6192, + "step": 15060 + }, + { + "epoch": 30.28, + "learning_rate": 5.0249999999999995e-05, + "loss": 4.6135, + "step": 15080 + }, + { + "epoch": 30.32, + "learning_rate": 5.0316666666666665e-05, + "loss": 4.6119, + "step": 15100 + }, + { + "epoch": 30.36, + "learning_rate": 5.0383333333333336e-05, + "loss": 4.5851, + "step": 15120 + }, + { + "epoch": 30.4, + "learning_rate": 5.045e-05, + "loss": 4.6253, + "step": 15140 + }, + { + "epoch": 30.44, + "learning_rate": 5.051666666666667e-05, + "loss": 4.6137, + "step": 15160 + }, + { + "epoch": 30.48, + "learning_rate": 5.058333333333334e-05, + "loss": 4.6016, + "step": 15180 + }, + { + "epoch": 30.52, + "learning_rate": 5.065e-05, + "loss": 4.6243, + "step": 15200 + }, + { + "epoch": 30.56, + "learning_rate": 5.071666666666667e-05, + "loss": 4.5901, + "step": 15220 + }, + { + "epoch": 30.6, + "learning_rate": 5.078333333333334e-05, + "loss": 4.6136, + "step": 15240 + }, + { + "epoch": 30.64, + "learning_rate": 5.0849999999999996e-05, + "loss": 4.5704, + "step": 15260 + }, + { + "epoch": 30.68, + "learning_rate": 5.0916666666666666e-05, + "loss": 4.6086, + "step": 15280 + }, + { + "epoch": 30.72, + "learning_rate": 5.098333333333334e-05, + "loss": 4.6009, + "step": 15300 + }, + { + "epoch": 30.76, + "learning_rate": 5.105e-05, + "loss": 4.5917, + "step": 15320 + }, + { + "epoch": 30.8, + "learning_rate": 5.111666666666667e-05, + "loss": 4.5815, + "step": 15340 + }, + { + "epoch": 30.84, + "learning_rate": 5.118333333333334e-05, + "loss": 4.5926, + "step": 15360 + }, + { + "epoch": 30.88, + "learning_rate": 5.125e-05, + "loss": 4.5926, + "step": 15380 + }, + { + "epoch": 30.92, + "learning_rate": 5.131666666666667e-05, + "loss": 4.5879, + "step": 15400 + }, + { + "epoch": 30.96, + "learning_rate": 5.138333333333334e-05, + "loss": 4.5764, + "step": 15420 + }, + { + "epoch": 31.0, + "learning_rate": 5.145e-05, + "loss": 4.572, + "step": 15440 + }, + { + "epoch": 31.04, + "learning_rate": 5.151666666666667e-05, + "loss": 4.564, + "step": 15460 + }, + { + "epoch": 31.08, + "learning_rate": 5.158333333333334e-05, + "loss": 4.5577, + "step": 15480 + }, + { + "epoch": 31.12, + "learning_rate": 5.1649999999999995e-05, + "loss": 4.5503, + "step": 15500 + }, + { + "epoch": 31.16, + "learning_rate": 5.1716666666666666e-05, + "loss": 4.521, + "step": 15520 + }, + { + "epoch": 31.2, + "learning_rate": 5.1783333333333336e-05, + "loss": 4.5433, + "step": 15540 + }, + { + "epoch": 31.24, + "learning_rate": 5.185e-05, + "loss": 4.5435, + "step": 15560 + }, + { + "epoch": 31.29, + "learning_rate": 5.191666666666667e-05, + "loss": 4.5523, + "step": 15580 + }, + { + "epoch": 31.33, + "learning_rate": 5.198333333333334e-05, + "loss": 4.5384, + "step": 15600 + }, + { + "epoch": 31.37, + "learning_rate": 5.205e-05, + "loss": 4.5295, + "step": 15620 + }, + { + "epoch": 31.41, + "learning_rate": 5.211666666666667e-05, + "loss": 4.542, + "step": 15640 + }, + { + "epoch": 31.45, + "learning_rate": 5.218333333333334e-05, + "loss": 4.5207, + "step": 15660 + }, + { + "epoch": 31.49, + "learning_rate": 5.2249999999999996e-05, + "loss": 4.5242, + "step": 15680 + }, + { + "epoch": 31.53, + "learning_rate": 5.231666666666667e-05, + "loss": 4.5676, + "step": 15700 + }, + { + "epoch": 31.57, + "learning_rate": 5.238333333333334e-05, + "loss": 4.5258, + "step": 15720 + }, + { + "epoch": 31.61, + "learning_rate": 5.245e-05, + "loss": 4.556, + "step": 15740 + }, + { + "epoch": 31.65, + "learning_rate": 5.251666666666667e-05, + "loss": 4.5266, + "step": 15760 + }, + { + "epoch": 31.69, + "learning_rate": 5.258333333333334e-05, + "loss": 4.5364, + "step": 15780 + }, + { + "epoch": 31.73, + "learning_rate": 5.265e-05, + "loss": 4.5458, + "step": 15800 + }, + { + "epoch": 31.77, + "learning_rate": 5.271666666666667e-05, + "loss": 4.5294, + "step": 15820 + }, + { + "epoch": 31.81, + "learning_rate": 5.278333333333334e-05, + "loss": 4.541, + "step": 15840 + }, + { + "epoch": 31.85, + "learning_rate": 5.285e-05, + "loss": 4.5221, + "step": 15860 + }, + { + "epoch": 31.89, + "learning_rate": 5.291666666666667e-05, + "loss": 4.5199, + "step": 15880 + }, + { + "epoch": 31.93, + "learning_rate": 5.298333333333334e-05, + "loss": 4.5239, + "step": 15900 + }, + { + "epoch": 31.97, + "learning_rate": 5.305e-05, + "loss": 4.5298, + "step": 15920 + }, + { + "epoch": 32.01, + "learning_rate": 5.311666666666667e-05, + "loss": 4.4998, + "step": 15940 + }, + { + "epoch": 32.05, + "learning_rate": 5.318333333333334e-05, + "loss": 4.4966, + "step": 15960 + }, + { + "epoch": 32.09, + "learning_rate": 5.325e-05, + "loss": 4.4686, + "step": 15980 + }, + { + "epoch": 32.13, + "learning_rate": 5.331666666666667e-05, + "loss": 4.4641, + "step": 16000 + }, + { + "epoch": 32.13, + "eval_loss": 4.4028239250183105, + "eval_runtime": 34.0987, + "eval_samples_per_second": 22.464, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.0021504568339371557, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 4.5329271830577315e-06, + "step": 16000 + }, + { + "epoch": 32.17, + "learning_rate": 5.338333333333334e-05, + "loss": 4.4769, + "step": 16020 + }, + { + "epoch": 32.21, + "learning_rate": 5.345e-05, + "loss": 4.4706, + "step": 16040 + }, + { + "epoch": 32.25, + "learning_rate": 5.3513333333333335e-05, + "loss": 4.4892, + "step": 16060 + }, + { + "epoch": 32.29, + "learning_rate": 5.3580000000000005e-05, + "loss": 4.4822, + "step": 16080 + }, + { + "epoch": 32.33, + "learning_rate": 5.364666666666667e-05, + "loss": 4.4802, + "step": 16100 + }, + { + "epoch": 32.37, + "learning_rate": 5.371333333333334e-05, + "loss": 4.4917, + "step": 16120 + }, + { + "epoch": 32.41, + "learning_rate": 5.378e-05, + "loss": 4.4908, + "step": 16140 + }, + { + "epoch": 32.45, + "learning_rate": 5.384666666666667e-05, + "loss": 4.4819, + "step": 16160 + }, + { + "epoch": 32.49, + "learning_rate": 5.391333333333334e-05, + "loss": 4.4899, + "step": 16180 + }, + { + "epoch": 32.53, + "learning_rate": 5.3979999999999995e-05, + "loss": 4.466, + "step": 16200 + }, + { + "epoch": 32.57, + "learning_rate": 5.4046666666666665e-05, + "loss": 4.4752, + "step": 16220 + }, + { + "epoch": 32.61, + "learning_rate": 5.4113333333333336e-05, + "loss": 4.479, + "step": 16240 + }, + { + "epoch": 32.65, + "learning_rate": 5.418e-05, + "loss": 4.4489, + "step": 16260 + }, + { + "epoch": 32.69, + "learning_rate": 5.424666666666667e-05, + "loss": 4.4683, + "step": 16280 + }, + { + "epoch": 32.73, + "learning_rate": 5.431333333333334e-05, + "loss": 4.4627, + "step": 16300 + }, + { + "epoch": 32.77, + "learning_rate": 5.438e-05, + "loss": 4.4932, + "step": 16320 + }, + { + "epoch": 32.81, + "learning_rate": 5.444666666666667e-05, + "loss": 4.4676, + "step": 16340 + }, + { + "epoch": 32.85, + "learning_rate": 5.451333333333334e-05, + "loss": 4.4755, + "step": 16360 + }, + { + "epoch": 32.89, + "learning_rate": 5.4579999999999996e-05, + "loss": 4.4675, + "step": 16380 + }, + { + "epoch": 32.93, + "learning_rate": 5.4646666666666666e-05, + "loss": 4.4648, + "step": 16400 + }, + { + "epoch": 32.97, + "learning_rate": 5.471333333333334e-05, + "loss": 4.4522, + "step": 16420 + }, + { + "epoch": 33.01, + "learning_rate": 5.478e-05, + "loss": 4.4586, + "step": 16440 + }, + { + "epoch": 33.05, + "learning_rate": 5.484666666666667e-05, + "loss": 4.4006, + "step": 16460 + }, + { + "epoch": 33.09, + "learning_rate": 5.491333333333334e-05, + "loss": 4.426, + "step": 16480 + }, + { + "epoch": 33.13, + "learning_rate": 5.498e-05, + "loss": 4.4313, + "step": 16500 + }, + { + "epoch": 33.17, + "learning_rate": 5.504666666666667e-05, + "loss": 4.4158, + "step": 16520 + }, + { + "epoch": 33.21, + "learning_rate": 5.511333333333334e-05, + "loss": 4.446, + "step": 16540 + }, + { + "epoch": 33.25, + "learning_rate": 5.518e-05, + "loss": 4.4201, + "step": 16560 + }, + { + "epoch": 33.29, + "learning_rate": 5.524666666666667e-05, + "loss": 4.4208, + "step": 16580 + }, + { + "epoch": 33.33, + "learning_rate": 5.531333333333334e-05, + "loss": 4.43, + "step": 16600 + }, + { + "epoch": 33.37, + "learning_rate": 5.538e-05, + "loss": 4.4252, + "step": 16620 + }, + { + "epoch": 33.41, + "learning_rate": 5.544666666666667e-05, + "loss": 4.4097, + "step": 16640 + }, + { + "epoch": 33.45, + "learning_rate": 5.5513333333333336e-05, + "loss": 4.4407, + "step": 16660 + }, + { + "epoch": 33.49, + "learning_rate": 5.558e-05, + "loss": 4.3998, + "step": 16680 + }, + { + "epoch": 33.53, + "learning_rate": 5.564666666666667e-05, + "loss": 4.4274, + "step": 16700 + }, + { + "epoch": 33.57, + "learning_rate": 5.571333333333334e-05, + "loss": 4.4036, + "step": 16720 + }, + { + "epoch": 33.61, + "learning_rate": 5.578e-05, + "loss": 4.4104, + "step": 16740 + }, + { + "epoch": 33.65, + "learning_rate": 5.584666666666667e-05, + "loss": 4.4004, + "step": 16760 + }, + { + "epoch": 33.69, + "learning_rate": 5.591333333333334e-05, + "loss": 4.4064, + "step": 16780 + }, + { + "epoch": 33.73, + "learning_rate": 5.5979999999999996e-05, + "loss": 4.4059, + "step": 16800 + }, + { + "epoch": 33.78, + "learning_rate": 5.6046666666666667e-05, + "loss": 4.4181, + "step": 16820 + }, + { + "epoch": 33.82, + "learning_rate": 5.611333333333334e-05, + "loss": 4.4037, + "step": 16840 + }, + { + "epoch": 33.86, + "learning_rate": 5.618e-05, + "loss": 4.4017, + "step": 16860 + }, + { + "epoch": 33.9, + "learning_rate": 5.624666666666667e-05, + "loss": 4.3999, + "step": 16880 + }, + { + "epoch": 33.94, + "learning_rate": 5.631333333333334e-05, + "loss": 4.4304, + "step": 16900 + }, + { + "epoch": 33.98, + "learning_rate": 5.6376666666666665e-05, + "loss": 4.4195, + "step": 16920 + }, + { + "epoch": 34.02, + "learning_rate": 5.6443333333333336e-05, + "loss": 4.3814, + "step": 16940 + }, + { + "epoch": 34.06, + "learning_rate": 5.6510000000000006e-05, + "loss": 4.363, + "step": 16960 + }, + { + "epoch": 34.1, + "learning_rate": 5.657666666666667e-05, + "loss": 4.3332, + "step": 16980 + }, + { + "epoch": 34.14, + "learning_rate": 5.664333333333334e-05, + "loss": 4.3572, + "step": 17000 + }, + { + "epoch": 34.14, + "eval_loss": 4.296210765838623, + "eval_runtime": 32.5174, + "eval_samples_per_second": 23.557, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.002247042532030213, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.1332317957644328e-05, + "step": 17000 + }, + { + "epoch": 34.18, + "learning_rate": 5.6710000000000004e-05, + "loss": 4.3563, + "step": 17020 + }, + { + "epoch": 34.22, + "learning_rate": 5.677666666666667e-05, + "loss": 4.3667, + "step": 17040 + }, + { + "epoch": 34.26, + "learning_rate": 5.684333333333334e-05, + "loss": 4.3711, + "step": 17060 + }, + { + "epoch": 34.3, + "learning_rate": 5.691000000000001e-05, + "loss": 4.3467, + "step": 17080 + }, + { + "epoch": 34.34, + "learning_rate": 5.6976666666666666e-05, + "loss": 4.3624, + "step": 17100 + }, + { + "epoch": 34.38, + "learning_rate": 5.704333333333334e-05, + "loss": 4.364, + "step": 17120 + }, + { + "epoch": 34.42, + "learning_rate": 5.711000000000001e-05, + "loss": 4.3821, + "step": 17140 + }, + { + "epoch": 34.46, + "learning_rate": 5.7176666666666664e-05, + "loss": 4.3408, + "step": 17160 + }, + { + "epoch": 34.5, + "learning_rate": 5.7243333333333335e-05, + "loss": 4.3607, + "step": 17180 + }, + { + "epoch": 34.54, + "learning_rate": 5.7310000000000005e-05, + "loss": 4.3704, + "step": 17200 + }, + { + "epoch": 34.58, + "learning_rate": 5.737666666666667e-05, + "loss": 4.364, + "step": 17220 + }, + { + "epoch": 34.62, + "learning_rate": 5.744333333333334e-05, + "loss": 4.3566, + "step": 17240 + }, + { + "epoch": 34.66, + "learning_rate": 5.7509999999999997e-05, + "loss": 4.3506, + "step": 17260 + }, + { + "epoch": 34.7, + "learning_rate": 5.757666666666667e-05, + "loss": 4.3562, + "step": 17280 + }, + { + "epoch": 34.74, + "learning_rate": 5.764333333333334e-05, + "loss": 4.3597, + "step": 17300 + }, + { + "epoch": 34.78, + "learning_rate": 5.7709999999999995e-05, + "loss": 4.3516, + "step": 17320 + }, + { + "epoch": 34.82, + "learning_rate": 5.7776666666666665e-05, + "loss": 4.3441, + "step": 17340 + }, + { + "epoch": 34.86, + "learning_rate": 5.7843333333333336e-05, + "loss": 4.3676, + "step": 17360 + }, + { + "epoch": 34.9, + "learning_rate": 5.791e-05, + "loss": 4.3512, + "step": 17380 + }, + { + "epoch": 34.94, + "learning_rate": 5.797666666666667e-05, + "loss": 4.3619, + "step": 17400 + }, + { + "epoch": 34.98, + "learning_rate": 5.804333333333334e-05, + "loss": 4.3475, + "step": 17420 + }, + { + "epoch": 35.02, + "learning_rate": 5.811e-05, + "loss": 4.3188, + "step": 17440 + }, + { + "epoch": 35.06, + "learning_rate": 5.817666666666667e-05, + "loss": 4.291, + "step": 17460 + }, + { + "epoch": 35.1, + "learning_rate": 5.824333333333334e-05, + "loss": 4.3046, + "step": 17480 + }, + { + "epoch": 35.14, + "learning_rate": 5.8309999999999996e-05, + "loss": 4.3014, + "step": 17500 + }, + { + "epoch": 35.18, + "learning_rate": 5.8376666666666666e-05, + "loss": 4.3016, + "step": 17520 + }, + { + "epoch": 35.22, + "learning_rate": 5.844333333333334e-05, + "loss": 4.3051, + "step": 17540 + }, + { + "epoch": 35.26, + "learning_rate": 5.851e-05, + "loss": 4.3054, + "step": 17560 + }, + { + "epoch": 35.3, + "learning_rate": 5.857666666666667e-05, + "loss": 4.314, + "step": 17580 + }, + { + "epoch": 35.34, + "learning_rate": 5.864333333333334e-05, + "loss": 4.3073, + "step": 17600 + }, + { + "epoch": 35.38, + "learning_rate": 5.871e-05, + "loss": 4.2885, + "step": 17620 + }, + { + "epoch": 35.42, + "learning_rate": 5.877666666666667e-05, + "loss": 4.3109, + "step": 17640 + }, + { + "epoch": 35.46, + "learning_rate": 5.884333333333334e-05, + "loss": 4.2974, + "step": 17660 + }, + { + "epoch": 35.5, + "learning_rate": 5.891e-05, + "loss": 4.2932, + "step": 17680 + }, + { + "epoch": 35.54, + "learning_rate": 5.897666666666667e-05, + "loss": 4.3302, + "step": 17700 + }, + { + "epoch": 35.58, + "learning_rate": 5.904333333333334e-05, + "loss": 4.303, + "step": 17720 + }, + { + "epoch": 35.62, + "learning_rate": 5.911e-05, + "loss": 4.2902, + "step": 17740 + }, + { + "epoch": 35.66, + "learning_rate": 5.917666666666667e-05, + "loss": 4.3103, + "step": 17760 + }, + { + "epoch": 35.7, + "learning_rate": 5.924333333333334e-05, + "loss": 4.3037, + "step": 17780 + }, + { + "epoch": 35.74, + "learning_rate": 5.931e-05, + "loss": 4.2967, + "step": 17800 + }, + { + "epoch": 35.78, + "learning_rate": 5.937666666666667e-05, + "loss": 4.3096, + "step": 17820 + }, + { + "epoch": 35.82, + "learning_rate": 5.944333333333334e-05, + "loss": 4.3128, + "step": 17840 + }, + { + "epoch": 35.86, + "learning_rate": 5.951e-05, + "loss": 4.3163, + "step": 17860 + }, + { + "epoch": 35.9, + "learning_rate": 5.957666666666667e-05, + "loss": 4.2994, + "step": 17880 + }, + { + "epoch": 35.94, + "learning_rate": 5.964333333333334e-05, + "loss": 4.3078, + "step": 17900 + }, + { + "epoch": 35.98, + "learning_rate": 5.971e-05, + "loss": 4.2918, + "step": 17920 + }, + { + "epoch": 36.02, + "learning_rate": 5.9776666666666666e-05, + "loss": 4.2625, + "step": 17940 + }, + { + "epoch": 36.06, + "learning_rate": 5.984333333333334e-05, + "loss": 4.2644, + "step": 17960 + }, + { + "epoch": 36.1, + "learning_rate": 5.991e-05, + "loss": 4.2557, + "step": 17980 + }, + { + "epoch": 36.14, + "learning_rate": 5.997666666666667e-05, + "loss": 4.2426, + "step": 18000 + }, + { + "epoch": 36.14, + "eval_loss": 4.198193073272705, + "eval_runtime": 32.4417, + "eval_samples_per_second": 23.612, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0028524994513553937, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0, + "step": 18000 + }, + { + "epoch": 36.18, + "learning_rate": 6.004333333333334e-05, + "loss": 4.2724, + "step": 18020 + }, + { + "epoch": 36.22, + "learning_rate": 6.011e-05, + "loss": 4.2565, + "step": 18040 + }, + { + "epoch": 36.27, + "learning_rate": 6.017666666666667e-05, + "loss": 4.2645, + "step": 18060 + }, + { + "epoch": 36.31, + "learning_rate": 6.024333333333334e-05, + "loss": 4.2643, + "step": 18080 + }, + { + "epoch": 36.35, + "learning_rate": 6.031e-05, + "loss": 4.2492, + "step": 18100 + }, + { + "epoch": 36.39, + "learning_rate": 6.037666666666667e-05, + "loss": 4.2445, + "step": 18120 + }, + { + "epoch": 36.43, + "learning_rate": 6.044333333333334e-05, + "loss": 4.2449, + "step": 18140 + }, + { + "epoch": 36.47, + "learning_rate": 6.051e-05, + "loss": 4.267, + "step": 18160 + }, + { + "epoch": 36.51, + "learning_rate": 6.057666666666667e-05, + "loss": 4.2402, + "step": 18180 + }, + { + "epoch": 36.55, + "learning_rate": 6.064333333333334e-05, + "loss": 4.2401, + "step": 18200 + }, + { + "epoch": 36.59, + "learning_rate": 6.071e-05, + "loss": 4.2587, + "step": 18220 + }, + { + "epoch": 36.63, + "learning_rate": 6.077666666666667e-05, + "loss": 4.2426, + "step": 18240 + }, + { + "epoch": 36.67, + "learning_rate": 6.084333333333334e-05, + "loss": 4.2459, + "step": 18260 + }, + { + "epoch": 36.71, + "learning_rate": 6.091e-05, + "loss": 4.2525, + "step": 18280 + }, + { + "epoch": 36.75, + "learning_rate": 6.097666666666667e-05, + "loss": 4.231, + "step": 18300 + }, + { + "epoch": 36.79, + "learning_rate": 6.104333333333334e-05, + "loss": 4.2347, + "step": 18320 + }, + { + "epoch": 36.83, + "learning_rate": 6.111e-05, + "loss": 4.2379, + "step": 18340 + }, + { + "epoch": 36.87, + "learning_rate": 6.117666666666667e-05, + "loss": 4.246, + "step": 18360 + }, + { + "epoch": 36.91, + "learning_rate": 6.124333333333334e-05, + "loss": 4.2606, + "step": 18380 + }, + { + "epoch": 36.95, + "learning_rate": 6.131e-05, + "loss": 4.237, + "step": 18400 + }, + { + "epoch": 36.99, + "learning_rate": 6.137666666666667e-05, + "loss": 4.2194, + "step": 18420 + }, + { + "epoch": 37.03, + "learning_rate": 6.144333333333334e-05, + "loss": 4.1954, + "step": 18440 + }, + { + "epoch": 37.07, + "learning_rate": 6.151e-05, + "loss": 4.1866, + "step": 18460 + }, + { + "epoch": 37.11, + "learning_rate": 6.157666666666668e-05, + "loss": 4.2049, + "step": 18480 + }, + { + "epoch": 37.15, + "learning_rate": 6.164333333333334e-05, + "loss": 4.1978, + "step": 18500 + }, + { + "epoch": 37.19, + "learning_rate": 6.171e-05, + "loss": 4.1877, + "step": 18520 + }, + { + "epoch": 37.23, + "learning_rate": 6.177666666666667e-05, + "loss": 4.199, + "step": 18540 + }, + { + "epoch": 37.27, + "learning_rate": 6.184333333333333e-05, + "loss": 4.2134, + "step": 18560 + }, + { + "epoch": 37.31, + "learning_rate": 6.191e-05, + "loss": 4.2168, + "step": 18580 + }, + { + "epoch": 37.35, + "learning_rate": 6.197666666666667e-05, + "loss": 4.2081, + "step": 18600 + }, + { + "epoch": 37.39, + "learning_rate": 6.204333333333334e-05, + "loss": 4.2021, + "step": 18620 + }, + { + "epoch": 37.43, + "learning_rate": 6.211e-05, + "loss": 4.1857, + "step": 18640 + }, + { + "epoch": 37.47, + "learning_rate": 6.217666666666668e-05, + "loss": 4.2129, + "step": 18660 + }, + { + "epoch": 37.51, + "learning_rate": 6.224333333333333e-05, + "loss": 4.1929, + "step": 18680 + }, + { + "epoch": 37.55, + "learning_rate": 6.231e-05, + "loss": 4.1866, + "step": 18700 + }, + { + "epoch": 37.59, + "learning_rate": 6.237666666666667e-05, + "loss": 4.2012, + "step": 18720 + }, + { + "epoch": 37.63, + "learning_rate": 6.244333333333333e-05, + "loss": 4.1875, + "step": 18740 + }, + { + "epoch": 37.67, + "learning_rate": 6.251e-05, + "loss": 4.1891, + "step": 18760 + }, + { + "epoch": 37.71, + "learning_rate": 6.257666666666667e-05, + "loss": 4.1939, + "step": 18780 + }, + { + "epoch": 37.75, + "learning_rate": 6.264333333333334e-05, + "loss": 4.1932, + "step": 18800 + }, + { + "epoch": 37.79, + "learning_rate": 6.271e-05, + "loss": 4.1955, + "step": 18820 + }, + { + "epoch": 37.83, + "learning_rate": 6.277666666666668e-05, + "loss": 4.1946, + "step": 18840 + }, + { + "epoch": 37.87, + "learning_rate": 6.284333333333333e-05, + "loss": 4.2007, + "step": 18860 + }, + { + "epoch": 37.91, + "learning_rate": 6.291e-05, + "loss": 4.1935, + "step": 18880 + }, + { + "epoch": 37.95, + "learning_rate": 6.297666666666667e-05, + "loss": 4.1898, + "step": 18900 + }, + { + "epoch": 37.99, + "learning_rate": 6.304333333333333e-05, + "loss": 4.1893, + "step": 18920 + }, + { + "epoch": 38.03, + "learning_rate": 6.311e-05, + "loss": 4.1452, + "step": 18940 + }, + { + "epoch": 38.07, + "learning_rate": 6.317666666666667e-05, + "loss": 4.1223, + "step": 18960 + }, + { + "epoch": 38.11, + "learning_rate": 6.324333333333334e-05, + "loss": 4.1634, + "step": 18980 + }, + { + "epoch": 38.15, + "learning_rate": 6.331e-05, + "loss": 4.1555, + "step": 19000 + }, + { + "epoch": 38.15, + "eval_loss": 4.106661319732666, + "eval_runtime": 32.7512, + "eval_samples_per_second": 23.388, + "eval_steps_per_second": 0.122, + "eval_tse_ndup": 0.002448615240441789, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0, + "step": 19000 + }, + { + "epoch": 38.19, + "learning_rate": 6.337666666666668e-05, + "loss": 4.1348, + "step": 19020 + }, + { + "epoch": 38.23, + "learning_rate": 6.344333333333333e-05, + "loss": 4.1313, + "step": 19040 + }, + { + "epoch": 38.27, + "learning_rate": 6.351000000000001e-05, + "loss": 4.1251, + "step": 19060 + }, + { + "epoch": 38.31, + "learning_rate": 6.357666666666667e-05, + "loss": 4.1353, + "step": 19080 + }, + { + "epoch": 38.35, + "learning_rate": 6.364333333333333e-05, + "loss": 4.1662, + "step": 19100 + }, + { + "epoch": 38.39, + "learning_rate": 6.371e-05, + "loss": 4.1618, + "step": 19120 + }, + { + "epoch": 38.43, + "learning_rate": 6.377666666666668e-05, + "loss": 4.1583, + "step": 19140 + }, + { + "epoch": 38.47, + "learning_rate": 6.384333333333334e-05, + "loss": 4.1558, + "step": 19160 + }, + { + "epoch": 38.51, + "learning_rate": 6.391e-05, + "loss": 4.1593, + "step": 19180 + }, + { + "epoch": 38.55, + "learning_rate": 6.397666666666668e-05, + "loss": 4.1346, + "step": 19200 + }, + { + "epoch": 38.59, + "learning_rate": 6.404333333333333e-05, + "loss": 4.1476, + "step": 19220 + }, + { + "epoch": 38.63, + "learning_rate": 6.411000000000001e-05, + "loss": 4.1533, + "step": 19240 + }, + { + "epoch": 38.67, + "learning_rate": 6.417666666666667e-05, + "loss": 4.167, + "step": 19260 + }, + { + "epoch": 38.71, + "learning_rate": 6.424333333333334e-05, + "loss": 4.1483, + "step": 19280 + }, + { + "epoch": 38.76, + "learning_rate": 6.431e-05, + "loss": 4.1324, + "step": 19300 + }, + { + "epoch": 38.8, + "learning_rate": 6.437666666666668e-05, + "loss": 4.1519, + "step": 19320 + }, + { + "epoch": 38.84, + "learning_rate": 6.444333333333333e-05, + "loss": 4.1457, + "step": 19340 + }, + { + "epoch": 38.88, + "learning_rate": 6.451e-05, + "loss": 4.1255, + "step": 19360 + }, + { + "epoch": 38.92, + "learning_rate": 6.457333333333334e-05, + "loss": 4.1573, + "step": 19380 + }, + { + "epoch": 38.96, + "learning_rate": 6.464e-05, + "loss": 4.1486, + "step": 19400 + }, + { + "epoch": 39.0, + "learning_rate": 6.470666666666667e-05, + "loss": 4.1297, + "step": 19420 + }, + { + "epoch": 39.04, + "learning_rate": 6.477333333333335e-05, + "loss": 4.0869, + "step": 19440 + }, + { + "epoch": 39.08, + "learning_rate": 6.484e-05, + "loss": 4.0962, + "step": 19460 + }, + { + "epoch": 39.12, + "learning_rate": 6.490666666666667e-05, + "loss": 4.0775, + "step": 19480 + }, + { + "epoch": 39.16, + "learning_rate": 6.497333333333334e-05, + "loss": 4.0859, + "step": 19500 + }, + { + "epoch": 39.2, + "learning_rate": 6.504e-05, + "loss": 4.0939, + "step": 19520 + }, + { + "epoch": 39.24, + "learning_rate": 6.510666666666668e-05, + "loss": 4.1048, + "step": 19540 + }, + { + "epoch": 39.28, + "learning_rate": 6.517333333333334e-05, + "loss": 4.0989, + "step": 19560 + }, + { + "epoch": 39.32, + "learning_rate": 6.524e-05, + "loss": 4.0775, + "step": 19580 + }, + { + "epoch": 39.36, + "learning_rate": 6.530666666666667e-05, + "loss": 4.1095, + "step": 19600 + }, + { + "epoch": 39.4, + "learning_rate": 6.537333333333335e-05, + "loss": 4.093, + "step": 19620 + }, + { + "epoch": 39.44, + "learning_rate": 6.544e-05, + "loss": 4.1023, + "step": 19640 + }, + { + "epoch": 39.48, + "learning_rate": 6.550666666666667e-05, + "loss": 4.0857, + "step": 19660 + }, + { + "epoch": 39.52, + "learning_rate": 6.557333333333332e-05, + "loss": 4.1066, + "step": 19680 + }, + { + "epoch": 39.56, + "learning_rate": 6.564e-05, + "loss": 4.1177, + "step": 19700 + }, + { + "epoch": 39.6, + "learning_rate": 6.570666666666667e-05, + "loss": 4.1041, + "step": 19720 + }, + { + "epoch": 39.64, + "learning_rate": 6.577333333333333e-05, + "loss": 4.0998, + "step": 19740 + }, + { + "epoch": 39.68, + "learning_rate": 6.584e-05, + "loss": 4.0898, + "step": 19760 + }, + { + "epoch": 39.72, + "learning_rate": 6.590666666666667e-05, + "loss": 4.1224, + "step": 19780 + }, + { + "epoch": 39.76, + "learning_rate": 6.597333333333333e-05, + "loss": 4.1242, + "step": 19800 + }, + { + "epoch": 39.8, + "learning_rate": 6.604e-05, + "loss": 4.0929, + "step": 19820 + }, + { + "epoch": 39.84, + "learning_rate": 6.610666666666668e-05, + "loss": 4.0869, + "step": 19840 + }, + { + "epoch": 39.88, + "learning_rate": 6.617333333333333e-05, + "loss": 4.096, + "step": 19860 + }, + { + "epoch": 39.92, + "learning_rate": 6.624e-05, + "loss": 4.1162, + "step": 19880 + }, + { + "epoch": 39.96, + "learning_rate": 6.630666666666667e-05, + "loss": 4.0646, + "step": 19900 + }, + { + "epoch": 40.0, + "learning_rate": 6.637333333333333e-05, + "loss": 4.1091, + "step": 19920 + }, + { + "epoch": 40.04, + "learning_rate": 6.644000000000001e-05, + "loss": 4.0301, + "step": 19940 + }, + { + "epoch": 40.08, + "learning_rate": 6.650666666666667e-05, + "loss": 4.0443, + "step": 19960 + }, + { + "epoch": 40.12, + "learning_rate": 6.657333333333334e-05, + "loss": 4.0432, + "step": 19980 + }, + { + "epoch": 40.16, + "learning_rate": 6.664e-05, + "loss": 4.0319, + "step": 20000 + }, + { + "epoch": 40.16, + "eval_loss": 4.018675327301025, + "eval_runtime": 32.4719, + "eval_samples_per_second": 23.59, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0033304058197341554, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.2664635915288658e-06, + "step": 20000 + }, + { + "epoch": 40.2, + "learning_rate": 6.670666666666668e-05, + "loss": 4.0535, + "step": 20020 + }, + { + "epoch": 40.24, + "learning_rate": 6.677333333333333e-05, + "loss": 4.043, + "step": 20040 + }, + { + "epoch": 40.28, + "learning_rate": 6.684e-05, + "loss": 4.0428, + "step": 20060 + }, + { + "epoch": 40.32, + "learning_rate": 6.690666666666667e-05, + "loss": 4.0547, + "step": 20080 + }, + { + "epoch": 40.36, + "learning_rate": 6.697333333333333e-05, + "loss": 4.0274, + "step": 20100 + }, + { + "epoch": 40.4, + "learning_rate": 6.704000000000001e-05, + "loss": 4.0534, + "step": 20120 + }, + { + "epoch": 40.44, + "learning_rate": 6.710666666666667e-05, + "loss": 4.0354, + "step": 20140 + }, + { + "epoch": 40.48, + "learning_rate": 6.717333333333334e-05, + "loss": 4.06, + "step": 20160 + }, + { + "epoch": 40.52, + "learning_rate": 6.724e-05, + "loss": 4.0473, + "step": 20180 + }, + { + "epoch": 40.56, + "learning_rate": 6.730666666666668e-05, + "loss": 4.0405, + "step": 20200 + }, + { + "epoch": 40.6, + "learning_rate": 6.737333333333333e-05, + "loss": 4.0629, + "step": 20220 + }, + { + "epoch": 40.64, + "learning_rate": 6.744e-05, + "loss": 4.0669, + "step": 20240 + }, + { + "epoch": 40.68, + "learning_rate": 6.750666666666667e-05, + "loss": 4.0893, + "step": 20260 + }, + { + "epoch": 40.72, + "learning_rate": 6.757333333333333e-05, + "loss": 4.055, + "step": 20280 + }, + { + "epoch": 40.76, + "learning_rate": 6.764000000000001e-05, + "loss": 4.0617, + "step": 20300 + }, + { + "epoch": 40.8, + "learning_rate": 6.770666666666667e-05, + "loss": 4.0506, + "step": 20320 + }, + { + "epoch": 40.84, + "learning_rate": 6.777333333333334e-05, + "loss": 4.0448, + "step": 20340 + }, + { + "epoch": 40.88, + "learning_rate": 6.784e-05, + "loss": 4.0524, + "step": 20360 + }, + { + "epoch": 40.92, + "learning_rate": 6.790666666666668e-05, + "loss": 4.0546, + "step": 20380 + }, + { + "epoch": 40.96, + "learning_rate": 6.797333333333333e-05, + "loss": 4.0277, + "step": 20400 + }, + { + "epoch": 41.0, + "learning_rate": 6.804e-05, + "loss": 4.0433, + "step": 20420 + }, + { + "epoch": 41.04, + "learning_rate": 6.810666666666667e-05, + "loss": 3.9592, + "step": 20440 + }, + { + "epoch": 41.08, + "learning_rate": 6.817333333333333e-05, + "loss": 3.9922, + "step": 20460 + }, + { + "epoch": 41.12, + "learning_rate": 6.824e-05, + "loss": 3.9995, + "step": 20480 + }, + { + "epoch": 41.16, + "learning_rate": 6.830666666666667e-05, + "loss": 3.9954, + "step": 20500 + }, + { + "epoch": 41.2, + "learning_rate": 6.837333333333334e-05, + "loss": 4.0018, + "step": 20520 + }, + { + "epoch": 41.24, + "learning_rate": 6.844e-05, + "loss": 4.0142, + "step": 20540 + }, + { + "epoch": 41.29, + "learning_rate": 6.850666666666668e-05, + "loss": 4.0254, + "step": 20560 + }, + { + "epoch": 41.33, + "learning_rate": 6.857333333333333e-05, + "loss": 3.9994, + "step": 20580 + }, + { + "epoch": 41.37, + "learning_rate": 6.864000000000001e-05, + "loss": 4.0269, + "step": 20600 + }, + { + "epoch": 41.41, + "learning_rate": 6.870666666666667e-05, + "loss": 4.0123, + "step": 20620 + }, + { + "epoch": 41.45, + "learning_rate": 6.877333333333333e-05, + "loss": 4.0168, + "step": 20640 + }, + { + "epoch": 41.49, + "learning_rate": 6.884e-05, + "loss": 4.0047, + "step": 20660 + }, + { + "epoch": 41.53, + "learning_rate": 6.890666666666668e-05, + "loss": 4.0189, + "step": 20680 + }, + { + "epoch": 41.57, + "learning_rate": 6.897333333333334e-05, + "loss": 3.9824, + "step": 20700 + }, + { + "epoch": 41.61, + "learning_rate": 6.904e-05, + "loss": 3.9982, + "step": 20720 + }, + { + "epoch": 41.65, + "learning_rate": 6.910666666666668e-05, + "loss": 4.0346, + "step": 20740 + }, + { + "epoch": 41.69, + "learning_rate": 6.917333333333333e-05, + "loss": 4.0127, + "step": 20760 + }, + { + "epoch": 41.73, + "learning_rate": 6.924000000000001e-05, + "loss": 3.9963, + "step": 20780 + }, + { + "epoch": 41.77, + "learning_rate": 6.930666666666667e-05, + "loss": 4.0061, + "step": 20800 + }, + { + "epoch": 41.81, + "learning_rate": 6.937333333333334e-05, + "loss": 4.0078, + "step": 20820 + }, + { + "epoch": 41.85, + "learning_rate": 6.944e-05, + "loss": 3.9968, + "step": 20840 + }, + { + "epoch": 41.89, + "learning_rate": 6.950666666666668e-05, + "loss": 4.0199, + "step": 20860 + }, + { + "epoch": 41.93, + "learning_rate": 6.957333333333334e-05, + "loss": 4.0061, + "step": 20880 + }, + { + "epoch": 41.97, + "learning_rate": 6.964e-05, + "loss": 4.0063, + "step": 20900 + }, + { + "epoch": 42.01, + "learning_rate": 6.970666666666667e-05, + "loss": 3.9757, + "step": 20920 + }, + { + "epoch": 42.05, + "learning_rate": 6.977333333333333e-05, + "loss": 3.9581, + "step": 20940 + }, + { + "epoch": 42.09, + "learning_rate": 6.984000000000001e-05, + "loss": 3.9491, + "step": 20960 + }, + { + "epoch": 42.13, + "learning_rate": 6.990666666666666e-05, + "loss": 3.9629, + "step": 20980 + }, + { + "epoch": 42.17, + "learning_rate": 6.997333333333334e-05, + "loss": 3.9731, + "step": 21000 + }, + { + "epoch": 42.17, + "eval_loss": 3.9499671459198, + "eval_runtime": 32.528, + "eval_samples_per_second": 23.549, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0027746028120584306, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0, + "step": 21000 + }, + { + "epoch": 42.21, + "learning_rate": 7.004e-05, + "loss": 3.9347, + "step": 21020 + }, + { + "epoch": 42.25, + "learning_rate": 7.010666666666666e-05, + "loss": 3.9578, + "step": 21040 + }, + { + "epoch": 42.29, + "learning_rate": 7.017333333333334e-05, + "loss": 3.9593, + "step": 21060 + }, + { + "epoch": 42.33, + "learning_rate": 7.024e-05, + "loss": 3.9154, + "step": 21080 + }, + { + "epoch": 42.37, + "learning_rate": 7.030666666666667e-05, + "loss": 3.9559, + "step": 21100 + }, + { + "epoch": 42.41, + "learning_rate": 7.037333333333333e-05, + "loss": 3.9721, + "step": 21120 + }, + { + "epoch": 42.45, + "learning_rate": 7.044000000000001e-05, + "loss": 3.9571, + "step": 21140 + }, + { + "epoch": 42.49, + "learning_rate": 7.050666666666666e-05, + "loss": 3.9844, + "step": 21160 + }, + { + "epoch": 42.53, + "learning_rate": 7.057333333333334e-05, + "loss": 3.9576, + "step": 21180 + }, + { + "epoch": 42.57, + "learning_rate": 7.064e-05, + "loss": 3.9654, + "step": 21200 + }, + { + "epoch": 42.61, + "learning_rate": 7.070666666666667e-05, + "loss": 3.9682, + "step": 21220 + }, + { + "epoch": 42.65, + "learning_rate": 7.077333333333334e-05, + "loss": 3.9699, + "step": 21240 + }, + { + "epoch": 42.69, + "learning_rate": 7.084e-05, + "loss": 3.9765, + "step": 21260 + }, + { + "epoch": 42.73, + "learning_rate": 7.090666666666667e-05, + "loss": 3.9662, + "step": 21280 + }, + { + "epoch": 42.77, + "learning_rate": 7.097333333333333e-05, + "loss": 3.9751, + "step": 21300 + }, + { + "epoch": 42.81, + "learning_rate": 7.104000000000001e-05, + "loss": 3.9617, + "step": 21320 + }, + { + "epoch": 42.85, + "learning_rate": 7.110666666666666e-05, + "loss": 3.9642, + "step": 21340 + }, + { + "epoch": 42.89, + "learning_rate": 7.117333333333334e-05, + "loss": 3.9492, + "step": 21360 + }, + { + "epoch": 42.93, + "learning_rate": 7.124e-05, + "loss": 3.9557, + "step": 21380 + }, + { + "epoch": 42.97, + "learning_rate": 7.130333333333334e-05, + "loss": 3.9612, + "step": 21400 + }, + { + "epoch": 43.01, + "learning_rate": 7.137e-05, + "loss": 3.9295, + "step": 21420 + }, + { + "epoch": 43.05, + "learning_rate": 7.143666666666668e-05, + "loss": 3.8931, + "step": 21440 + }, + { + "epoch": 43.09, + "learning_rate": 7.150333333333333e-05, + "loss": 3.908, + "step": 21460 + }, + { + "epoch": 43.13, + "learning_rate": 7.157000000000001e-05, + "loss": 3.9157, + "step": 21480 + }, + { + "epoch": 43.17, + "learning_rate": 7.163666666666667e-05, + "loss": 3.8825, + "step": 21500 + }, + { + "epoch": 43.21, + "learning_rate": 7.170333333333334e-05, + "loss": 3.9151, + "step": 21520 + }, + { + "epoch": 43.25, + "learning_rate": 7.177e-05, + "loss": 3.9145, + "step": 21540 + }, + { + "epoch": 43.29, + "learning_rate": 7.183666666666668e-05, + "loss": 3.9158, + "step": 21560 + }, + { + "epoch": 43.33, + "learning_rate": 7.190333333333334e-05, + "loss": 3.9196, + "step": 21580 + }, + { + "epoch": 43.37, + "learning_rate": 7.197e-05, + "loss": 3.9109, + "step": 21600 + }, + { + "epoch": 43.41, + "learning_rate": 7.203666666666668e-05, + "loss": 3.9166, + "step": 21620 + }, + { + "epoch": 43.45, + "learning_rate": 7.210333333333333e-05, + "loss": 3.9109, + "step": 21640 + }, + { + "epoch": 43.49, + "learning_rate": 7.217000000000001e-05, + "loss": 3.9181, + "step": 21660 + }, + { + "epoch": 43.53, + "learning_rate": 7.223666666666667e-05, + "loss": 3.9112, + "step": 21680 + }, + { + "epoch": 43.57, + "learning_rate": 7.230333333333334e-05, + "loss": 3.9266, + "step": 21700 + }, + { + "epoch": 43.61, + "learning_rate": 7.237e-05, + "loss": 3.9021, + "step": 21720 + }, + { + "epoch": 43.65, + "learning_rate": 7.243666666666668e-05, + "loss": 3.9113, + "step": 21740 + }, + { + "epoch": 43.69, + "learning_rate": 7.250333333333333e-05, + "loss": 3.9242, + "step": 21760 + }, + { + "epoch": 43.73, + "learning_rate": 7.257e-05, + "loss": 3.9302, + "step": 21780 + }, + { + "epoch": 43.78, + "learning_rate": 7.263666666666667e-05, + "loss": 3.9338, + "step": 21800 + }, + { + "epoch": 43.82, + "learning_rate": 7.270333333333333e-05, + "loss": 3.9207, + "step": 21820 + }, + { + "epoch": 43.86, + "learning_rate": 7.277000000000001e-05, + "loss": 3.9152, + "step": 21840 + }, + { + "epoch": 43.9, + "learning_rate": 7.283666666666667e-05, + "loss": 3.9263, + "step": 21860 + }, + { + "epoch": 43.94, + "learning_rate": 7.290333333333334e-05, + "loss": 3.9504, + "step": 21880 + }, + { + "epoch": 43.98, + "learning_rate": 7.297e-05, + "loss": 3.9133, + "step": 21900 + }, + { + "epoch": 44.02, + "learning_rate": 7.303666666666668e-05, + "loss": 3.893, + "step": 21920 + }, + { + "epoch": 44.06, + "learning_rate": 7.310333333333333e-05, + "loss": 3.8788, + "step": 21940 + }, + { + "epoch": 44.1, + "learning_rate": 7.317e-05, + "loss": 3.8724, + "step": 21960 + }, + { + "epoch": 44.14, + "learning_rate": 7.323666666666667e-05, + "loss": 3.8612, + "step": 21980 + }, + { + "epoch": 44.18, + "learning_rate": 7.330333333333333e-05, + "loss": 3.8646, + "step": 22000 + }, + { + "epoch": 44.18, + "eval_loss": 3.8803529739379883, + "eval_runtime": 32.5993, + "eval_samples_per_second": 23.497, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0021499095711561477, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.3598781549173194e-05, + "step": 22000 + }, + { + "epoch": 44.22, + "learning_rate": 7.337000000000001e-05, + "loss": 3.8711, + "step": 22020 + }, + { + "epoch": 44.26, + "learning_rate": 7.343666666666666e-05, + "loss": 3.8724, + "step": 22040 + }, + { + "epoch": 44.3, + "learning_rate": 7.350333333333334e-05, + "loss": 3.8671, + "step": 22060 + }, + { + "epoch": 44.34, + "learning_rate": 7.357e-05, + "loss": 3.8693, + "step": 22080 + }, + { + "epoch": 44.38, + "learning_rate": 7.363666666666667e-05, + "loss": 3.8917, + "step": 22100 + }, + { + "epoch": 44.42, + "learning_rate": 7.370333333333333e-05, + "loss": 3.8865, + "step": 22120 + }, + { + "epoch": 44.46, + "learning_rate": 7.377000000000001e-05, + "loss": 3.8756, + "step": 22140 + }, + { + "epoch": 44.5, + "learning_rate": 7.383666666666667e-05, + "loss": 3.8787, + "step": 22160 + }, + { + "epoch": 44.54, + "learning_rate": 7.390333333333333e-05, + "loss": 3.8648, + "step": 22180 + }, + { + "epoch": 44.58, + "learning_rate": 7.397000000000001e-05, + "loss": 3.8939, + "step": 22200 + }, + { + "epoch": 44.62, + "learning_rate": 7.403666666666666e-05, + "loss": 3.8842, + "step": 22220 + }, + { + "epoch": 44.66, + "learning_rate": 7.410333333333334e-05, + "loss": 3.8496, + "step": 22240 + }, + { + "epoch": 44.7, + "learning_rate": 7.417e-05, + "loss": 3.8817, + "step": 22260 + }, + { + "epoch": 44.74, + "learning_rate": 7.423666666666667e-05, + "loss": 3.8806, + "step": 22280 + }, + { + "epoch": 44.78, + "learning_rate": 7.430333333333333e-05, + "loss": 3.8792, + "step": 22300 + }, + { + "epoch": 44.82, + "learning_rate": 7.437000000000001e-05, + "loss": 3.8775, + "step": 22320 + }, + { + "epoch": 44.86, + "learning_rate": 7.443666666666667e-05, + "loss": 3.887, + "step": 22340 + }, + { + "epoch": 44.9, + "learning_rate": 7.450333333333334e-05, + "loss": 3.9009, + "step": 22360 + }, + { + "epoch": 44.94, + "learning_rate": 7.457000000000001e-05, + "loss": 3.8633, + "step": 22380 + }, + { + "epoch": 44.98, + "learning_rate": 7.463666666666666e-05, + "loss": 3.8754, + "step": 22400 + }, + { + "epoch": 45.02, + "learning_rate": 7.470333333333334e-05, + "loss": 3.8348, + "step": 22420 + }, + { + "epoch": 45.06, + "learning_rate": 7.477e-05, + "loss": 3.8494, + "step": 22440 + }, + { + "epoch": 45.1, + "learning_rate": 7.483666666666667e-05, + "loss": 3.8215, + "step": 22460 + }, + { + "epoch": 45.14, + "learning_rate": 7.490333333333333e-05, + "loss": 3.8003, + "step": 22480 + }, + { + "epoch": 45.18, + "learning_rate": 7.497000000000001e-05, + "loss": 3.817, + "step": 22500 + }, + { + "epoch": 45.22, + "learning_rate": 7.503666666666667e-05, + "loss": 3.8352, + "step": 22520 + }, + { + "epoch": 45.26, + "learning_rate": 7.510333333333334e-05, + "loss": 3.8496, + "step": 22540 + }, + { + "epoch": 45.3, + "learning_rate": 7.517000000000001e-05, + "loss": 3.8394, + "step": 22560 + }, + { + "epoch": 45.34, + "learning_rate": 7.523666666666666e-05, + "loss": 3.8336, + "step": 22580 + }, + { + "epoch": 45.38, + "learning_rate": 7.530333333333334e-05, + "loss": 3.817, + "step": 22600 + }, + { + "epoch": 45.42, + "learning_rate": 7.537e-05, + "loss": 3.8421, + "step": 22620 + }, + { + "epoch": 45.46, + "learning_rate": 7.543666666666667e-05, + "loss": 3.8365, + "step": 22640 + }, + { + "epoch": 45.5, + "learning_rate": 7.550333333333333e-05, + "loss": 3.835, + "step": 22660 + }, + { + "epoch": 45.54, + "learning_rate": 7.557000000000001e-05, + "loss": 3.8246, + "step": 22680 + }, + { + "epoch": 45.58, + "learning_rate": 7.563666666666666e-05, + "loss": 3.836, + "step": 22700 + }, + { + "epoch": 45.62, + "learning_rate": 7.570333333333334e-05, + "loss": 3.834, + "step": 22720 + }, + { + "epoch": 45.66, + "learning_rate": 7.577e-05, + "loss": 3.8289, + "step": 22740 + }, + { + "epoch": 45.7, + "learning_rate": 7.583666666666667e-05, + "loss": 3.8492, + "step": 22760 + }, + { + "epoch": 45.74, + "learning_rate": 7.590333333333334e-05, + "loss": 3.8391, + "step": 22780 + }, + { + "epoch": 45.78, + "learning_rate": 7.597e-05, + "loss": 3.8489, + "step": 22800 + }, + { + "epoch": 45.82, + "learning_rate": 7.603666666666667e-05, + "loss": 3.8504, + "step": 22820 + }, + { + "epoch": 45.86, + "learning_rate": 7.610333333333333e-05, + "loss": 3.8382, + "step": 22840 + }, + { + "epoch": 45.9, + "learning_rate": 7.617000000000001e-05, + "loss": 3.8493, + "step": 22860 + }, + { + "epoch": 45.94, + "learning_rate": 7.623666666666666e-05, + "loss": 3.8554, + "step": 22880 + }, + { + "epoch": 45.98, + "learning_rate": 7.630333333333334e-05, + "loss": 3.8416, + "step": 22900 + }, + { + "epoch": 46.02, + "learning_rate": 7.637e-05, + "loss": 3.815, + "step": 22920 + }, + { + "epoch": 46.06, + "learning_rate": 7.643666666666667e-05, + "loss": 3.7739, + "step": 22940 + }, + { + "epoch": 46.1, + "learning_rate": 7.650333333333334e-05, + "loss": 3.7905, + "step": 22960 + }, + { + "epoch": 46.14, + "learning_rate": 7.657000000000001e-05, + "loss": 3.8, + "step": 22980 + }, + { + "epoch": 46.18, + "learning_rate": 7.663666666666667e-05, + "loss": 3.7958, + "step": 23000 + }, + { + "epoch": 46.18, + "eval_loss": 3.8215503692626953, + "eval_runtime": 32.5771, + "eval_samples_per_second": 23.513, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.002767212612605263, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.799390774586597e-06, + "step": 23000 + }, + { + "epoch": 46.22, + "learning_rate": 7.670333333333333e-05, + "loss": 3.7742, + "step": 23020 + }, + { + "epoch": 46.27, + "learning_rate": 7.677000000000001e-05, + "loss": 3.7743, + "step": 23040 + }, + { + "epoch": 46.31, + "learning_rate": 7.683666666666666e-05, + "loss": 3.7995, + "step": 23060 + }, + { + "epoch": 46.35, + "learning_rate": 7.690333333333334e-05, + "loss": 3.778, + "step": 23080 + }, + { + "epoch": 46.39, + "learning_rate": 7.697e-05, + "loss": 3.7773, + "step": 23100 + }, + { + "epoch": 46.43, + "learning_rate": 7.703666666666667e-05, + "loss": 3.804, + "step": 23120 + }, + { + "epoch": 46.47, + "learning_rate": 7.710333333333334e-05, + "loss": 3.8084, + "step": 23140 + }, + { + "epoch": 46.51, + "learning_rate": 7.717000000000001e-05, + "loss": 3.7656, + "step": 23160 + }, + { + "epoch": 46.55, + "learning_rate": 7.723666666666667e-05, + "loss": 3.8094, + "step": 23180 + }, + { + "epoch": 46.59, + "learning_rate": 7.730333333333334e-05, + "loss": 3.8024, + "step": 23200 + }, + { + "epoch": 46.63, + "learning_rate": 7.737000000000001e-05, + "loss": 3.8068, + "step": 23220 + }, + { + "epoch": 46.67, + "learning_rate": 7.743666666666666e-05, + "loss": 3.7976, + "step": 23240 + }, + { + "epoch": 46.71, + "learning_rate": 7.750333333333334e-05, + "loss": 3.8117, + "step": 23260 + }, + { + "epoch": 46.75, + "learning_rate": 7.757e-05, + "loss": 3.7884, + "step": 23280 + }, + { + "epoch": 46.79, + "learning_rate": 7.763666666666667e-05, + "loss": 3.7971, + "step": 23300 + }, + { + "epoch": 46.83, + "learning_rate": 7.770333333333335e-05, + "loss": 3.7913, + "step": 23320 + }, + { + "epoch": 46.87, + "learning_rate": 7.777e-05, + "loss": 3.8081, + "step": 23340 + }, + { + "epoch": 46.91, + "learning_rate": 7.783666666666667e-05, + "loss": 3.8055, + "step": 23360 + }, + { + "epoch": 46.95, + "learning_rate": 7.790333333333334e-05, + "loss": 3.8265, + "step": 23380 + }, + { + "epoch": 46.99, + "learning_rate": 7.797e-05, + "loss": 3.8149, + "step": 23400 + }, + { + "epoch": 47.03, + "learning_rate": 7.803666666666666e-05, + "loss": 3.7513, + "step": 23420 + }, + { + "epoch": 47.07, + "learning_rate": 7.810333333333334e-05, + "loss": 3.7436, + "step": 23440 + }, + { + "epoch": 47.11, + "learning_rate": 7.817e-05, + "loss": 3.7285, + "step": 23460 + }, + { + "epoch": 47.15, + "learning_rate": 7.823666666666667e-05, + "loss": 3.7651, + "step": 23480 + }, + { + "epoch": 47.19, + "learning_rate": 7.830333333333333e-05, + "loss": 3.7491, + "step": 23500 + }, + { + "epoch": 47.23, + "learning_rate": 7.837e-05, + "loss": 3.7566, + "step": 23520 + }, + { + "epoch": 47.27, + "learning_rate": 7.843666666666667e-05, + "loss": 3.7406, + "step": 23540 + }, + { + "epoch": 47.31, + "learning_rate": 7.850333333333334e-05, + "loss": 3.7364, + "step": 23560 + }, + { + "epoch": 47.35, + "learning_rate": 7.856666666666666e-05, + "loss": 3.7503, + "step": 23580 + }, + { + "epoch": 47.39, + "learning_rate": 7.863333333333334e-05, + "loss": 3.7644, + "step": 23600 + }, + { + "epoch": 47.43, + "learning_rate": 7.87e-05, + "loss": 3.7443, + "step": 23620 + }, + { + "epoch": 47.47, + "learning_rate": 7.876666666666667e-05, + "loss": 3.7693, + "step": 23640 + }, + { + "epoch": 47.51, + "learning_rate": 7.883333333333334e-05, + "loss": 3.773, + "step": 23660 + }, + { + "epoch": 47.55, + "learning_rate": 7.890000000000001e-05, + "loss": 3.7728, + "step": 23680 + }, + { + "epoch": 47.59, + "learning_rate": 7.896666666666667e-05, + "loss": 3.7753, + "step": 23700 + }, + { + "epoch": 47.63, + "learning_rate": 7.903333333333333e-05, + "loss": 3.7418, + "step": 23720 + }, + { + "epoch": 47.67, + "learning_rate": 7.910000000000001e-05, + "loss": 3.7626, + "step": 23740 + }, + { + "epoch": 47.71, + "learning_rate": 7.916666666666666e-05, + "loss": 3.7528, + "step": 23760 + }, + { + "epoch": 47.75, + "learning_rate": 7.923333333333334e-05, + "loss": 3.7831, + "step": 23780 + }, + { + "epoch": 47.79, + "learning_rate": 7.93e-05, + "loss": 3.7657, + "step": 23800 + }, + { + "epoch": 47.83, + "learning_rate": 7.936666666666667e-05, + "loss": 3.7565, + "step": 23820 + }, + { + "epoch": 47.87, + "learning_rate": 7.943333333333333e-05, + "loss": 3.7516, + "step": 23840 + }, + { + "epoch": 47.91, + "learning_rate": 7.950000000000001e-05, + "loss": 3.7433, + "step": 23860 + }, + { + "epoch": 47.95, + "learning_rate": 7.956666666666667e-05, + "loss": 3.7644, + "step": 23880 + }, + { + "epoch": 47.99, + "learning_rate": 7.963333333333334e-05, + "loss": 3.7764, + "step": 23900 + }, + { + "epoch": 48.03, + "learning_rate": 7.970000000000001e-05, + "loss": 3.7162, + "step": 23920 + }, + { + "epoch": 48.07, + "learning_rate": 7.976666666666666e-05, + "loss": 3.6797, + "step": 23940 + }, + { + "epoch": 48.11, + "learning_rate": 7.983333333333334e-05, + "loss": 3.7208, + "step": 23960 + }, + { + "epoch": 48.15, + "learning_rate": 7.99e-05, + "loss": 3.7182, + "step": 23980 + }, + { + "epoch": 48.19, + "learning_rate": 7.996666666666667e-05, + "loss": 3.6981, + "step": 24000 + }, + { + "epoch": 48.19, + "eval_loss": 3.7694156169891357, + "eval_runtime": 32.5347, + "eval_samples_per_second": 23.544, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0030633166246210003, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0, + "step": 24000 + }, + { + "epoch": 48.23, + "learning_rate": 8.003333333333333e-05, + "loss": 3.6959, + "step": 24020 + }, + { + "epoch": 48.27, + "learning_rate": 8.010000000000001e-05, + "loss": 3.7227, + "step": 24040 + }, + { + "epoch": 48.31, + "learning_rate": 8.016666666666667e-05, + "loss": 3.708, + "step": 24060 + }, + { + "epoch": 48.35, + "learning_rate": 8.023333333333334e-05, + "loss": 3.7206, + "step": 24080 + }, + { + "epoch": 48.39, + "learning_rate": 8.030000000000001e-05, + "loss": 3.7151, + "step": 24100 + }, + { + "epoch": 48.43, + "learning_rate": 8.036666666666666e-05, + "loss": 3.7136, + "step": 24120 + }, + { + "epoch": 48.47, + "learning_rate": 8.043333333333334e-05, + "loss": 3.7496, + "step": 24140 + }, + { + "epoch": 48.51, + "learning_rate": 8.05e-05, + "loss": 3.7178, + "step": 24160 + }, + { + "epoch": 48.55, + "learning_rate": 8.056666666666667e-05, + "loss": 3.7038, + "step": 24180 + }, + { + "epoch": 48.59, + "learning_rate": 8.063333333333333e-05, + "loss": 3.7042, + "step": 24200 + }, + { + "epoch": 48.63, + "learning_rate": 8.070000000000001e-05, + "loss": 3.7337, + "step": 24220 + }, + { + "epoch": 48.67, + "learning_rate": 8.076666666666667e-05, + "loss": 3.7231, + "step": 24240 + }, + { + "epoch": 48.71, + "learning_rate": 8.083333333333334e-05, + "loss": 3.7348, + "step": 24260 + }, + { + "epoch": 48.76, + "learning_rate": 8.090000000000001e-05, + "loss": 3.7384, + "step": 24280 + }, + { + "epoch": 48.8, + "learning_rate": 8.096666666666667e-05, + "loss": 3.7513, + "step": 24300 + }, + { + "epoch": 48.84, + "learning_rate": 8.103333333333334e-05, + "loss": 3.7508, + "step": 24320 + }, + { + "epoch": 48.88, + "learning_rate": 8.11e-05, + "loss": 3.7425, + "step": 24340 + }, + { + "epoch": 48.92, + "learning_rate": 8.116666666666667e-05, + "loss": 3.7278, + "step": 24360 + }, + { + "epoch": 48.96, + "learning_rate": 8.123333333333333e-05, + "loss": 3.7474, + "step": 24380 + }, + { + "epoch": 49.0, + "learning_rate": 8.13e-05, + "loss": 3.7444, + "step": 24400 + }, + { + "epoch": 49.04, + "learning_rate": 8.136666666666667e-05, + "loss": 3.6663, + "step": 24420 + }, + { + "epoch": 49.08, + "learning_rate": 8.143333333333334e-05, + "loss": 3.6616, + "step": 24440 + }, + { + "epoch": 49.12, + "learning_rate": 8.15e-05, + "loss": 3.6607, + "step": 24460 + }, + { + "epoch": 49.16, + "learning_rate": 8.156666666666667e-05, + "loss": 3.6699, + "step": 24480 + }, + { + "epoch": 49.2, + "learning_rate": 8.163333333333334e-05, + "loss": 3.6584, + "step": 24500 + }, + { + "epoch": 49.24, + "learning_rate": 8.17e-05, + "loss": 3.6967, + "step": 24520 + }, + { + "epoch": 49.28, + "learning_rate": 8.176666666666667e-05, + "loss": 3.6886, + "step": 24540 + }, + { + "epoch": 49.32, + "learning_rate": 8.183333333333333e-05, + "loss": 3.6859, + "step": 24560 + }, + { + "epoch": 49.36, + "learning_rate": 8.19e-05, + "loss": 3.6853, + "step": 24580 + }, + { + "epoch": 49.4, + "learning_rate": 8.196666666666668e-05, + "loss": 3.6781, + "step": 24600 + }, + { + "epoch": 49.44, + "learning_rate": 8.203333333333334e-05, + "loss": 3.6954, + "step": 24620 + }, + { + "epoch": 49.48, + "learning_rate": 8.21e-05, + "loss": 3.677, + "step": 24640 + }, + { + "epoch": 49.52, + "learning_rate": 8.216666666666667e-05, + "loss": 3.6849, + "step": 24660 + }, + { + "epoch": 49.56, + "learning_rate": 8.223333333333334e-05, + "loss": 3.6698, + "step": 24680 + }, + { + "epoch": 49.6, + "learning_rate": 8.23e-05, + "loss": 3.6965, + "step": 24700 + }, + { + "epoch": 49.64, + "learning_rate": 8.236666666666667e-05, + "loss": 3.6824, + "step": 24720 + }, + { + "epoch": 49.68, + "learning_rate": 8.243333333333334e-05, + "loss": 3.715, + "step": 24740 + }, + { + "epoch": 49.72, + "learning_rate": 8.25e-05, + "loss": 3.7037, + "step": 24760 + }, + { + "epoch": 49.76, + "learning_rate": 8.256666666666666e-05, + "loss": 3.7027, + "step": 24780 + }, + { + "epoch": 49.8, + "learning_rate": 8.263333333333334e-05, + "loss": 3.705, + "step": 24800 + }, + { + "epoch": 49.84, + "learning_rate": 8.27e-05, + "loss": 3.6721, + "step": 24820 + }, + { + "epoch": 49.88, + "learning_rate": 8.276666666666667e-05, + "loss": 3.6897, + "step": 24840 + }, + { + "epoch": 49.92, + "learning_rate": 8.283333333333335e-05, + "loss": 3.7083, + "step": 24860 + }, + { + "epoch": 49.96, + "learning_rate": 8.29e-05, + "loss": 3.6997, + "step": 24880 + }, + { + "epoch": 50.0, + "learning_rate": 8.296666666666667e-05, + "loss": 3.7052, + "step": 24900 + }, + { + "epoch": 50.04, + "learning_rate": 8.303333333333334e-05, + "loss": 3.6212, + "step": 24920 + }, + { + "epoch": 50.08, + "learning_rate": 8.31e-05, + "loss": 3.6255, + "step": 24940 + }, + { + "epoch": 50.12, + "learning_rate": 8.316666666666666e-05, + "loss": 3.6267, + "step": 24960 + }, + { + "epoch": 50.16, + "learning_rate": 8.323333333333334e-05, + "loss": 3.635, + "step": 24980 + }, + { + "epoch": 50.2, + "learning_rate": 8.33e-05, + "loss": 3.6487, + "step": 25000 + }, + { + "epoch": 50.2, + "eval_loss": 3.708888053894043, + "eval_runtime": 34.7091, + "eval_samples_per_second": 22.069, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.004554271903409982, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 4.936163036765321e-06, + "step": 25000 + }, + { + "epoch": 50.24, + "learning_rate": 8.336666666666667e-05, + "loss": 3.6509, + "step": 25020 + }, + { + "epoch": 50.28, + "learning_rate": 8.343333333333335e-05, + "loss": 3.6466, + "step": 25040 + }, + { + "epoch": 50.32, + "learning_rate": 8.35e-05, + "loss": 3.6402, + "step": 25060 + }, + { + "epoch": 50.36, + "learning_rate": 8.356666666666667e-05, + "loss": 3.6469, + "step": 25080 + }, + { + "epoch": 50.4, + "learning_rate": 8.363333333333334e-05, + "loss": 3.6667, + "step": 25100 + }, + { + "epoch": 50.44, + "learning_rate": 8.37e-05, + "loss": 3.6638, + "step": 25120 + }, + { + "epoch": 50.48, + "learning_rate": 8.376666666666667e-05, + "loss": 3.6522, + "step": 25140 + }, + { + "epoch": 50.52, + "learning_rate": 8.383333333333334e-05, + "loss": 3.6422, + "step": 25160 + }, + { + "epoch": 50.56, + "learning_rate": 8.39e-05, + "loss": 3.6635, + "step": 25180 + }, + { + "epoch": 50.6, + "learning_rate": 8.396666666666667e-05, + "loss": 3.6736, + "step": 25200 + }, + { + "epoch": 50.64, + "learning_rate": 8.403333333333335e-05, + "loss": 3.6459, + "step": 25220 + }, + { + "epoch": 50.68, + "learning_rate": 8.41e-05, + "loss": 3.6434, + "step": 25240 + }, + { + "epoch": 50.72, + "learning_rate": 8.416666666666668e-05, + "loss": 3.6744, + "step": 25260 + }, + { + "epoch": 50.76, + "learning_rate": 8.423333333333334e-05, + "loss": 3.6646, + "step": 25280 + }, + { + "epoch": 50.8, + "learning_rate": 8.43e-05, + "loss": 3.636, + "step": 25300 + }, + { + "epoch": 50.84, + "learning_rate": 8.436666666666667e-05, + "loss": 3.6692, + "step": 25320 + }, + { + "epoch": 50.88, + "learning_rate": 8.443333333333334e-05, + "loss": 3.6646, + "step": 25340 + }, + { + "epoch": 50.92, + "learning_rate": 8.450000000000001e-05, + "loss": 3.6648, + "step": 25360 + }, + { + "epoch": 50.96, + "learning_rate": 8.456666666666667e-05, + "loss": 3.6507, + "step": 25380 + }, + { + "epoch": 51.0, + "learning_rate": 8.463333333333335e-05, + "loss": 3.6493, + "step": 25400 + }, + { + "epoch": 51.04, + "learning_rate": 8.47e-05, + "loss": 3.5824, + "step": 25420 + }, + { + "epoch": 51.08, + "learning_rate": 8.476666666666668e-05, + "loss": 3.581, + "step": 25440 + }, + { + "epoch": 51.12, + "learning_rate": 8.483333333333334e-05, + "loss": 3.6218, + "step": 25460 + }, + { + "epoch": 51.16, + "learning_rate": 8.49e-05, + "loss": 3.6151, + "step": 25480 + }, + { + "epoch": 51.2, + "learning_rate": 8.496666666666667e-05, + "loss": 3.6108, + "step": 25500 + }, + { + "epoch": 51.24, + "learning_rate": 8.503333333333334e-05, + "loss": 3.6011, + "step": 25520 + }, + { + "epoch": 51.29, + "learning_rate": 8.510000000000001e-05, + "loss": 3.5977, + "step": 25540 + }, + { + "epoch": 51.33, + "learning_rate": 8.516666666666667e-05, + "loss": 3.6228, + "step": 25560 + }, + { + "epoch": 51.37, + "learning_rate": 8.523e-05, + "loss": 3.6207, + "step": 25580 + }, + { + "epoch": 51.41, + "learning_rate": 8.529666666666667e-05, + "loss": 3.606, + "step": 25600 + }, + { + "epoch": 51.45, + "learning_rate": 8.536333333333334e-05, + "loss": 3.6303, + "step": 25620 + }, + { + "epoch": 51.49, + "learning_rate": 8.543e-05, + "loss": 3.6253, + "step": 25640 + }, + { + "epoch": 51.53, + "learning_rate": 8.549666666666666e-05, + "loss": 3.618, + "step": 25660 + }, + { + "epoch": 51.57, + "learning_rate": 8.556333333333334e-05, + "loss": 3.6212, + "step": 25680 + }, + { + "epoch": 51.61, + "learning_rate": 8.563e-05, + "loss": 3.628, + "step": 25700 + }, + { + "epoch": 51.65, + "learning_rate": 8.569666666666667e-05, + "loss": 3.6239, + "step": 25720 + }, + { + "epoch": 51.69, + "learning_rate": 8.576333333333335e-05, + "loss": 3.6402, + "step": 25740 + }, + { + "epoch": 51.73, + "learning_rate": 8.583e-05, + "loss": 3.6343, + "step": 25760 + }, + { + "epoch": 51.77, + "learning_rate": 8.589666666666667e-05, + "loss": 3.6354, + "step": 25780 + }, + { + "epoch": 51.81, + "learning_rate": 8.596333333333334e-05, + "loss": 3.6474, + "step": 25800 + }, + { + "epoch": 51.85, + "learning_rate": 8.603e-05, + "loss": 3.6263, + "step": 25820 + }, + { + "epoch": 51.89, + "learning_rate": 8.609666666666667e-05, + "loss": 3.6385, + "step": 25840 + }, + { + "epoch": 51.93, + "learning_rate": 8.616333333333334e-05, + "loss": 3.635, + "step": 25860 + }, + { + "epoch": 51.97, + "learning_rate": 8.623e-05, + "loss": 3.6145, + "step": 25880 + }, + { + "epoch": 52.01, + "learning_rate": 8.629666666666667e-05, + "loss": 3.6082, + "step": 25900 + }, + { + "epoch": 52.05, + "learning_rate": 8.636333333333335e-05, + "loss": 3.5575, + "step": 25920 + }, + { + "epoch": 52.09, + "learning_rate": 8.643e-05, + "loss": 3.5642, + "step": 25940 + }, + { + "epoch": 52.13, + "learning_rate": 8.649666666666667e-05, + "loss": 3.569, + "step": 25960 + }, + { + "epoch": 52.17, + "learning_rate": 8.656333333333334e-05, + "loss": 3.5663, + "step": 25980 + }, + { + "epoch": 52.21, + "learning_rate": 8.663e-05, + "loss": 3.567, + "step": 26000 + }, + { + "epoch": 52.21, + "eval_loss": 3.665741205215454, + "eval_runtime": 32.3862, + "eval_samples_per_second": 23.652, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.003515021049887783, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.799390774586597e-06, + "step": 26000 + }, + { + "epoch": 52.25, + "learning_rate": 8.669666666666667e-05, + "loss": 3.5662, + "step": 26020 + }, + { + "epoch": 52.29, + "learning_rate": 8.676333333333334e-05, + "loss": 3.5894, + "step": 26040 + }, + { + "epoch": 52.33, + "learning_rate": 8.683e-05, + "loss": 3.5913, + "step": 26060 + }, + { + "epoch": 52.37, + "learning_rate": 8.689666666666667e-05, + "loss": 3.5772, + "step": 26080 + }, + { + "epoch": 52.41, + "learning_rate": 8.696333333333333e-05, + "loss": 3.5737, + "step": 26100 + }, + { + "epoch": 52.45, + "learning_rate": 8.703e-05, + "loss": 3.5932, + "step": 26120 + }, + { + "epoch": 52.49, + "learning_rate": 8.709666666666668e-05, + "loss": 3.5873, + "step": 26140 + }, + { + "epoch": 52.53, + "learning_rate": 8.716333333333334e-05, + "loss": 3.5869, + "step": 26160 + }, + { + "epoch": 52.57, + "learning_rate": 8.723e-05, + "loss": 3.582, + "step": 26180 + }, + { + "epoch": 52.61, + "learning_rate": 8.729666666666667e-05, + "loss": 3.5949, + "step": 26200 + }, + { + "epoch": 52.65, + "learning_rate": 8.736333333333334e-05, + "loss": 3.5939, + "step": 26220 + }, + { + "epoch": 52.69, + "learning_rate": 8.743e-05, + "loss": 3.6013, + "step": 26240 + }, + { + "epoch": 52.73, + "learning_rate": 8.749666666666667e-05, + "loss": 3.6035, + "step": 26260 + }, + { + "epoch": 52.77, + "learning_rate": 8.756333333333334e-05, + "loss": 3.5847, + "step": 26280 + }, + { + "epoch": 52.81, + "learning_rate": 8.763e-05, + "loss": 3.5922, + "step": 26300 + }, + { + "epoch": 52.85, + "learning_rate": 8.769666666666668e-05, + "loss": 3.6056, + "step": 26320 + }, + { + "epoch": 52.89, + "learning_rate": 8.776333333333334e-05, + "loss": 3.6005, + "step": 26340 + }, + { + "epoch": 52.93, + "learning_rate": 8.783e-05, + "loss": 3.5864, + "step": 26360 + }, + { + "epoch": 52.97, + "learning_rate": 8.789666666666667e-05, + "loss": 3.6081, + "step": 26380 + }, + { + "epoch": 53.01, + "learning_rate": 8.796333333333335e-05, + "loss": 3.5767, + "step": 26400 + }, + { + "epoch": 53.05, + "learning_rate": 8.803e-05, + "loss": 3.5176, + "step": 26420 + }, + { + "epoch": 53.09, + "learning_rate": 8.809666666666667e-05, + "loss": 3.5493, + "step": 26440 + }, + { + "epoch": 53.13, + "learning_rate": 8.816333333333334e-05, + "loss": 3.5555, + "step": 26460 + }, + { + "epoch": 53.17, + "learning_rate": 8.823e-05, + "loss": 3.5312, + "step": 26480 + }, + { + "epoch": 53.21, + "learning_rate": 8.829666666666668e-05, + "loss": 3.5566, + "step": 26500 + }, + { + "epoch": 53.25, + "learning_rate": 8.836333333333334e-05, + "loss": 3.54, + "step": 26520 + }, + { + "epoch": 53.29, + "learning_rate": 8.843e-05, + "loss": 3.5346, + "step": 26540 + }, + { + "epoch": 53.33, + "learning_rate": 8.849666666666667e-05, + "loss": 3.5457, + "step": 26560 + }, + { + "epoch": 53.37, + "learning_rate": 8.856333333333335e-05, + "loss": 3.5526, + "step": 26580 + }, + { + "epoch": 53.41, + "learning_rate": 8.863e-05, + "loss": 3.5482, + "step": 26600 + }, + { + "epoch": 53.45, + "learning_rate": 8.869666666666667e-05, + "loss": 3.5507, + "step": 26620 + }, + { + "epoch": 53.49, + "learning_rate": 8.876333333333334e-05, + "loss": 3.5469, + "step": 26640 + }, + { + "epoch": 53.53, + "learning_rate": 8.883e-05, + "loss": 3.5616, + "step": 26660 + }, + { + "epoch": 53.57, + "learning_rate": 8.889666666666668e-05, + "loss": 3.5448, + "step": 26680 + }, + { + "epoch": 53.61, + "learning_rate": 8.896333333333334e-05, + "loss": 3.5699, + "step": 26700 + }, + { + "epoch": 53.65, + "learning_rate": 8.903e-05, + "loss": 3.5541, + "step": 26720 + }, + { + "epoch": 53.69, + "learning_rate": 8.909666666666667e-05, + "loss": 3.5683, + "step": 26740 + }, + { + "epoch": 53.73, + "learning_rate": 8.916333333333335e-05, + "loss": 3.5795, + "step": 26760 + }, + { + "epoch": 53.78, + "learning_rate": 8.923e-05, + "loss": 3.5725, + "step": 26780 + }, + { + "epoch": 53.82, + "learning_rate": 8.929666666666668e-05, + "loss": 3.5517, + "step": 26800 + }, + { + "epoch": 53.86, + "learning_rate": 8.936333333333334e-05, + "loss": 3.5871, + "step": 26820 + }, + { + "epoch": 53.9, + "learning_rate": 8.943e-05, + "loss": 3.5482, + "step": 26840 + }, + { + "epoch": 53.94, + "learning_rate": 8.949666666666667e-05, + "loss": 3.5609, + "step": 26860 + }, + { + "epoch": 53.98, + "learning_rate": 8.956333333333333e-05, + "loss": 3.5725, + "step": 26880 + }, + { + "epoch": 54.02, + "learning_rate": 8.963000000000001e-05, + "loss": 3.5513, + "step": 26900 + }, + { + "epoch": 54.06, + "learning_rate": 8.969666666666667e-05, + "loss": 3.4831, + "step": 26920 + }, + { + "epoch": 54.1, + "learning_rate": 8.976333333333334e-05, + "loss": 3.471, + "step": 26940 + }, + { + "epoch": 54.14, + "learning_rate": 8.983e-05, + "loss": 3.4957, + "step": 26960 + }, + { + "epoch": 54.18, + "learning_rate": 8.989666666666668e-05, + "loss": 3.4937, + "step": 26980 + }, + { + "epoch": 54.22, + "learning_rate": 8.996333333333333e-05, + "loss": 3.5183, + "step": 27000 + }, + { + "epoch": 54.22, + "eval_loss": 3.6193368434906006, + "eval_runtime": 32.525, + "eval_samples_per_second": 23.551, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.003092754429949435, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 9.065854366115463e-06, + "step": 27000 + }, + { + "epoch": 54.26, + "learning_rate": 9.003e-05, + "loss": 3.5114, + "step": 27020 + }, + { + "epoch": 54.3, + "learning_rate": 9.009666666666667e-05, + "loss": 3.5257, + "step": 27040 + }, + { + "epoch": 54.34, + "learning_rate": 9.016333333333333e-05, + "loss": 3.522, + "step": 27060 + }, + { + "epoch": 54.38, + "learning_rate": 9.023000000000001e-05, + "loss": 3.5359, + "step": 27080 + }, + { + "epoch": 54.42, + "learning_rate": 9.029666666666667e-05, + "loss": 3.5093, + "step": 27100 + }, + { + "epoch": 54.46, + "learning_rate": 9.036333333333334e-05, + "loss": 3.5311, + "step": 27120 + }, + { + "epoch": 54.5, + "learning_rate": 9.043e-05, + "loss": 3.5336, + "step": 27140 + }, + { + "epoch": 54.54, + "learning_rate": 9.049666666666668e-05, + "loss": 3.5278, + "step": 27160 + }, + { + "epoch": 54.58, + "learning_rate": 9.056333333333333e-05, + "loss": 3.5449, + "step": 27180 + }, + { + "epoch": 54.62, + "learning_rate": 9.063e-05, + "loss": 3.5038, + "step": 27200 + }, + { + "epoch": 54.66, + "learning_rate": 9.069666666666667e-05, + "loss": 3.5303, + "step": 27220 + }, + { + "epoch": 54.7, + "learning_rate": 9.076333333333333e-05, + "loss": 3.5039, + "step": 27240 + }, + { + "epoch": 54.74, + "learning_rate": 9.083000000000001e-05, + "loss": 3.5457, + "step": 27260 + }, + { + "epoch": 54.78, + "learning_rate": 9.089666666666667e-05, + "loss": 3.5362, + "step": 27280 + }, + { + "epoch": 54.82, + "learning_rate": 9.096333333333334e-05, + "loss": 3.5444, + "step": 27300 + }, + { + "epoch": 54.86, + "learning_rate": 9.103e-05, + "loss": 3.5395, + "step": 27320 + }, + { + "epoch": 54.9, + "learning_rate": 9.109666666666668e-05, + "loss": 3.5494, + "step": 27340 + }, + { + "epoch": 54.94, + "learning_rate": 9.116333333333333e-05, + "loss": 3.5328, + "step": 27360 + }, + { + "epoch": 54.98, + "learning_rate": 9.123e-05, + "loss": 3.5156, + "step": 27380 + }, + { + "epoch": 55.02, + "learning_rate": 9.129666666666667e-05, + "loss": 3.521, + "step": 27400 + }, + { + "epoch": 55.06, + "learning_rate": 9.136e-05, + "loss": 3.4545, + "step": 27420 + }, + { + "epoch": 55.1, + "learning_rate": 9.142666666666667e-05, + "loss": 3.4398, + "step": 27440 + }, + { + "epoch": 55.14, + "learning_rate": 9.149333333333335e-05, + "loss": 3.4454, + "step": 27460 + }, + { + "epoch": 55.18, + "learning_rate": 9.156e-05, + "loss": 3.466, + "step": 27480 + }, + { + "epoch": 55.22, + "learning_rate": 9.162666666666667e-05, + "loss": 3.4879, + "step": 27500 + }, + { + "epoch": 55.26, + "learning_rate": 9.169333333333334e-05, + "loss": 3.4678, + "step": 27520 + }, + { + "epoch": 55.3, + "learning_rate": 9.176e-05, + "loss": 3.4856, + "step": 27540 + }, + { + "epoch": 55.34, + "learning_rate": 9.182666666666667e-05, + "loss": 3.4659, + "step": 27560 + }, + { + "epoch": 55.38, + "learning_rate": 9.189333333333334e-05, + "loss": 3.5032, + "step": 27580 + }, + { + "epoch": 55.42, + "learning_rate": 9.196000000000001e-05, + "loss": 3.473, + "step": 27600 + }, + { + "epoch": 55.46, + "learning_rate": 9.202666666666667e-05, + "loss": 3.5081, + "step": 27620 + }, + { + "epoch": 55.5, + "learning_rate": 9.209333333333335e-05, + "loss": 3.5104, + "step": 27640 + }, + { + "epoch": 55.54, + "learning_rate": 9.216e-05, + "loss": 3.4946, + "step": 27660 + }, + { + "epoch": 55.58, + "learning_rate": 9.222666666666668e-05, + "loss": 3.4886, + "step": 27680 + }, + { + "epoch": 55.62, + "learning_rate": 9.229333333333334e-05, + "loss": 3.5019, + "step": 27700 + }, + { + "epoch": 55.66, + "learning_rate": 9.236e-05, + "loss": 3.519, + "step": 27720 + }, + { + "epoch": 55.7, + "learning_rate": 9.242666666666667e-05, + "loss": 3.5166, + "step": 27740 + }, + { + "epoch": 55.74, + "learning_rate": 9.249333333333334e-05, + "loss": 3.5029, + "step": 27760 + }, + { + "epoch": 55.78, + "learning_rate": 9.256000000000001e-05, + "loss": 3.5102, + "step": 27780 + }, + { + "epoch": 55.82, + "learning_rate": 9.262666666666667e-05, + "loss": 3.5131, + "step": 27800 + }, + { + "epoch": 55.86, + "learning_rate": 9.269333333333335e-05, + "loss": 3.4995, + "step": 27820 + }, + { + "epoch": 55.9, + "learning_rate": 9.276e-05, + "loss": 3.528, + "step": 27840 + }, + { + "epoch": 55.94, + "learning_rate": 9.282666666666668e-05, + "loss": 3.5206, + "step": 27860 + }, + { + "epoch": 55.98, + "learning_rate": 9.289333333333334e-05, + "loss": 3.5038, + "step": 27880 + }, + { + "epoch": 56.02, + "learning_rate": 9.296e-05, + "loss": 3.4693, + "step": 27900 + }, + { + "epoch": 56.06, + "learning_rate": 9.302666666666667e-05, + "loss": 3.4263, + "step": 27920 + }, + { + "epoch": 56.1, + "learning_rate": 9.309333333333333e-05, + "loss": 3.4514, + "step": 27940 + }, + { + "epoch": 56.14, + "learning_rate": 9.316000000000001e-05, + "loss": 3.4402, + "step": 27960 + }, + { + "epoch": 56.18, + "learning_rate": 9.322666666666667e-05, + "loss": 3.4529, + "step": 27980 + }, + { + "epoch": 56.22, + "learning_rate": 9.329333333333334e-05, + "loss": 3.4303, + "step": 28000 + }, + { + "epoch": 56.22, + "eval_loss": 3.5819454193115234, + "eval_runtime": 34.419, + "eval_samples_per_second": 22.255, + "eval_steps_per_second": 0.116, + "eval_tse_ndup": 0.0037477743156356714, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.4961434583568385e-06, + "step": 28000 + }, + { + "epoch": 56.27, + "learning_rate": 9.336e-05, + "loss": 3.441, + "step": 28020 + }, + { + "epoch": 56.31, + "learning_rate": 9.342666666666668e-05, + "loss": 3.4594, + "step": 28040 + }, + { + "epoch": 56.35, + "learning_rate": 9.349333333333333e-05, + "loss": 3.4511, + "step": 28060 + }, + { + "epoch": 56.39, + "learning_rate": 9.356e-05, + "loss": 3.4652, + "step": 28080 + }, + { + "epoch": 56.43, + "learning_rate": 9.362666666666667e-05, + "loss": 3.4743, + "step": 28100 + }, + { + "epoch": 56.47, + "learning_rate": 9.369333333333333e-05, + "loss": 3.4793, + "step": 28120 + }, + { + "epoch": 56.51, + "learning_rate": 9.376e-05, + "loss": 3.4523, + "step": 28140 + }, + { + "epoch": 56.55, + "learning_rate": 9.382666666666667e-05, + "loss": 3.4678, + "step": 28160 + }, + { + "epoch": 56.59, + "learning_rate": 9.389333333333334e-05, + "loss": 3.4712, + "step": 28180 + }, + { + "epoch": 56.63, + "learning_rate": 9.396e-05, + "loss": 3.4606, + "step": 28200 + }, + { + "epoch": 56.67, + "learning_rate": 9.402666666666668e-05, + "loss": 3.4742, + "step": 28220 + }, + { + "epoch": 56.71, + "learning_rate": 9.409333333333333e-05, + "loss": 3.487, + "step": 28240 + }, + { + "epoch": 56.75, + "learning_rate": 9.416e-05, + "loss": 3.5037, + "step": 28260 + }, + { + "epoch": 56.79, + "learning_rate": 9.422666666666667e-05, + "loss": 3.4766, + "step": 28280 + }, + { + "epoch": 56.83, + "learning_rate": 9.429333333333333e-05, + "loss": 3.478, + "step": 28300 + }, + { + "epoch": 56.87, + "learning_rate": 9.436e-05, + "loss": 3.4842, + "step": 28320 + }, + { + "epoch": 56.91, + "learning_rate": 9.442666666666668e-05, + "loss": 3.4781, + "step": 28340 + }, + { + "epoch": 56.95, + "learning_rate": 9.449333333333334e-05, + "loss": 3.4991, + "step": 28360 + }, + { + "epoch": 56.99, + "learning_rate": 9.456e-05, + "loss": 3.4682, + "step": 28380 + }, + { + "epoch": 57.03, + "learning_rate": 9.462666666666668e-05, + "loss": 3.4166, + "step": 28400 + }, + { + "epoch": 57.07, + "learning_rate": 9.469333333333333e-05, + "loss": 3.3915, + "step": 28420 + }, + { + "epoch": 57.11, + "learning_rate": 9.476000000000001e-05, + "loss": 3.4032, + "step": 28440 + }, + { + "epoch": 57.15, + "learning_rate": 9.482666666666667e-05, + "loss": 3.4245, + "step": 28460 + }, + { + "epoch": 57.19, + "learning_rate": 9.489333333333334e-05, + "loss": 3.4492, + "step": 28480 + }, + { + "epoch": 57.23, + "learning_rate": 9.496e-05, + "loss": 3.4192, + "step": 28500 + }, + { + "epoch": 57.27, + "learning_rate": 9.502666666666668e-05, + "loss": 3.4105, + "step": 28520 + }, + { + "epoch": 57.31, + "learning_rate": 9.509333333333334e-05, + "loss": 3.4265, + "step": 28540 + }, + { + "epoch": 57.35, + "learning_rate": 9.516e-05, + "loss": 3.4249, + "step": 28560 + }, + { + "epoch": 57.39, + "learning_rate": 9.522666666666668e-05, + "loss": 3.4205, + "step": 28580 + }, + { + "epoch": 57.43, + "learning_rate": 9.529333333333333e-05, + "loss": 3.4443, + "step": 28600 + }, + { + "epoch": 57.47, + "learning_rate": 9.536000000000001e-05, + "loss": 3.4374, + "step": 28620 + }, + { + "epoch": 57.51, + "learning_rate": 9.542666666666667e-05, + "loss": 3.4333, + "step": 28640 + }, + { + "epoch": 57.55, + "learning_rate": 9.549333333333334e-05, + "loss": 3.424, + "step": 28660 + }, + { + "epoch": 57.59, + "learning_rate": 9.556e-05, + "loss": 3.4512, + "step": 28680 + }, + { + "epoch": 57.63, + "learning_rate": 9.562666666666668e-05, + "loss": 3.4511, + "step": 28700 + }, + { + "epoch": 57.67, + "learning_rate": 9.569333333333334e-05, + "loss": 3.4468, + "step": 28720 + }, + { + "epoch": 57.71, + "learning_rate": 9.576e-05, + "loss": 3.4525, + "step": 28740 + }, + { + "epoch": 57.75, + "learning_rate": 9.582666666666668e-05, + "loss": 3.4426, + "step": 28760 + }, + { + "epoch": 57.79, + "learning_rate": 9.589333333333333e-05, + "loss": 3.4616, + "step": 28780 + }, + { + "epoch": 57.83, + "learning_rate": 9.596000000000001e-05, + "loss": 3.4642, + "step": 28800 + }, + { + "epoch": 57.87, + "learning_rate": 9.602666666666667e-05, + "loss": 3.4632, + "step": 28820 + }, + { + "epoch": 57.91, + "learning_rate": 9.609333333333334e-05, + "loss": 3.4555, + "step": 28840 + }, + { + "epoch": 57.95, + "learning_rate": 9.616e-05, + "loss": 3.449, + "step": 28860 + }, + { + "epoch": 57.99, + "learning_rate": 9.622666666666668e-05, + "loss": 3.4371, + "step": 28880 + }, + { + "epoch": 58.03, + "learning_rate": 9.629333333333334e-05, + "loss": 3.3892, + "step": 28900 + }, + { + "epoch": 58.07, + "learning_rate": 9.636e-05, + "loss": 3.3735, + "step": 28920 + }, + { + "epoch": 58.11, + "learning_rate": 9.642666666666667e-05, + "loss": 3.3795, + "step": 28940 + }, + { + "epoch": 58.15, + "learning_rate": 9.649333333333333e-05, + "loss": 3.3724, + "step": 28960 + }, + { + "epoch": 58.19, + "learning_rate": 9.656000000000001e-05, + "loss": 3.367, + "step": 28980 + }, + { + "epoch": 58.23, + "learning_rate": 9.662666666666667e-05, + "loss": 3.4298, + "step": 29000 + }, + { + "epoch": 58.23, + "eval_loss": 3.543978214263916, + "eval_runtime": 34.265, + "eval_samples_per_second": 22.355, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.0030940438107753946, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.858930098357058e-06, + "step": 29000 + }, + { + "epoch": 58.27, + "learning_rate": 9.669333333333334e-05, + "loss": 3.3914, + "step": 29020 + }, + { + "epoch": 58.31, + "learning_rate": 9.676e-05, + "loss": 3.4077, + "step": 29040 + }, + { + "epoch": 58.35, + "learning_rate": 9.682666666666668e-05, + "loss": 3.3884, + "step": 29060 + }, + { + "epoch": 58.39, + "learning_rate": 9.689333333333333e-05, + "loss": 3.3959, + "step": 29080 + }, + { + "epoch": 58.43, + "learning_rate": 9.696000000000001e-05, + "loss": 3.4057, + "step": 29100 + }, + { + "epoch": 58.47, + "learning_rate": 9.702666666666667e-05, + "loss": 3.4132, + "step": 29120 + }, + { + "epoch": 58.51, + "learning_rate": 9.709333333333333e-05, + "loss": 3.4031, + "step": 29140 + }, + { + "epoch": 58.55, + "learning_rate": 9.716000000000001e-05, + "loss": 3.4311, + "step": 29160 + }, + { + "epoch": 58.59, + "learning_rate": 9.722666666666666e-05, + "loss": 3.4227, + "step": 29180 + }, + { + "epoch": 58.63, + "learning_rate": 9.729333333333334e-05, + "loss": 3.4277, + "step": 29200 + }, + { + "epoch": 58.67, + "learning_rate": 9.736e-05, + "loss": 3.4303, + "step": 29220 + }, + { + "epoch": 58.71, + "learning_rate": 9.742666666666667e-05, + "loss": 3.4255, + "step": 29240 + }, + { + "epoch": 58.76, + "learning_rate": 9.749333333333333e-05, + "loss": 3.4348, + "step": 29260 + }, + { + "epoch": 58.8, + "learning_rate": 9.756000000000001e-05, + "loss": 3.4223, + "step": 29280 + }, + { + "epoch": 58.84, + "learning_rate": 9.762666666666667e-05, + "loss": 3.4252, + "step": 29300 + }, + { + "epoch": 58.88, + "learning_rate": 9.769333333333334e-05, + "loss": 3.4369, + "step": 29320 + }, + { + "epoch": 58.92, + "learning_rate": 9.776000000000001e-05, + "loss": 3.425, + "step": 29340 + }, + { + "epoch": 58.96, + "learning_rate": 9.782666666666666e-05, + "loss": 3.4159, + "step": 29360 + }, + { + "epoch": 59.0, + "learning_rate": 9.789333333333334e-05, + "loss": 3.4257, + "step": 29380 + }, + { + "epoch": 59.04, + "learning_rate": 9.796e-05, + "loss": 3.3579, + "step": 29400 + }, + { + "epoch": 59.08, + "learning_rate": 9.802666666666667e-05, + "loss": 3.3561, + "step": 29420 + }, + { + "epoch": 59.12, + "learning_rate": 9.809333333333333e-05, + "loss": 3.3375, + "step": 29440 + }, + { + "epoch": 59.16, + "learning_rate": 9.816000000000001e-05, + "loss": 3.3639, + "step": 29460 + }, + { + "epoch": 59.2, + "learning_rate": 9.822666666666667e-05, + "loss": 3.3674, + "step": 29480 + }, + { + "epoch": 59.24, + "learning_rate": 9.829333333333334e-05, + "loss": 3.3444, + "step": 29500 + }, + { + "epoch": 59.28, + "learning_rate": 9.836000000000001e-05, + "loss": 3.3795, + "step": 29520 + }, + { + "epoch": 59.32, + "learning_rate": 9.842666666666666e-05, + "loss": 3.3766, + "step": 29540 + }, + { + "epoch": 59.36, + "learning_rate": 9.849333333333334e-05, + "loss": 3.3729, + "step": 29560 + }, + { + "epoch": 59.4, + "learning_rate": 9.856e-05, + "loss": 3.385, + "step": 29580 + }, + { + "epoch": 59.44, + "learning_rate": 9.862666666666667e-05, + "loss": 3.3762, + "step": 29600 + }, + { + "epoch": 59.48, + "learning_rate": 9.869333333333333e-05, + "loss": 3.3695, + "step": 29620 + }, + { + "epoch": 59.52, + "learning_rate": 9.876000000000001e-05, + "loss": 3.3782, + "step": 29640 + }, + { + "epoch": 59.56, + "learning_rate": 9.882666666666667e-05, + "loss": 3.3869, + "step": 29660 + }, + { + "epoch": 59.6, + "learning_rate": 9.889333333333334e-05, + "loss": 3.3827, + "step": 29680 + }, + { + "epoch": 59.64, + "learning_rate": 9.896000000000001e-05, + "loss": 3.3884, + "step": 29700 + }, + { + "epoch": 59.68, + "learning_rate": 9.902666666666666e-05, + "loss": 3.4089, + "step": 29720 + }, + { + "epoch": 59.72, + "learning_rate": 9.909333333333334e-05, + "loss": 3.3945, + "step": 29740 + }, + { + "epoch": 59.76, + "learning_rate": 9.916e-05, + "loss": 3.4083, + "step": 29760 + }, + { + "epoch": 59.8, + "learning_rate": 9.922666666666667e-05, + "loss": 3.412, + "step": 29780 + }, + { + "epoch": 59.84, + "learning_rate": 9.929333333333333e-05, + "loss": 3.3954, + "step": 29800 + }, + { + "epoch": 59.88, + "learning_rate": 9.936000000000001e-05, + "loss": 3.3937, + "step": 29820 + }, + { + "epoch": 59.92, + "learning_rate": 9.942666666666667e-05, + "loss": 3.4168, + "step": 29840 + }, + { + "epoch": 59.96, + "learning_rate": 9.949333333333334e-05, + "loss": 3.3937, + "step": 29860 + }, + { + "epoch": 60.0, + "learning_rate": 9.956e-05, + "loss": 3.4015, + "step": 29880 + }, + { + "epoch": 60.04, + "learning_rate": 9.962666666666667e-05, + "loss": 3.3123, + "step": 29900 + }, + { + "epoch": 60.08, + "learning_rate": 9.969333333333334e-05, + "loss": 3.3176, + "step": 29920 + }, + { + "epoch": 60.12, + "learning_rate": 9.976000000000001e-05, + "loss": 3.3188, + "step": 29940 + }, + { + "epoch": 60.16, + "learning_rate": 9.982666666666667e-05, + "loss": 3.3403, + "step": 29960 + }, + { + "epoch": 60.2, + "learning_rate": 9.989333333333333e-05, + "loss": 3.3459, + "step": 29980 + }, + { + "epoch": 60.24, + "learning_rate": 9.996000000000001e-05, + "loss": 3.3547, + "step": 30000 + }, + { + "epoch": 60.24, + "eval_loss": 3.5147206783294678, + "eval_runtime": 34.2489, + "eval_samples_per_second": 22.366, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.004077897177159217, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.799390774586597e-06, + "step": 30000 + }, + { + "epoch": 60.28, + "learning_rate": 9.999999677727208e-05, + "loss": 3.326, + "step": 30020 + }, + { + "epoch": 60.32, + "learning_rate": 9.99999605215876e-05, + "loss": 3.3468, + "step": 30040 + }, + { + "epoch": 60.36, + "learning_rate": 9.999988398183804e-05, + "loss": 3.3376, + "step": 30060 + }, + { + "epoch": 60.4, + "learning_rate": 9.999976715808505e-05, + "loss": 3.3675, + "step": 30080 + }, + { + "epoch": 60.44, + "learning_rate": 9.999961005042278e-05, + "loss": 3.358, + "step": 30100 + }, + { + "epoch": 60.48, + "learning_rate": 9.999941265897779e-05, + "loss": 3.3522, + "step": 30120 + }, + { + "epoch": 60.52, + "learning_rate": 9.999917498390912e-05, + "loss": 3.3772, + "step": 30140 + }, + { + "epoch": 60.56, + "learning_rate": 9.999889702540826e-05, + "loss": 3.3682, + "step": 30160 + }, + { + "epoch": 60.6, + "learning_rate": 9.999857878369916e-05, + "loss": 3.3663, + "step": 30180 + }, + { + "epoch": 60.64, + "learning_rate": 9.99982202590382e-05, + "loss": 3.3637, + "step": 30200 + }, + { + "epoch": 60.68, + "learning_rate": 9.999782145171425e-05, + "loss": 3.368, + "step": 30220 + }, + { + "epoch": 60.72, + "learning_rate": 9.999738236204865e-05, + "loss": 3.381, + "step": 30240 + }, + { + "epoch": 60.76, + "learning_rate": 9.999690299039513e-05, + "loss": 3.3683, + "step": 30260 + }, + { + "epoch": 60.8, + "learning_rate": 9.999638333713993e-05, + "loss": 3.3531, + "step": 30280 + }, + { + "epoch": 60.84, + "learning_rate": 9.99958234027017e-05, + "loss": 3.3772, + "step": 30300 + }, + { + "epoch": 60.88, + "learning_rate": 9.99952231875316e-05, + "loss": 3.3757, + "step": 30320 + }, + { + "epoch": 60.92, + "learning_rate": 9.999458269211321e-05, + "loss": 3.376, + "step": 30340 + }, + { + "epoch": 60.96, + "learning_rate": 9.999390191696255e-05, + "loss": 3.3823, + "step": 30360 + }, + { + "epoch": 61.0, + "learning_rate": 9.99931808626281e-05, + "loss": 3.3515, + "step": 30380 + }, + { + "epoch": 61.04, + "learning_rate": 9.999241952969083e-05, + "loss": 3.2709, + "step": 30400 + }, + { + "epoch": 61.08, + "learning_rate": 9.99916179187641e-05, + "loss": 3.2939, + "step": 30420 + }, + { + "epoch": 61.12, + "learning_rate": 9.999077603049378e-05, + "loss": 3.3137, + "step": 30440 + }, + { + "epoch": 61.16, + "learning_rate": 9.998989386555814e-05, + "loss": 3.3036, + "step": 30460 + }, + { + "epoch": 61.2, + "learning_rate": 9.998897142466797e-05, + "loss": 3.3012, + "step": 30480 + }, + { + "epoch": 61.24, + "learning_rate": 9.99880087085664e-05, + "loss": 3.3207, + "step": 30500 + }, + { + "epoch": 61.29, + "learning_rate": 9.998700571802912e-05, + "loss": 3.3248, + "step": 30520 + }, + { + "epoch": 61.33, + "learning_rate": 9.998596245386422e-05, + "loss": 3.3072, + "step": 30540 + }, + { + "epoch": 61.37, + "learning_rate": 9.998487891691219e-05, + "loss": 3.3247, + "step": 30560 + }, + { + "epoch": 61.41, + "learning_rate": 9.998375510804609e-05, + "loss": 3.336, + "step": 30580 + }, + { + "epoch": 61.45, + "learning_rate": 9.998259102817129e-05, + "loss": 3.3163, + "step": 30600 + }, + { + "epoch": 61.49, + "learning_rate": 9.99813866782257e-05, + "loss": 3.3386, + "step": 30620 + }, + { + "epoch": 61.53, + "learning_rate": 9.998014205917963e-05, + "loss": 3.3454, + "step": 30640 + }, + { + "epoch": 61.57, + "learning_rate": 9.997885717203586e-05, + "loss": 3.3343, + "step": 30660 + }, + { + "epoch": 61.61, + "learning_rate": 9.997753201782961e-05, + "loss": 3.3289, + "step": 30680 + }, + { + "epoch": 61.65, + "learning_rate": 9.997616659762848e-05, + "loss": 3.3368, + "step": 30700 + }, + { + "epoch": 61.69, + "learning_rate": 9.997476091253261e-05, + "loss": 3.3442, + "step": 30720 + }, + { + "epoch": 61.73, + "learning_rate": 9.997331496367455e-05, + "loss": 3.3307, + "step": 30740 + }, + { + "epoch": 61.77, + "learning_rate": 9.997182875221922e-05, + "loss": 3.3358, + "step": 30760 + }, + { + "epoch": 61.81, + "learning_rate": 9.997030227936408e-05, + "loss": 3.3184, + "step": 30780 + }, + { + "epoch": 61.85, + "learning_rate": 9.996873554633897e-05, + "loss": 3.3497, + "step": 30800 + }, + { + "epoch": 61.89, + "learning_rate": 9.996712855440616e-05, + "loss": 3.3705, + "step": 30820 + }, + { + "epoch": 61.93, + "learning_rate": 9.99654813048604e-05, + "loss": 3.3516, + "step": 30840 + }, + { + "epoch": 61.97, + "learning_rate": 9.996379379902883e-05, + "loss": 3.3588, + "step": 30860 + }, + { + "epoch": 62.01, + "learning_rate": 9.996206603827105e-05, + "loss": 3.3081, + "step": 30880 + }, + { + "epoch": 62.05, + "learning_rate": 9.99602980239791e-05, + "loss": 3.2414, + "step": 30900 + }, + { + "epoch": 62.09, + "learning_rate": 9.99584897575774e-05, + "loss": 3.2668, + "step": 30920 + }, + { + "epoch": 62.13, + "learning_rate": 9.995664124052288e-05, + "loss": 3.2673, + "step": 30940 + }, + { + "epoch": 62.17, + "learning_rate": 9.995475247430483e-05, + "loss": 3.2741, + "step": 30960 + }, + { + "epoch": 62.21, + "learning_rate": 9.995282346044501e-05, + "loss": 3.2683, + "step": 30980 + }, + { + "epoch": 62.25, + "learning_rate": 9.995085420049757e-05, + "loss": 3.2909, + "step": 31000 + }, + { + "epoch": 62.25, + "eval_loss": 3.472719430923462, + "eval_runtime": 34.199, + "eval_samples_per_second": 22.398, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.004517904171379088, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 5.439512619669278e-05, + "step": 31000 + }, + { + "epoch": 62.29, + "learning_rate": 9.994884469604912e-05, + "loss": 3.269, + "step": 31020 + }, + { + "epoch": 62.33, + "learning_rate": 9.99467949487187e-05, + "loss": 3.2857, + "step": 31040 + }, + { + "epoch": 62.37, + "learning_rate": 9.994470496015773e-05, + "loss": 3.2992, + "step": 31060 + }, + { + "epoch": 62.41, + "learning_rate": 9.994257473205009e-05, + "loss": 3.2951, + "step": 31080 + }, + { + "epoch": 62.45, + "learning_rate": 9.994040426611204e-05, + "loss": 3.2939, + "step": 31100 + }, + { + "epoch": 62.49, + "learning_rate": 9.993819356409233e-05, + "loss": 3.3046, + "step": 31120 + }, + { + "epoch": 62.53, + "learning_rate": 9.993594262777203e-05, + "loss": 3.2982, + "step": 31140 + }, + { + "epoch": 62.57, + "learning_rate": 9.993365145896473e-05, + "loss": 3.2915, + "step": 31160 + }, + { + "epoch": 62.61, + "learning_rate": 9.993132005951635e-05, + "loss": 3.3135, + "step": 31180 + }, + { + "epoch": 62.65, + "learning_rate": 9.992894843130526e-05, + "loss": 3.3076, + "step": 31200 + }, + { + "epoch": 62.69, + "learning_rate": 9.992653657624225e-05, + "loss": 3.3025, + "step": 31220 + }, + { + "epoch": 62.73, + "learning_rate": 9.992408449627051e-05, + "loss": 3.3287, + "step": 31240 + }, + { + "epoch": 62.77, + "learning_rate": 9.992159219336561e-05, + "loss": 3.3267, + "step": 31260 + }, + { + "epoch": 62.81, + "learning_rate": 9.991905966953559e-05, + "loss": 3.3288, + "step": 31280 + }, + { + "epoch": 62.85, + "learning_rate": 9.991648692682083e-05, + "loss": 3.3305, + "step": 31300 + }, + { + "epoch": 62.89, + "learning_rate": 9.991387396729415e-05, + "loss": 3.3249, + "step": 31320 + }, + { + "epoch": 62.93, + "learning_rate": 9.99112207930608e-05, + "loss": 3.328, + "step": 31340 + }, + { + "epoch": 62.97, + "learning_rate": 9.990852740625831e-05, + "loss": 3.3317, + "step": 31360 + }, + { + "epoch": 63.01, + "learning_rate": 9.990579380905678e-05, + "loss": 3.3109, + "step": 31380 + }, + { + "epoch": 63.05, + "learning_rate": 9.990302000365858e-05, + "loss": 3.2172, + "step": 31400 + }, + { + "epoch": 63.09, + "learning_rate": 9.990034764772323e-05, + "loss": 3.2413, + "step": 31420 + }, + { + "epoch": 63.13, + "learning_rate": 9.98974954427989e-05, + "loss": 3.2467, + "step": 31440 + }, + { + "epoch": 63.17, + "learning_rate": 9.989460303636373e-05, + "loss": 3.239, + "step": 31460 + }, + { + "epoch": 63.21, + "learning_rate": 9.989167043074812e-05, + "loss": 3.2643, + "step": 31480 + }, + { + "epoch": 63.25, + "learning_rate": 9.988869762831478e-05, + "loss": 3.2649, + "step": 31500 + }, + { + "epoch": 63.29, + "learning_rate": 9.988568463145886e-05, + "loss": 3.2774, + "step": 31520 + }, + { + "epoch": 63.33, + "learning_rate": 9.988263144260788e-05, + "loss": 3.2511, + "step": 31540 + }, + { + "epoch": 63.37, + "learning_rate": 9.987953806422173e-05, + "loss": 3.2782, + "step": 31560 + }, + { + "epoch": 63.41, + "learning_rate": 9.98764044987927e-05, + "loss": 3.2736, + "step": 31580 + }, + { + "epoch": 63.45, + "learning_rate": 9.987323074884543e-05, + "loss": 3.2659, + "step": 31600 + }, + { + "epoch": 63.49, + "learning_rate": 9.987001681693697e-05, + "loss": 3.2844, + "step": 31620 + }, + { + "epoch": 63.53, + "learning_rate": 9.986676270565674e-05, + "loss": 3.2815, + "step": 31640 + }, + { + "epoch": 63.57, + "learning_rate": 9.986346841762648e-05, + "loss": 3.2817, + "step": 31660 + }, + { + "epoch": 63.61, + "learning_rate": 9.986013395550036e-05, + "loss": 3.274, + "step": 31680 + }, + { + "epoch": 63.65, + "learning_rate": 9.98567593219649e-05, + "loss": 3.2815, + "step": 31700 + }, + { + "epoch": 63.69, + "learning_rate": 9.985334451973896e-05, + "loss": 3.2765, + "step": 31720 + }, + { + "epoch": 63.73, + "learning_rate": 9.984988955157381e-05, + "loss": 3.2823, + "step": 31740 + }, + { + "epoch": 63.78, + "learning_rate": 9.984639442025306e-05, + "loss": 3.2991, + "step": 31760 + }, + { + "epoch": 63.82, + "learning_rate": 9.984285912859263e-05, + "loss": 3.2877, + "step": 31780 + }, + { + "epoch": 63.86, + "learning_rate": 9.98392836794409e-05, + "loss": 3.3074, + "step": 31800 + }, + { + "epoch": 63.9, + "learning_rate": 9.98356680756785e-05, + "loss": 3.2998, + "step": 31820 + }, + { + "epoch": 63.94, + "learning_rate": 9.983201232021847e-05, + "loss": 3.2996, + "step": 31840 + }, + { + "epoch": 63.98, + "learning_rate": 9.982831641600618e-05, + "loss": 3.3012, + "step": 31860 + }, + { + "epoch": 64.02, + "learning_rate": 9.982458036601939e-05, + "loss": 3.2715, + "step": 31880 + }, + { + "epoch": 64.06, + "learning_rate": 9.982080417326811e-05, + "loss": 3.1864, + "step": 31900 + }, + { + "epoch": 64.1, + "learning_rate": 9.98169878407948e-05, + "loss": 3.2261, + "step": 31920 + }, + { + "epoch": 64.14, + "learning_rate": 9.981313137167419e-05, + "loss": 3.2119, + "step": 31940 + }, + { + "epoch": 64.18, + "learning_rate": 9.980923476901333e-05, + "loss": 3.2365, + "step": 31960 + }, + { + "epoch": 64.22, + "learning_rate": 9.980529803595172e-05, + "loss": 3.2367, + "step": 31980 + }, + { + "epoch": 64.26, + "learning_rate": 9.980132117566106e-05, + "loss": 3.24, + "step": 32000 + }, + { + "epoch": 64.26, + "eval_loss": 3.442429780960083, + "eval_runtime": 34.15, + "eval_samples_per_second": 22.43, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.003033775938362608, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.3598781549173195e-05, + "step": 32000 + }, + { + "epoch": 64.3, + "learning_rate": 9.979730419134546e-05, + "loss": 3.2347, + "step": 32020 + }, + { + "epoch": 64.34, + "learning_rate": 9.979324708624129e-05, + "loss": 3.2324, + "step": 32040 + }, + { + "epoch": 64.38, + "learning_rate": 9.978914986361734e-05, + "loss": 3.2395, + "step": 32060 + }, + { + "epoch": 64.42, + "learning_rate": 9.978501252677464e-05, + "loss": 3.2305, + "step": 32080 + }, + { + "epoch": 64.46, + "learning_rate": 9.978083507904659e-05, + "loss": 3.2592, + "step": 32100 + }, + { + "epoch": 64.5, + "learning_rate": 9.977661752379883e-05, + "loss": 3.2598, + "step": 32120 + }, + { + "epoch": 64.54, + "learning_rate": 9.977235986442943e-05, + "loss": 3.2655, + "step": 32140 + }, + { + "epoch": 64.58, + "learning_rate": 9.976806210436868e-05, + "loss": 3.2568, + "step": 32160 + }, + { + "epoch": 64.62, + "learning_rate": 9.97637242470792e-05, + "loss": 3.2583, + "step": 32180 + }, + { + "epoch": 64.66, + "learning_rate": 9.975934629605595e-05, + "loss": 3.2559, + "step": 32200 + }, + { + "epoch": 64.7, + "learning_rate": 9.975492825482615e-05, + "loss": 3.2632, + "step": 32220 + }, + { + "epoch": 64.74, + "learning_rate": 9.975047012694933e-05, + "loss": 3.2619, + "step": 32240 + }, + { + "epoch": 64.78, + "learning_rate": 9.974597191601733e-05, + "loss": 3.257, + "step": 32260 + }, + { + "epoch": 64.82, + "learning_rate": 9.97414336256543e-05, + "loss": 3.2741, + "step": 32280 + }, + { + "epoch": 64.86, + "learning_rate": 9.97368552595166e-05, + "loss": 3.2825, + "step": 32300 + }, + { + "epoch": 64.9, + "learning_rate": 9.9732236821293e-05, + "loss": 3.2737, + "step": 32320 + }, + { + "epoch": 64.94, + "learning_rate": 9.972757831470445e-05, + "loss": 3.2662, + "step": 32340 + }, + { + "epoch": 64.98, + "learning_rate": 9.972287974350425e-05, + "loss": 3.2667, + "step": 32360 + }, + { + "epoch": 65.02, + "learning_rate": 9.971814111147793e-05, + "loss": 3.2121, + "step": 32380 + }, + { + "epoch": 65.06, + "learning_rate": 9.971336242244333e-05, + "loss": 3.2034, + "step": 32400 + }, + { + "epoch": 65.1, + "learning_rate": 9.970854368025057e-05, + "loss": 3.1867, + "step": 32420 + }, + { + "epoch": 65.14, + "learning_rate": 9.970368488878202e-05, + "loss": 3.169, + "step": 32440 + }, + { + "epoch": 65.18, + "learning_rate": 9.969878605195231e-05, + "loss": 3.2003, + "step": 32460 + }, + { + "epoch": 65.22, + "learning_rate": 9.969384717370834e-05, + "loss": 3.1943, + "step": 32480 + }, + { + "epoch": 65.26, + "learning_rate": 9.968886825802926e-05, + "loss": 3.1981, + "step": 32500 + }, + { + "epoch": 65.3, + "learning_rate": 9.968384930892653e-05, + "loss": 3.2159, + "step": 32520 + }, + { + "epoch": 65.34, + "learning_rate": 9.967879033044382e-05, + "loss": 3.2019, + "step": 32540 + }, + { + "epoch": 65.38, + "learning_rate": 9.967369132665703e-05, + "loss": 3.2311, + "step": 32560 + }, + { + "epoch": 65.42, + "learning_rate": 9.966855230167436e-05, + "loss": 3.2331, + "step": 32580 + }, + { + "epoch": 65.46, + "learning_rate": 9.966337325963624e-05, + "loss": 3.2223, + "step": 32600 + }, + { + "epoch": 65.5, + "learning_rate": 9.96581542047153e-05, + "loss": 3.2066, + "step": 32620 + }, + { + "epoch": 65.54, + "learning_rate": 9.965289514111644e-05, + "loss": 3.2304, + "step": 32640 + }, + { + "epoch": 65.58, + "learning_rate": 9.96475960730768e-05, + "loss": 3.2308, + "step": 32660 + }, + { + "epoch": 65.62, + "learning_rate": 9.964225700486576e-05, + "loss": 3.2466, + "step": 32680 + }, + { + "epoch": 65.66, + "learning_rate": 9.96368779407849e-05, + "loss": 3.2336, + "step": 32700 + }, + { + "epoch": 65.7, + "learning_rate": 9.963145888516803e-05, + "loss": 3.2461, + "step": 32720 + }, + { + "epoch": 65.74, + "learning_rate": 9.96259998423812e-05, + "loss": 3.2524, + "step": 32740 + }, + { + "epoch": 65.78, + "learning_rate": 9.962050081682261e-05, + "loss": 3.2419, + "step": 32760 + }, + { + "epoch": 65.82, + "learning_rate": 9.96149618129228e-05, + "loss": 3.2402, + "step": 32780 + }, + { + "epoch": 65.86, + "learning_rate": 9.96093828351444e-05, + "loss": 3.2513, + "step": 32800 + }, + { + "epoch": 65.9, + "learning_rate": 9.960376388798232e-05, + "loss": 3.2527, + "step": 32820 + }, + { + "epoch": 65.94, + "learning_rate": 9.959810497596361e-05, + "loss": 3.2652, + "step": 32840 + }, + { + "epoch": 65.98, + "learning_rate": 9.959240610364757e-05, + "loss": 3.2228, + "step": 32860 + }, + { + "epoch": 66.02, + "learning_rate": 9.958666727562568e-05, + "loss": 3.1984, + "step": 32880 + }, + { + "epoch": 66.06, + "learning_rate": 9.95808884965216e-05, + "loss": 3.1355, + "step": 32900 + }, + { + "epoch": 66.1, + "learning_rate": 9.957506977099123e-05, + "loss": 3.1519, + "step": 32920 + }, + { + "epoch": 66.14, + "learning_rate": 9.956921110372255e-05, + "loss": 3.1557, + "step": 32940 + }, + { + "epoch": 66.18, + "learning_rate": 9.956331249943583e-05, + "loss": 3.2067, + "step": 32960 + }, + { + "epoch": 66.22, + "learning_rate": 9.955737396288343e-05, + "loss": 3.1896, + "step": 32980 + }, + { + "epoch": 66.27, + "learning_rate": 9.955139549884995e-05, + "loss": 3.1833, + "step": 33000 + }, + { + "epoch": 66.27, + "eval_loss": 3.4064080715179443, + "eval_runtime": 34.592, + "eval_samples_per_second": 22.144, + "eval_steps_per_second": 0.116, + "eval_tse_ndup": 0.0036108256657534973, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.5865245140702057e-05, + "step": 33000 + }, + { + "epoch": 66.31, + "learning_rate": 9.954537711215212e-05, + "loss": 3.2035, + "step": 33020 + }, + { + "epoch": 66.35, + "learning_rate": 9.953931880763885e-05, + "loss": 3.194, + "step": 33040 + }, + { + "epoch": 66.39, + "learning_rate": 9.953322059019118e-05, + "loss": 3.1952, + "step": 33060 + }, + { + "epoch": 66.43, + "learning_rate": 9.95270824647224e-05, + "loss": 3.2167, + "step": 33080 + }, + { + "epoch": 66.47, + "learning_rate": 9.952090443617783e-05, + "loss": 3.1999, + "step": 33100 + }, + { + "epoch": 66.51, + "learning_rate": 9.951468650953499e-05, + "loss": 3.2021, + "step": 33120 + }, + { + "epoch": 66.55, + "learning_rate": 9.95084286898036e-05, + "loss": 3.2029, + "step": 33140 + }, + { + "epoch": 66.59, + "learning_rate": 9.950213098202543e-05, + "loss": 3.2069, + "step": 33160 + }, + { + "epoch": 66.63, + "learning_rate": 9.949579339127444e-05, + "loss": 3.1867, + "step": 33180 + }, + { + "epoch": 66.67, + "learning_rate": 9.948941592265673e-05, + "loss": 3.2052, + "step": 33200 + }, + { + "epoch": 66.71, + "learning_rate": 9.948299858131049e-05, + "loss": 3.2045, + "step": 33220 + }, + { + "epoch": 66.75, + "learning_rate": 9.947654137240606e-05, + "loss": 3.2151, + "step": 33240 + }, + { + "epoch": 66.79, + "learning_rate": 9.94700443011459e-05, + "loss": 3.2122, + "step": 33260 + }, + { + "epoch": 66.83, + "learning_rate": 9.946350737276459e-05, + "loss": 3.2219, + "step": 33280 + }, + { + "epoch": 66.87, + "learning_rate": 9.945693059252881e-05, + "loss": 3.242, + "step": 33300 + }, + { + "epoch": 66.91, + "learning_rate": 9.945031396573734e-05, + "loss": 3.2395, + "step": 33320 + }, + { + "epoch": 66.95, + "learning_rate": 9.94436574977211e-05, + "loss": 3.2189, + "step": 33340 + }, + { + "epoch": 66.99, + "learning_rate": 9.943696119384305e-05, + "loss": 3.2346, + "step": 33360 + }, + { + "epoch": 67.03, + "learning_rate": 9.943022505949833e-05, + "loss": 3.1595, + "step": 33380 + }, + { + "epoch": 67.07, + "learning_rate": 9.942344910011407e-05, + "loss": 3.1317, + "step": 33400 + }, + { + "epoch": 67.11, + "learning_rate": 9.941663332114959e-05, + "loss": 3.1256, + "step": 33420 + }, + { + "epoch": 67.15, + "learning_rate": 9.941012145324835e-05, + "loss": 3.1375, + "step": 33440 + }, + { + "epoch": 67.19, + "learning_rate": 9.94032280419261e-05, + "loss": 3.1477, + "step": 33460 + }, + { + "epoch": 67.23, + "learning_rate": 9.939629482731533e-05, + "loss": 3.1549, + "step": 33480 + }, + { + "epoch": 67.27, + "learning_rate": 9.938932181500205e-05, + "loss": 3.1402, + "step": 33500 + }, + { + "epoch": 67.31, + "learning_rate": 9.938230901060423e-05, + "loss": 3.1778, + "step": 33520 + }, + { + "epoch": 67.35, + "learning_rate": 9.937525641977202e-05, + "loss": 3.1773, + "step": 33540 + }, + { + "epoch": 67.39, + "learning_rate": 9.936816404818751e-05, + "loss": 3.1828, + "step": 33560 + }, + { + "epoch": 67.43, + "learning_rate": 9.936103190156494e-05, + "loss": 3.1967, + "step": 33580 + }, + { + "epoch": 67.47, + "learning_rate": 9.935385998565053e-05, + "loss": 3.1896, + "step": 33600 + }, + { + "epoch": 67.51, + "learning_rate": 9.934664830622258e-05, + "loss": 3.174, + "step": 33620 + }, + { + "epoch": 67.55, + "learning_rate": 9.933939686909137e-05, + "loss": 3.1905, + "step": 33640 + }, + { + "epoch": 67.59, + "learning_rate": 9.933210568009929e-05, + "loss": 3.2043, + "step": 33660 + }, + { + "epoch": 67.63, + "learning_rate": 9.932477474512072e-05, + "loss": 3.1838, + "step": 33680 + }, + { + "epoch": 67.67, + "learning_rate": 9.931740407006205e-05, + "loss": 3.1873, + "step": 33700 + }, + { + "epoch": 67.71, + "learning_rate": 9.930999366086169e-05, + "loss": 3.2052, + "step": 33720 + }, + { + "epoch": 67.75, + "learning_rate": 9.930254352349008e-05, + "loss": 3.1873, + "step": 33740 + }, + { + "epoch": 67.79, + "learning_rate": 9.929505366394967e-05, + "loss": 3.1994, + "step": 33760 + }, + { + "epoch": 67.83, + "learning_rate": 9.92875240882749e-05, + "loss": 3.1895, + "step": 33780 + }, + { + "epoch": 67.87, + "learning_rate": 9.92799548025322e-05, + "loss": 3.1835, + "step": 33800 + }, + { + "epoch": 67.91, + "learning_rate": 9.927234581282004e-05, + "loss": 3.2167, + "step": 33820 + }, + { + "epoch": 67.95, + "learning_rate": 9.92646971252688e-05, + "loss": 3.1902, + "step": 33840 + }, + { + "epoch": 67.99, + "learning_rate": 9.925700874604093e-05, + "loss": 3.1904, + "step": 33860 + }, + { + "epoch": 68.03, + "learning_rate": 9.92492806813308e-05, + "loss": 3.1183, + "step": 33880 + }, + { + "epoch": 68.07, + "learning_rate": 9.924151293736476e-05, + "loss": 3.1173, + "step": 33900 + }, + { + "epoch": 68.11, + "learning_rate": 9.923370552040116e-05, + "loss": 3.1226, + "step": 33920 + }, + { + "epoch": 68.15, + "learning_rate": 9.922585843673031e-05, + "loss": 3.1213, + "step": 33940 + }, + { + "epoch": 68.19, + "learning_rate": 9.921797169267443e-05, + "loss": 3.1254, + "step": 33960 + }, + { + "epoch": 68.23, + "learning_rate": 9.921004529458772e-05, + "loss": 3.1092, + "step": 33980 + }, + { + "epoch": 68.27, + "learning_rate": 9.920207924885638e-05, + "loss": 3.15, + "step": 34000 + }, + { + "epoch": 68.27, + "eval_loss": 3.377257823944092, + "eval_runtime": 32.456, + "eval_samples_per_second": 23.601, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0033100573882608566, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.5865245140702057e-05, + "step": 34000 + }, + { + "epoch": 68.31, + "learning_rate": 9.919407356189849e-05, + "loss": 3.1504, + "step": 34020 + }, + { + "epoch": 68.35, + "learning_rate": 9.918602824016408e-05, + "loss": 3.1453, + "step": 34040 + }, + { + "epoch": 68.39, + "learning_rate": 9.91779432901351e-05, + "loss": 3.1545, + "step": 34060 + }, + { + "epoch": 68.43, + "learning_rate": 9.916981871832549e-05, + "loss": 3.1396, + "step": 34080 + }, + { + "epoch": 68.47, + "learning_rate": 9.916165453128106e-05, + "loss": 3.1399, + "step": 34100 + }, + { + "epoch": 68.51, + "learning_rate": 9.915345073557953e-05, + "loss": 3.1532, + "step": 34120 + }, + { + "epoch": 68.55, + "learning_rate": 9.914520733783055e-05, + "loss": 3.1623, + "step": 34140 + }, + { + "epoch": 68.59, + "learning_rate": 9.913692434467569e-05, + "loss": 3.1767, + "step": 34160 + }, + { + "epoch": 68.63, + "learning_rate": 9.912860176278842e-05, + "loss": 3.1742, + "step": 34180 + }, + { + "epoch": 68.67, + "learning_rate": 9.912023959887408e-05, + "loss": 3.181, + "step": 34200 + }, + { + "epoch": 68.71, + "learning_rate": 9.91118378596699e-05, + "loss": 3.1757, + "step": 34220 + }, + { + "epoch": 68.76, + "learning_rate": 9.910339655194502e-05, + "loss": 3.1899, + "step": 34240 + }, + { + "epoch": 68.8, + "learning_rate": 9.909491568250047e-05, + "loss": 3.1717, + "step": 34260 + }, + { + "epoch": 68.84, + "learning_rate": 9.90863952581691e-05, + "loss": 3.1542, + "step": 34280 + }, + { + "epoch": 68.88, + "learning_rate": 9.907783528581568e-05, + "loss": 3.1626, + "step": 34300 + }, + { + "epoch": 68.92, + "learning_rate": 9.906923577233682e-05, + "loss": 3.1796, + "step": 34320 + }, + { + "epoch": 68.96, + "learning_rate": 9.906059672466101e-05, + "loss": 3.1792, + "step": 34340 + }, + { + "epoch": 69.0, + "learning_rate": 9.905191814974854e-05, + "loss": 3.2085, + "step": 34360 + }, + { + "epoch": 69.04, + "learning_rate": 9.90432000545916e-05, + "loss": 3.0709, + "step": 34380 + }, + { + "epoch": 69.08, + "learning_rate": 9.903444244621423e-05, + "loss": 3.0754, + "step": 34400 + }, + { + "epoch": 69.12, + "learning_rate": 9.902564533167222e-05, + "loss": 3.0979, + "step": 34420 + }, + { + "epoch": 69.16, + "learning_rate": 9.901680871805329e-05, + "loss": 3.1183, + "step": 34440 + }, + { + "epoch": 69.2, + "learning_rate": 9.900793261247692e-05, + "loss": 3.1097, + "step": 34460 + }, + { + "epoch": 69.24, + "learning_rate": 9.899901702209445e-05, + "loss": 3.1084, + "step": 34480 + }, + { + "epoch": 69.28, + "learning_rate": 9.899006195408898e-05, + "loss": 3.1209, + "step": 34500 + }, + { + "epoch": 69.32, + "learning_rate": 9.898106741567545e-05, + "loss": 3.1414, + "step": 34520 + }, + { + "epoch": 69.36, + "learning_rate": 9.897203341410062e-05, + "loss": 3.1434, + "step": 34540 + }, + { + "epoch": 69.4, + "learning_rate": 9.8962959956643e-05, + "loss": 3.1281, + "step": 34560 + }, + { + "epoch": 69.44, + "learning_rate": 9.895384705061292e-05, + "loss": 3.142, + "step": 34580 + }, + { + "epoch": 69.48, + "learning_rate": 9.894469470335251e-05, + "loss": 3.1372, + "step": 34600 + }, + { + "epoch": 69.52, + "learning_rate": 9.89355029222356e-05, + "loss": 3.1263, + "step": 34620 + }, + { + "epoch": 69.56, + "learning_rate": 9.892627171466786e-05, + "loss": 3.1402, + "step": 34640 + }, + { + "epoch": 69.6, + "learning_rate": 9.891700108808674e-05, + "loss": 3.1607, + "step": 34660 + }, + { + "epoch": 69.64, + "learning_rate": 9.890769104996137e-05, + "loss": 3.1368, + "step": 34680 + }, + { + "epoch": 69.68, + "learning_rate": 9.889834160779271e-05, + "loss": 3.1639, + "step": 34700 + }, + { + "epoch": 69.72, + "learning_rate": 9.888895276911342e-05, + "loss": 3.1468, + "step": 34720 + }, + { + "epoch": 69.76, + "learning_rate": 9.887952454148792e-05, + "loss": 3.152, + "step": 34740 + }, + { + "epoch": 69.8, + "learning_rate": 9.88700569325124e-05, + "loss": 3.1412, + "step": 34760 + }, + { + "epoch": 69.84, + "learning_rate": 9.886054994981466e-05, + "loss": 3.1594, + "step": 34780 + }, + { + "epoch": 69.88, + "learning_rate": 9.88510036010544e-05, + "loss": 3.144, + "step": 34800 + }, + { + "epoch": 69.92, + "learning_rate": 9.884141789392285e-05, + "loss": 3.1563, + "step": 34820 + }, + { + "epoch": 69.96, + "learning_rate": 9.883179283614311e-05, + "loss": 3.1517, + "step": 34840 + }, + { + "epoch": 70.0, + "learning_rate": 9.882212843546988e-05, + "loss": 3.1415, + "step": 34860 + }, + { + "epoch": 70.04, + "learning_rate": 9.881242469968962e-05, + "loss": 3.0655, + "step": 34880 + }, + { + "epoch": 70.08, + "learning_rate": 9.880268163662042e-05, + "loss": 3.0566, + "step": 34900 + }, + { + "epoch": 70.12, + "learning_rate": 9.879289925411212e-05, + "loss": 3.072, + "step": 34920 + }, + { + "epoch": 70.16, + "learning_rate": 9.878307756004619e-05, + "loss": 3.0809, + "step": 34940 + }, + { + "epoch": 70.2, + "learning_rate": 9.87732165623358e-05, + "loss": 3.0748, + "step": 34960 + }, + { + "epoch": 70.24, + "learning_rate": 9.876331626892576e-05, + "loss": 3.0804, + "step": 34980 + }, + { + "epoch": 70.28, + "learning_rate": 9.875337668779259e-05, + "loss": 3.1111, + "step": 35000 + }, + { + "epoch": 70.28, + "eval_loss": 3.3502800464630127, + "eval_runtime": 34.661, + "eval_samples_per_second": 22.1, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.003550789684460403, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.493109950681752e-05, + "step": 35000 + }, + { + "epoch": 70.32, + "learning_rate": 9.874339782694441e-05, + "loss": 3.0931, + "step": 35020 + }, + { + "epoch": 70.36, + "learning_rate": 9.873337969442101e-05, + "loss": 3.1004, + "step": 35040 + }, + { + "epoch": 70.4, + "learning_rate": 9.872332229829383e-05, + "loss": 3.1185, + "step": 35060 + }, + { + "epoch": 70.44, + "learning_rate": 9.871322564666592e-05, + "loss": 3.1029, + "step": 35080 + }, + { + "epoch": 70.48, + "learning_rate": 9.870308974767197e-05, + "loss": 3.1085, + "step": 35100 + }, + { + "epoch": 70.52, + "learning_rate": 9.869291460947829e-05, + "loss": 3.1154, + "step": 35120 + }, + { + "epoch": 70.56, + "learning_rate": 9.86827002402828e-05, + "loss": 3.132, + "step": 35140 + }, + { + "epoch": 70.6, + "learning_rate": 9.867244664831507e-05, + "loss": 3.1104, + "step": 35160 + }, + { + "epoch": 70.64, + "learning_rate": 9.866215384183619e-05, + "loss": 3.1237, + "step": 35180 + }, + { + "epoch": 70.68, + "learning_rate": 9.86518218291389e-05, + "loss": 3.1239, + "step": 35200 + }, + { + "epoch": 70.72, + "learning_rate": 9.864145061854753e-05, + "loss": 3.1342, + "step": 35220 + }, + { + "epoch": 70.76, + "learning_rate": 9.863104021841796e-05, + "loss": 3.1371, + "step": 35240 + }, + { + "epoch": 70.8, + "learning_rate": 9.862059063713767e-05, + "loss": 3.1453, + "step": 35260 + }, + { + "epoch": 70.84, + "learning_rate": 9.861010188312568e-05, + "loss": 3.1426, + "step": 35280 + }, + { + "epoch": 70.88, + "learning_rate": 9.859957396483262e-05, + "loss": 3.1445, + "step": 35300 + }, + { + "epoch": 70.92, + "learning_rate": 9.858900689074064e-05, + "loss": 3.1455, + "step": 35320 + }, + { + "epoch": 70.96, + "learning_rate": 9.857840066936341e-05, + "loss": 3.1393, + "step": 35340 + }, + { + "epoch": 71.0, + "learning_rate": 9.856775530924621e-05, + "loss": 3.1426, + "step": 35360 + }, + { + "epoch": 71.04, + "learning_rate": 9.855707081896579e-05, + "loss": 3.0355, + "step": 35380 + }, + { + "epoch": 71.08, + "learning_rate": 9.854634720713047e-05, + "loss": 3.0443, + "step": 35400 + }, + { + "epoch": 71.12, + "learning_rate": 9.853558448238003e-05, + "loss": 3.0502, + "step": 35420 + }, + { + "epoch": 71.16, + "learning_rate": 9.852478265338585e-05, + "loss": 3.056, + "step": 35440 + }, + { + "epoch": 71.2, + "learning_rate": 9.851394172885074e-05, + "loss": 3.0442, + "step": 35460 + }, + { + "epoch": 71.24, + "learning_rate": 9.850306171750905e-05, + "loss": 3.0735, + "step": 35480 + }, + { + "epoch": 71.29, + "learning_rate": 9.84921426281266e-05, + "loss": 3.0828, + "step": 35500 + }, + { + "epoch": 71.33, + "learning_rate": 9.848118446950072e-05, + "loss": 3.0725, + "step": 35520 + }, + { + "epoch": 71.37, + "learning_rate": 9.847073803896042e-05, + "loss": 3.0733, + "step": 35540 + }, + { + "epoch": 71.41, + "learning_rate": 9.845970372073228e-05, + "loss": 3.0962, + "step": 35560 + }, + { + "epoch": 71.45, + "learning_rate": 9.844863035939615e-05, + "loss": 3.0906, + "step": 35580 + }, + { + "epoch": 71.49, + "learning_rate": 9.843751796387363e-05, + "loss": 3.1018, + "step": 35600 + }, + { + "epoch": 71.53, + "learning_rate": 9.842636654311778e-05, + "loss": 3.098, + "step": 35620 + }, + { + "epoch": 71.57, + "learning_rate": 9.841517610611309e-05, + "loss": 3.1129, + "step": 35640 + }, + { + "epoch": 71.61, + "learning_rate": 9.840394666187546e-05, + "loss": 3.1123, + "step": 35660 + }, + { + "epoch": 71.65, + "learning_rate": 9.839267821945232e-05, + "loss": 3.1045, + "step": 35680 + }, + { + "epoch": 71.69, + "learning_rate": 9.838137078792243e-05, + "loss": 3.111, + "step": 35700 + }, + { + "epoch": 71.73, + "learning_rate": 9.837002437639593e-05, + "loss": 3.1092, + "step": 35720 + }, + { + "epoch": 71.77, + "learning_rate": 9.835863899401448e-05, + "loss": 3.1083, + "step": 35740 + }, + { + "epoch": 71.81, + "learning_rate": 9.834721464995105e-05, + "loss": 3.1255, + "step": 35760 + }, + { + "epoch": 71.85, + "learning_rate": 9.833575135341002e-05, + "loss": 3.1141, + "step": 35780 + }, + { + "epoch": 71.89, + "learning_rate": 9.832424911362719e-05, + "loss": 3.1084, + "step": 35800 + }, + { + "epoch": 71.93, + "learning_rate": 9.831270793986968e-05, + "loss": 3.1229, + "step": 35820 + }, + { + "epoch": 71.97, + "learning_rate": 9.830112784143601e-05, + "loss": 3.1167, + "step": 35840 + }, + { + "epoch": 72.01, + "learning_rate": 9.828950882765608e-05, + "loss": 3.1029, + "step": 35860 + }, + { + "epoch": 72.05, + "learning_rate": 9.827785090789109e-05, + "loss": 3.0329, + "step": 35880 + }, + { + "epoch": 72.09, + "learning_rate": 9.826615409153362e-05, + "loss": 3.0221, + "step": 35900 + }, + { + "epoch": 72.13, + "learning_rate": 9.82544183880076e-05, + "loss": 3.0364, + "step": 35920 + }, + { + "epoch": 72.17, + "learning_rate": 9.824264380676826e-05, + "loss": 3.0317, + "step": 35940 + }, + { + "epoch": 72.21, + "learning_rate": 9.823083035730218e-05, + "loss": 3.0568, + "step": 35960 + }, + { + "epoch": 72.25, + "learning_rate": 9.821897804912722e-05, + "loss": 3.0484, + "step": 35980 + }, + { + "epoch": 72.29, + "learning_rate": 9.820708689179259e-05, + "loss": 3.0692, + "step": 36000 + }, + { + "epoch": 72.29, + "eval_loss": 3.3271844387054443, + "eval_runtime": 34.1008, + "eval_samples_per_second": 22.463, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.004147898141934031, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.3828461416001169e-05, + "step": 36000 + }, + { + "epoch": 72.33, + "learning_rate": 9.819515689487877e-05, + "loss": 3.0693, + "step": 36020 + }, + { + "epoch": 72.37, + "learning_rate": 9.818318806799755e-05, + "loss": 3.0591, + "step": 36040 + }, + { + "epoch": 72.41, + "learning_rate": 9.817118042079201e-05, + "loss": 3.0918, + "step": 36060 + }, + { + "epoch": 72.45, + "learning_rate": 9.815913396293648e-05, + "loss": 3.0676, + "step": 36080 + }, + { + "epoch": 72.49, + "learning_rate": 9.814704870413657e-05, + "loss": 3.0751, + "step": 36100 + }, + { + "epoch": 72.53, + "learning_rate": 9.813492465412916e-05, + "loss": 3.079, + "step": 36120 + }, + { + "epoch": 72.57, + "learning_rate": 9.812276182268236e-05, + "loss": 3.076, + "step": 36140 + }, + { + "epoch": 72.61, + "learning_rate": 9.811056021959559e-05, + "loss": 3.0847, + "step": 36160 + }, + { + "epoch": 72.65, + "learning_rate": 9.809831985469942e-05, + "loss": 3.0833, + "step": 36180 + }, + { + "epoch": 72.69, + "learning_rate": 9.80860407378557e-05, + "loss": 3.0961, + "step": 36200 + }, + { + "epoch": 72.73, + "learning_rate": 9.807372287895748e-05, + "loss": 3.0964, + "step": 36220 + }, + { + "epoch": 72.77, + "learning_rate": 9.806136628792907e-05, + "loss": 3.1027, + "step": 36240 + }, + { + "epoch": 72.81, + "learning_rate": 9.804897097472594e-05, + "loss": 3.0719, + "step": 36260 + }, + { + "epoch": 72.85, + "learning_rate": 9.803653694933476e-05, + "loss": 3.098, + "step": 36280 + }, + { + "epoch": 72.89, + "learning_rate": 9.802406422177339e-05, + "loss": 3.1067, + "step": 36300 + }, + { + "epoch": 72.93, + "learning_rate": 9.801155280209091e-05, + "loss": 3.0924, + "step": 36320 + }, + { + "epoch": 72.97, + "learning_rate": 9.799900270036753e-05, + "loss": 3.0783, + "step": 36340 + }, + { + "epoch": 73.01, + "learning_rate": 9.798641392671463e-05, + "loss": 3.0699, + "step": 36360 + }, + { + "epoch": 73.05, + "learning_rate": 9.797378649127478e-05, + "loss": 3.0047, + "step": 36380 + }, + { + "epoch": 73.09, + "learning_rate": 9.796112040422165e-05, + "loss": 3.0041, + "step": 36400 + }, + { + "epoch": 73.13, + "learning_rate": 9.794841567576011e-05, + "loss": 3.0137, + "step": 36420 + }, + { + "epoch": 73.17, + "learning_rate": 9.79356723161261e-05, + "loss": 3.0278, + "step": 36440 + }, + { + "epoch": 73.21, + "learning_rate": 9.792289033558673e-05, + "loss": 3.0327, + "step": 36460 + }, + { + "epoch": 73.25, + "learning_rate": 9.791006974444022e-05, + "loss": 3.0424, + "step": 36480 + }, + { + "epoch": 73.29, + "learning_rate": 9.789721055301586e-05, + "loss": 3.0153, + "step": 36500 + }, + { + "epoch": 73.33, + "learning_rate": 9.788431277167408e-05, + "loss": 3.0601, + "step": 36520 + }, + { + "epoch": 73.37, + "learning_rate": 9.787137641080641e-05, + "loss": 3.0392, + "step": 36540 + }, + { + "epoch": 73.41, + "learning_rate": 9.785840148083543e-05, + "loss": 3.0527, + "step": 36560 + }, + { + "epoch": 73.45, + "learning_rate": 9.784538799221478e-05, + "loss": 3.0511, + "step": 36580 + }, + { + "epoch": 73.49, + "learning_rate": 9.783233595542923e-05, + "loss": 3.0584, + "step": 36600 + }, + { + "epoch": 73.53, + "learning_rate": 9.781924538099456e-05, + "loss": 3.0592, + "step": 36620 + }, + { + "epoch": 73.57, + "learning_rate": 9.78061162794576e-05, + "loss": 3.0493, + "step": 36640 + }, + { + "epoch": 73.61, + "learning_rate": 9.779294866139624e-05, + "loss": 3.0679, + "step": 36660 + }, + { + "epoch": 73.65, + "learning_rate": 9.777974253741938e-05, + "loss": 3.063, + "step": 36680 + }, + { + "epoch": 73.69, + "learning_rate": 9.776649791816698e-05, + "loss": 3.0708, + "step": 36700 + }, + { + "epoch": 73.73, + "learning_rate": 9.775321481430996e-05, + "loss": 3.0729, + "step": 36720 + }, + { + "epoch": 73.78, + "learning_rate": 9.773989323655028e-05, + "loss": 3.0582, + "step": 36740 + }, + { + "epoch": 73.82, + "learning_rate": 9.772653319562093e-05, + "loss": 3.0633, + "step": 36760 + }, + { + "epoch": 73.86, + "learning_rate": 9.771313470228583e-05, + "loss": 3.0814, + "step": 36780 + }, + { + "epoch": 73.9, + "learning_rate": 9.769969776733988e-05, + "loss": 3.102, + "step": 36800 + }, + { + "epoch": 73.94, + "learning_rate": 9.768622240160901e-05, + "loss": 3.0836, + "step": 36820 + }, + { + "epoch": 73.98, + "learning_rate": 9.767270861595005e-05, + "loss": 3.0764, + "step": 36840 + }, + { + "epoch": 74.02, + "learning_rate": 9.765915642125087e-05, + "loss": 3.0478, + "step": 36860 + }, + { + "epoch": 74.06, + "learning_rate": 9.764556582843016e-05, + "loss": 2.9825, + "step": 36880 + }, + { + "epoch": 74.1, + "learning_rate": 9.763193684843765e-05, + "loss": 2.9903, + "step": 36900 + }, + { + "epoch": 74.14, + "learning_rate": 9.761826949225397e-05, + "loss": 2.9747, + "step": 36920 + }, + { + "epoch": 74.18, + "learning_rate": 9.760456377089062e-05, + "loss": 2.9937, + "step": 36940 + }, + { + "epoch": 74.22, + "learning_rate": 9.759081969539012e-05, + "loss": 2.9986, + "step": 36960 + }, + { + "epoch": 74.26, + "learning_rate": 9.757703727682574e-05, + "loss": 3.0196, + "step": 36980 + }, + { + "epoch": 74.3, + "learning_rate": 9.75632165263018e-05, + "loss": 3.0151, + "step": 37000 + }, + { + "epoch": 74.3, + "eval_loss": 3.2996249198913574, + "eval_runtime": 32.4949, + "eval_samples_per_second": 23.573, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0022485641510888352, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0, + "step": 37000 + }, + { + "epoch": 74.34, + "learning_rate": 9.75493574549534e-05, + "loss": 3.0193, + "step": 37020 + }, + { + "epoch": 74.38, + "learning_rate": 9.753546007394654e-05, + "loss": 3.029, + "step": 37040 + }, + { + "epoch": 74.42, + "learning_rate": 9.752152439447808e-05, + "loss": 3.0389, + "step": 37060 + }, + { + "epoch": 74.46, + "learning_rate": 9.750755042777577e-05, + "loss": 3.0294, + "step": 37080 + }, + { + "epoch": 74.5, + "learning_rate": 9.749353818509817e-05, + "loss": 3.0392, + "step": 37100 + }, + { + "epoch": 74.54, + "learning_rate": 9.74794876777347e-05, + "loss": 3.0369, + "step": 37120 + }, + { + "epoch": 74.58, + "learning_rate": 9.746539891700556e-05, + "loss": 3.0714, + "step": 37140 + }, + { + "epoch": 74.62, + "learning_rate": 9.745127191426187e-05, + "loss": 3.0569, + "step": 37160 + }, + { + "epoch": 74.66, + "learning_rate": 9.743710668088545e-05, + "loss": 3.0607, + "step": 37180 + }, + { + "epoch": 74.7, + "learning_rate": 9.7422903228289e-05, + "loss": 3.0414, + "step": 37200 + }, + { + "epoch": 74.74, + "learning_rate": 9.7408661567916e-05, + "loss": 3.0644, + "step": 37220 + }, + { + "epoch": 74.78, + "learning_rate": 9.739438171124064e-05, + "loss": 3.0654, + "step": 37240 + }, + { + "epoch": 74.82, + "learning_rate": 9.7380063669768e-05, + "loss": 3.0747, + "step": 37260 + }, + { + "epoch": 74.86, + "learning_rate": 9.736570745503386e-05, + "loss": 3.0481, + "step": 37280 + }, + { + "epoch": 74.9, + "learning_rate": 9.735131307860475e-05, + "loss": 3.0671, + "step": 37300 + }, + { + "epoch": 74.94, + "learning_rate": 9.733688055207794e-05, + "loss": 3.0548, + "step": 37320 + }, + { + "epoch": 74.98, + "learning_rate": 9.732240988708148e-05, + "loss": 3.0532, + "step": 37340 + }, + { + "epoch": 75.02, + "learning_rate": 9.730790109527412e-05, + "loss": 3.0094, + "step": 37360 + }, + { + "epoch": 75.06, + "learning_rate": 9.729335418834534e-05, + "loss": 2.9633, + "step": 37380 + }, + { + "epoch": 75.1, + "learning_rate": 9.72787691780153e-05, + "loss": 2.956, + "step": 37400 + }, + { + "epoch": 75.14, + "learning_rate": 9.726414607603492e-05, + "loss": 2.9919, + "step": 37420 + }, + { + "epoch": 75.18, + "learning_rate": 9.724948489418571e-05, + "loss": 2.9959, + "step": 37440 + }, + { + "epoch": 75.22, + "learning_rate": 9.723478564427995e-05, + "loss": 2.9878, + "step": 37460 + }, + { + "epoch": 75.26, + "learning_rate": 9.722004833816056e-05, + "loss": 3.0033, + "step": 37480 + }, + { + "epoch": 75.3, + "learning_rate": 9.720527298770112e-05, + "loss": 2.9896, + "step": 37500 + }, + { + "epoch": 75.34, + "learning_rate": 9.719045960480588e-05, + "loss": 3.0096, + "step": 37520 + }, + { + "epoch": 75.38, + "learning_rate": 9.717560820140969e-05, + "loss": 3.0183, + "step": 37540 + }, + { + "epoch": 75.42, + "learning_rate": 9.716071878947805e-05, + "loss": 3.0273, + "step": 37560 + }, + { + "epoch": 75.46, + "learning_rate": 9.714579138100712e-05, + "loss": 3.0128, + "step": 37580 + }, + { + "epoch": 75.5, + "learning_rate": 9.713082598802364e-05, + "loss": 3.0249, + "step": 37600 + }, + { + "epoch": 75.54, + "learning_rate": 9.711582262258493e-05, + "loss": 3.033, + "step": 37620 + }, + { + "epoch": 75.58, + "learning_rate": 9.710078129677896e-05, + "loss": 3.0167, + "step": 37640 + }, + { + "epoch": 75.62, + "learning_rate": 9.708570202272422e-05, + "loss": 3.0468, + "step": 37660 + }, + { + "epoch": 75.66, + "learning_rate": 9.707058481256985e-05, + "loss": 3.0071, + "step": 37680 + }, + { + "epoch": 75.7, + "learning_rate": 9.705542967849547e-05, + "loss": 3.0225, + "step": 37700 + }, + { + "epoch": 75.74, + "learning_rate": 9.704023663271136e-05, + "loss": 3.0377, + "step": 37720 + }, + { + "epoch": 75.78, + "learning_rate": 9.702500568745821e-05, + "loss": 3.0273, + "step": 37740 + }, + { + "epoch": 75.82, + "learning_rate": 9.700973685500735e-05, + "loss": 3.0417, + "step": 37760 + }, + { + "epoch": 75.86, + "learning_rate": 9.699443014766063e-05, + "loss": 3.0476, + "step": 37780 + }, + { + "epoch": 75.9, + "learning_rate": 9.697985370529101e-05, + "loss": 3.0282, + "step": 37800 + }, + { + "epoch": 75.94, + "learning_rate": 9.696447317739598e-05, + "loss": 3.0677, + "step": 37820 + }, + { + "epoch": 75.98, + "learning_rate": 9.694905481107317e-05, + "loss": 3.0528, + "step": 37840 + }, + { + "epoch": 76.02, + "learning_rate": 9.693359861874491e-05, + "loss": 2.9857, + "step": 37860 + }, + { + "epoch": 76.06, + "learning_rate": 9.691810461286397e-05, + "loss": 2.9519, + "step": 37880 + }, + { + "epoch": 76.1, + "learning_rate": 9.690257280591358e-05, + "loss": 2.9586, + "step": 37900 + }, + { + "epoch": 76.14, + "learning_rate": 9.688700321040744e-05, + "loss": 2.9545, + "step": 37920 + }, + { + "epoch": 76.18, + "learning_rate": 9.687139583888972e-05, + "loss": 2.9574, + "step": 37940 + }, + { + "epoch": 76.22, + "learning_rate": 9.685575070393494e-05, + "loss": 2.9774, + "step": 37960 + }, + { + "epoch": 76.27, + "learning_rate": 9.684006781814814e-05, + "loss": 2.982, + "step": 37980 + }, + { + "epoch": 76.31, + "learning_rate": 9.682434719416473e-05, + "loss": 2.9887, + "step": 38000 + }, + { + "epoch": 76.31, + "eval_loss": 3.2820749282836914, + "eval_runtime": 32.4566, + "eval_samples_per_second": 23.601, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.003648645544944527, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 4.759573542210618e-05, + "step": 38000 + }, + { + "epoch": 76.35, + "learning_rate": 9.680858884465054e-05, + "loss": 2.9941, + "step": 38020 + }, + { + "epoch": 76.39, + "learning_rate": 9.679279278230179e-05, + "loss": 3.0156, + "step": 38040 + }, + { + "epoch": 76.43, + "learning_rate": 9.677695901984507e-05, + "loss": 3.0122, + "step": 38060 + }, + { + "epoch": 76.47, + "learning_rate": 9.676108757003735e-05, + "loss": 2.9893, + "step": 38080 + }, + { + "epoch": 76.51, + "learning_rate": 9.6745178445666e-05, + "loss": 3.0065, + "step": 38100 + }, + { + "epoch": 76.55, + "learning_rate": 9.672923165954868e-05, + "loss": 3.0045, + "step": 38120 + }, + { + "epoch": 76.59, + "learning_rate": 9.671324722453345e-05, + "loss": 3.0089, + "step": 38140 + }, + { + "epoch": 76.63, + "learning_rate": 9.669722515349868e-05, + "loss": 3.0217, + "step": 38160 + }, + { + "epoch": 76.67, + "learning_rate": 9.668116545935304e-05, + "loss": 3.0063, + "step": 38180 + }, + { + "epoch": 76.71, + "learning_rate": 9.666506815503558e-05, + "loss": 3.0103, + "step": 38200 + }, + { + "epoch": 76.75, + "learning_rate": 9.664893325351555e-05, + "loss": 3.0312, + "step": 38220 + }, + { + "epoch": 76.79, + "learning_rate": 9.66327607677926e-05, + "loss": 3.0006, + "step": 38240 + }, + { + "epoch": 76.83, + "learning_rate": 9.661655071089657e-05, + "loss": 3.0228, + "step": 38260 + }, + { + "epoch": 76.87, + "learning_rate": 9.660030309588764e-05, + "loss": 3.0238, + "step": 38280 + }, + { + "epoch": 76.91, + "learning_rate": 9.65840179358562e-05, + "loss": 3.0198, + "step": 38300 + }, + { + "epoch": 76.95, + "learning_rate": 9.656769524392292e-05, + "loss": 3.0225, + "step": 38320 + }, + { + "epoch": 76.99, + "learning_rate": 9.65513350332387e-05, + "loss": 3.0311, + "step": 38340 + }, + { + "epoch": 77.03, + "learning_rate": 9.653493731698467e-05, + "loss": 2.9647, + "step": 38360 + }, + { + "epoch": 77.07, + "learning_rate": 9.651850210837215e-05, + "loss": 2.9313, + "step": 38380 + }, + { + "epoch": 77.11, + "learning_rate": 9.65020294206427e-05, + "loss": 2.9422, + "step": 38400 + }, + { + "epoch": 77.15, + "learning_rate": 9.64855192670681e-05, + "loss": 2.9553, + "step": 38420 + }, + { + "epoch": 77.19, + "learning_rate": 9.646897166095024e-05, + "loss": 2.9464, + "step": 38440 + }, + { + "epoch": 77.23, + "learning_rate": 9.645238661562124e-05, + "loss": 2.9534, + "step": 38460 + }, + { + "epoch": 77.27, + "learning_rate": 9.643576414444338e-05, + "loss": 2.9644, + "step": 38480 + }, + { + "epoch": 77.31, + "learning_rate": 9.641910426080908e-05, + "loss": 2.9647, + "step": 38500 + }, + { + "epoch": 77.35, + "learning_rate": 9.64024069781409e-05, + "loss": 2.9672, + "step": 38520 + }, + { + "epoch": 77.39, + "learning_rate": 9.638567230989155e-05, + "loss": 2.9816, + "step": 38540 + }, + { + "epoch": 77.43, + "learning_rate": 9.636890026954385e-05, + "loss": 2.9895, + "step": 38560 + }, + { + "epoch": 77.47, + "learning_rate": 9.635209087061072e-05, + "loss": 2.9925, + "step": 38580 + }, + { + "epoch": 77.51, + "learning_rate": 9.63352441266352e-05, + "loss": 2.9766, + "step": 38600 + }, + { + "epoch": 77.55, + "learning_rate": 9.63183600511904e-05, + "loss": 2.9867, + "step": 38620 + }, + { + "epoch": 77.59, + "learning_rate": 9.630143865787951e-05, + "loss": 3.0005, + "step": 38640 + }, + { + "epoch": 77.63, + "learning_rate": 9.62844799603358e-05, + "loss": 2.9982, + "step": 38660 + }, + { + "epoch": 77.67, + "learning_rate": 9.62674839722226e-05, + "loss": 3.0178, + "step": 38680 + }, + { + "epoch": 77.71, + "learning_rate": 9.625045070723324e-05, + "loss": 2.9839, + "step": 38700 + }, + { + "epoch": 77.75, + "learning_rate": 9.623338017909113e-05, + "loss": 2.9991, + "step": 38720 + }, + { + "epoch": 77.79, + "learning_rate": 9.621627240154968e-05, + "loss": 3.0064, + "step": 38740 + }, + { + "epoch": 77.83, + "learning_rate": 9.619912738839233e-05, + "loss": 3.0219, + "step": 38760 + }, + { + "epoch": 77.87, + "learning_rate": 9.61819451534325e-05, + "loss": 3.025, + "step": 38780 + }, + { + "epoch": 77.91, + "learning_rate": 9.61647257105136e-05, + "loss": 2.9823, + "step": 38800 + }, + { + "epoch": 77.95, + "learning_rate": 9.614746907350904e-05, + "loss": 2.9937, + "step": 38820 + }, + { + "epoch": 77.99, + "learning_rate": 9.613017525632215e-05, + "loss": 3.0045, + "step": 38840 + }, + { + "epoch": 78.03, + "learning_rate": 9.611284427288628e-05, + "loss": 2.9306, + "step": 38860 + }, + { + "epoch": 78.07, + "learning_rate": 9.609547613716468e-05, + "loss": 2.9167, + "step": 38880 + }, + { + "epoch": 78.11, + "learning_rate": 9.607807086315055e-05, + "loss": 2.9186, + "step": 38900 + }, + { + "epoch": 78.15, + "learning_rate": 9.606062846486698e-05, + "loss": 2.9273, + "step": 38920 + }, + { + "epoch": 78.19, + "learning_rate": 9.604314895636702e-05, + "loss": 2.9386, + "step": 38940 + }, + { + "epoch": 78.23, + "learning_rate": 9.602563235173356e-05, + "loss": 2.9418, + "step": 38960 + }, + { + "epoch": 78.27, + "learning_rate": 9.600807866507946e-05, + "loss": 2.9469, + "step": 38980 + }, + { + "epoch": 78.31, + "learning_rate": 9.59904879105474e-05, + "loss": 2.976, + "step": 39000 + }, + { + "epoch": 78.31, + "eval_loss": 3.26127028465271, + "eval_runtime": 34.2926, + "eval_samples_per_second": 22.337, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.003476504865082289, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 3.500747904113057e-05, + "step": 39000 + }, + { + "epoch": 78.35, + "learning_rate": 9.597286010230989e-05, + "loss": 2.9441, + "step": 39020 + }, + { + "epoch": 78.39, + "learning_rate": 9.595519525456937e-05, + "loss": 2.9569, + "step": 39040 + }, + { + "epoch": 78.43, + "learning_rate": 9.593749338155809e-05, + "loss": 2.9659, + "step": 39060 + }, + { + "epoch": 78.47, + "learning_rate": 9.591975449753812e-05, + "loss": 2.9482, + "step": 39080 + }, + { + "epoch": 78.51, + "learning_rate": 9.590197861680137e-05, + "loss": 2.9667, + "step": 39100 + }, + { + "epoch": 78.55, + "learning_rate": 9.588416575366953e-05, + "loss": 2.9788, + "step": 39120 + }, + { + "epoch": 78.59, + "learning_rate": 9.58663159224941e-05, + "loss": 2.9627, + "step": 39140 + }, + { + "epoch": 78.63, + "learning_rate": 9.58484291376564e-05, + "loss": 2.9802, + "step": 39160 + }, + { + "epoch": 78.67, + "learning_rate": 9.583050541356747e-05, + "loss": 2.9825, + "step": 39180 + }, + { + "epoch": 78.71, + "learning_rate": 9.58125447646681e-05, + "loss": 2.9712, + "step": 39200 + }, + { + "epoch": 78.76, + "learning_rate": 9.579454720542889e-05, + "loss": 2.9771, + "step": 39220 + }, + { + "epoch": 78.8, + "learning_rate": 9.577651275035015e-05, + "loss": 3.003, + "step": 39240 + }, + { + "epoch": 78.84, + "learning_rate": 9.575844141396191e-05, + "loss": 2.9945, + "step": 39260 + }, + { + "epoch": 78.88, + "learning_rate": 9.574033321082392e-05, + "loss": 3.0115, + "step": 39280 + }, + { + "epoch": 78.92, + "learning_rate": 9.572218815552565e-05, + "loss": 2.9835, + "step": 39300 + }, + { + "epoch": 78.96, + "learning_rate": 9.57040062626862e-05, + "loss": 2.9967, + "step": 39320 + }, + { + "epoch": 79.0, + "learning_rate": 9.568578754695442e-05, + "loss": 3.0059, + "step": 39340 + }, + { + "epoch": 79.04, + "learning_rate": 9.566753202300881e-05, + "loss": 2.9099, + "step": 39360 + }, + { + "epoch": 79.08, + "learning_rate": 9.56492397055575e-05, + "loss": 2.9055, + "step": 39380 + }, + { + "epoch": 79.12, + "learning_rate": 9.563091060933829e-05, + "loss": 2.9025, + "step": 39400 + }, + { + "epoch": 79.16, + "learning_rate": 9.561254474911858e-05, + "loss": 2.9004, + "step": 39420 + }, + { + "epoch": 79.2, + "learning_rate": 9.559414213969543e-05, + "loss": 2.9223, + "step": 39440 + }, + { + "epoch": 79.24, + "learning_rate": 9.55757027958955e-05, + "loss": 2.9369, + "step": 39460 + }, + { + "epoch": 79.28, + "learning_rate": 9.555722673257501e-05, + "loss": 2.9522, + "step": 39480 + }, + { + "epoch": 79.32, + "learning_rate": 9.55387139646198e-05, + "loss": 2.9348, + "step": 39500 + }, + { + "epoch": 79.36, + "learning_rate": 9.552016450694529e-05, + "loss": 2.917, + "step": 39520 + }, + { + "epoch": 79.4, + "learning_rate": 9.550157837449643e-05, + "loss": 2.9467, + "step": 39540 + }, + { + "epoch": 79.44, + "learning_rate": 9.548295558224771e-05, + "loss": 2.9422, + "step": 39560 + }, + { + "epoch": 79.48, + "learning_rate": 9.546429614520322e-05, + "loss": 2.9477, + "step": 39580 + }, + { + "epoch": 79.52, + "learning_rate": 9.54456000783965e-05, + "loss": 2.9553, + "step": 39600 + }, + { + "epoch": 79.56, + "learning_rate": 9.542686739689065e-05, + "loss": 2.9578, + "step": 39620 + }, + { + "epoch": 79.6, + "learning_rate": 9.540809811577823e-05, + "loss": 2.9529, + "step": 39640 + }, + { + "epoch": 79.64, + "learning_rate": 9.538929225018133e-05, + "loss": 2.9576, + "step": 39660 + }, + { + "epoch": 79.68, + "learning_rate": 9.537044981525151e-05, + "loss": 2.9609, + "step": 39680 + }, + { + "epoch": 79.72, + "learning_rate": 9.535157082616975e-05, + "loss": 2.9714, + "step": 39700 + }, + { + "epoch": 79.76, + "learning_rate": 9.533265529814653e-05, + "loss": 2.9757, + "step": 39720 + }, + { + "epoch": 79.8, + "learning_rate": 9.531370324642175e-05, + "loss": 2.982, + "step": 39740 + }, + { + "epoch": 79.84, + "learning_rate": 9.529471468626472e-05, + "loss": 2.9656, + "step": 39760 + }, + { + "epoch": 79.88, + "learning_rate": 9.527568963297418e-05, + "loss": 2.9779, + "step": 39780 + }, + { + "epoch": 79.92, + "learning_rate": 9.525662810187827e-05, + "loss": 2.9783, + "step": 39800 + }, + { + "epoch": 79.96, + "learning_rate": 9.523753010833456e-05, + "loss": 2.9866, + "step": 39820 + }, + { + "epoch": 80.0, + "learning_rate": 9.521839566772993e-05, + "loss": 2.9894, + "step": 39840 + }, + { + "epoch": 80.04, + "learning_rate": 9.519922479548062e-05, + "loss": 2.889, + "step": 39860 + }, + { + "epoch": 80.08, + "learning_rate": 9.518001750703232e-05, + "loss": 2.8838, + "step": 39880 + }, + { + "epoch": 80.12, + "learning_rate": 9.516077381785994e-05, + "loss": 2.8894, + "step": 39900 + }, + { + "epoch": 80.16, + "learning_rate": 9.514149374346781e-05, + "loss": 2.8884, + "step": 39920 + }, + { + "epoch": 80.2, + "learning_rate": 9.512217729938953e-05, + "loss": 2.9035, + "step": 39940 + }, + { + "epoch": 80.24, + "learning_rate": 9.510282450118798e-05, + "loss": 2.9074, + "step": 39960 + }, + { + "epoch": 80.28, + "learning_rate": 9.508440568409111e-05, + "loss": 2.9303, + "step": 39980 + }, + { + "epoch": 80.32, + "learning_rate": 9.506498204022297e-05, + "loss": 2.9289, + "step": 40000 + }, + { + "epoch": 80.32, + "eval_loss": 3.2406046390533447, + "eval_runtime": 34.1861, + "eval_samples_per_second": 22.407, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.003786064256311405, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0001631853785900783, + "step": 40000 + }, + { + "epoch": 80.36, + "learning_rate": 9.504552208831279e-05, + "loss": 2.9236, + "step": 40020 + }, + { + "epoch": 80.4, + "learning_rate": 9.502602584403908e-05, + "loss": 2.9342, + "step": 40040 + }, + { + "epoch": 80.44, + "learning_rate": 9.50064933231096e-05, + "loss": 2.9398, + "step": 40060 + }, + { + "epoch": 80.48, + "learning_rate": 9.498692454126137e-05, + "loss": 2.9465, + "step": 40080 + }, + { + "epoch": 80.52, + "learning_rate": 9.496731951426059e-05, + "loss": 2.939, + "step": 40100 + }, + { + "epoch": 80.56, + "learning_rate": 9.494767825790269e-05, + "loss": 2.9324, + "step": 40120 + }, + { + "epoch": 80.6, + "learning_rate": 9.492800078801228e-05, + "loss": 2.9427, + "step": 40140 + }, + { + "epoch": 80.64, + "learning_rate": 9.490828712044312e-05, + "loss": 2.9425, + "step": 40160 + }, + { + "epoch": 80.68, + "learning_rate": 9.488853727107818e-05, + "loss": 2.9426, + "step": 40180 + }, + { + "epoch": 80.72, + "learning_rate": 9.486875125582955e-05, + "loss": 2.9397, + "step": 40200 + }, + { + "epoch": 80.76, + "learning_rate": 9.484892909063845e-05, + "loss": 2.9709, + "step": 40220 + }, + { + "epoch": 80.8, + "learning_rate": 9.482907079147524e-05, + "loss": 2.962, + "step": 40240 + }, + { + "epoch": 80.84, + "learning_rate": 9.480917637433942e-05, + "loss": 2.9646, + "step": 40260 + }, + { + "epoch": 80.88, + "learning_rate": 9.478924585525955e-05, + "loss": 2.9558, + "step": 40280 + }, + { + "epoch": 80.92, + "learning_rate": 9.476927925029326e-05, + "loss": 2.9407, + "step": 40300 + }, + { + "epoch": 80.96, + "learning_rate": 9.474927657552733e-05, + "loss": 2.9746, + "step": 40320 + }, + { + "epoch": 81.0, + "learning_rate": 9.472923784707752e-05, + "loss": 2.9458, + "step": 40340 + }, + { + "epoch": 81.04, + "learning_rate": 9.47091630810887e-05, + "loss": 2.8448, + "step": 40360 + }, + { + "epoch": 81.08, + "learning_rate": 9.468905229373472e-05, + "loss": 2.8732, + "step": 40380 + }, + { + "epoch": 81.12, + "learning_rate": 9.466890550121849e-05, + "loss": 2.8703, + "step": 40400 + }, + { + "epoch": 81.16, + "learning_rate": 9.464872271977189e-05, + "loss": 2.8854, + "step": 40420 + }, + { + "epoch": 81.2, + "learning_rate": 9.462850396565588e-05, + "loss": 2.899, + "step": 40440 + }, + { + "epoch": 81.24, + "learning_rate": 9.460824925516031e-05, + "loss": 2.9123, + "step": 40460 + }, + { + "epoch": 81.29, + "learning_rate": 9.458795860460403e-05, + "loss": 2.8992, + "step": 40480 + }, + { + "epoch": 81.33, + "learning_rate": 9.456763203033485e-05, + "loss": 2.9087, + "step": 40500 + }, + { + "epoch": 81.37, + "learning_rate": 9.454726954872954e-05, + "loss": 2.9122, + "step": 40520 + }, + { + "epoch": 81.41, + "learning_rate": 9.452687117619377e-05, + "loss": 2.9216, + "step": 40540 + }, + { + "epoch": 81.45, + "learning_rate": 9.450643692916216e-05, + "loss": 2.9037, + "step": 40560 + }, + { + "epoch": 81.49, + "learning_rate": 9.448596682409819e-05, + "loss": 2.9351, + "step": 40580 + }, + { + "epoch": 81.53, + "learning_rate": 9.446546087749425e-05, + "loss": 2.9026, + "step": 40600 + }, + { + "epoch": 81.57, + "learning_rate": 9.444491910587163e-05, + "loss": 2.9405, + "step": 40620 + }, + { + "epoch": 81.61, + "learning_rate": 9.442434152578046e-05, + "loss": 2.9294, + "step": 40640 + }, + { + "epoch": 81.65, + "learning_rate": 9.440372815379973e-05, + "loss": 2.9166, + "step": 40660 + }, + { + "epoch": 81.69, + "learning_rate": 9.438307900653724e-05, + "loss": 2.9495, + "step": 40680 + }, + { + "epoch": 81.73, + "learning_rate": 9.436239410062966e-05, + "loss": 2.9376, + "step": 40700 + }, + { + "epoch": 81.77, + "learning_rate": 9.434167345274244e-05, + "loss": 2.9399, + "step": 40720 + }, + { + "epoch": 81.81, + "learning_rate": 9.432091707956981e-05, + "loss": 2.9418, + "step": 40740 + }, + { + "epoch": 81.85, + "learning_rate": 9.430012499783482e-05, + "loss": 2.9517, + "step": 40760 + }, + { + "epoch": 81.89, + "learning_rate": 9.42792972242893e-05, + "loss": 2.9537, + "step": 40780 + }, + { + "epoch": 81.93, + "learning_rate": 9.425843377571377e-05, + "loss": 2.945, + "step": 40800 + }, + { + "epoch": 81.97, + "learning_rate": 9.423753466891754e-05, + "loss": 2.9447, + "step": 40820 + }, + { + "epoch": 82.01, + "learning_rate": 9.421659992073866e-05, + "loss": 2.9307, + "step": 40840 + }, + { + "epoch": 82.05, + "learning_rate": 9.419562954804388e-05, + "loss": 2.8401, + "step": 40860 + }, + { + "epoch": 82.09, + "learning_rate": 9.417462356772863e-05, + "loss": 2.862, + "step": 40880 + }, + { + "epoch": 82.13, + "learning_rate": 9.415358199671707e-05, + "loss": 2.8672, + "step": 40900 + }, + { + "epoch": 82.17, + "learning_rate": 9.4132504851962e-05, + "loss": 2.8718, + "step": 40920 + }, + { + "epoch": 82.21, + "learning_rate": 9.411139215044491e-05, + "loss": 2.8951, + "step": 40940 + }, + { + "epoch": 82.25, + "learning_rate": 9.409024390917591e-05, + "loss": 2.9005, + "step": 40960 + }, + { + "epoch": 82.29, + "learning_rate": 9.406906014519374e-05, + "loss": 2.8809, + "step": 40980 + }, + { + "epoch": 82.33, + "learning_rate": 9.404784087556582e-05, + "loss": 2.9078, + "step": 41000 + }, + { + "epoch": 82.33, + "eval_loss": 3.2119338512420654, + "eval_runtime": 34.0537, + "eval_samples_per_second": 22.494, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.003518717772293888, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 3.399695387293298e-05, + "step": 41000 + }, + { + "epoch": 82.37, + "learning_rate": 9.402658611738809e-05, + "loss": 2.8943, + "step": 41020 + }, + { + "epoch": 82.41, + "learning_rate": 9.400529588778513e-05, + "loss": 2.8835, + "step": 41040 + }, + { + "epoch": 82.45, + "learning_rate": 9.398397020391012e-05, + "loss": 2.9111, + "step": 41060 + }, + { + "epoch": 82.49, + "learning_rate": 9.396260908294476e-05, + "loss": 2.8911, + "step": 41080 + }, + { + "epoch": 82.53, + "learning_rate": 9.394121254209932e-05, + "loss": 2.9136, + "step": 41100 + }, + { + "epoch": 82.57, + "learning_rate": 9.391978059861261e-05, + "loss": 2.9107, + "step": 41120 + }, + { + "epoch": 82.61, + "learning_rate": 9.389831326975196e-05, + "loss": 2.9008, + "step": 41140 + }, + { + "epoch": 82.65, + "learning_rate": 9.38768105728132e-05, + "loss": 2.9042, + "step": 41160 + }, + { + "epoch": 82.69, + "learning_rate": 9.385527252512068e-05, + "loss": 2.9094, + "step": 41180 + }, + { + "epoch": 82.73, + "learning_rate": 9.38336991440272e-05, + "loss": 2.9249, + "step": 41200 + }, + { + "epoch": 82.77, + "learning_rate": 9.381209044691405e-05, + "loss": 2.9143, + "step": 41220 + }, + { + "epoch": 82.81, + "learning_rate": 9.379044645119098e-05, + "loss": 2.925, + "step": 41240 + }, + { + "epoch": 82.85, + "learning_rate": 9.376876717429615e-05, + "loss": 2.937, + "step": 41260 + }, + { + "epoch": 82.89, + "learning_rate": 9.374705263369617e-05, + "loss": 2.937, + "step": 41280 + }, + { + "epoch": 82.93, + "learning_rate": 9.372530284688603e-05, + "loss": 2.9281, + "step": 41300 + }, + { + "epoch": 82.97, + "learning_rate": 9.370351783138917e-05, + "loss": 2.9337, + "step": 41320 + }, + { + "epoch": 83.01, + "learning_rate": 9.36816976047574e-05, + "loss": 2.9131, + "step": 41340 + }, + { + "epoch": 83.05, + "learning_rate": 9.365984218457083e-05, + "loss": 2.8398, + "step": 41360 + }, + { + "epoch": 83.09, + "learning_rate": 9.3637951588438e-05, + "loss": 2.8443, + "step": 41380 + }, + { + "epoch": 83.13, + "learning_rate": 9.361602583399578e-05, + "loss": 2.8794, + "step": 41400 + }, + { + "epoch": 83.17, + "learning_rate": 9.359406493890934e-05, + "loss": 2.8691, + "step": 41420 + }, + { + "epoch": 83.21, + "learning_rate": 9.357206892087219e-05, + "loss": 2.8607, + "step": 41440 + }, + { + "epoch": 83.25, + "learning_rate": 9.355003779760612e-05, + "loss": 2.8774, + "step": 41460 + }, + { + "epoch": 83.29, + "learning_rate": 9.352797158686119e-05, + "loss": 2.8805, + "step": 41480 + }, + { + "epoch": 83.33, + "learning_rate": 9.350587030641578e-05, + "loss": 2.8667, + "step": 41500 + }, + { + "epoch": 83.37, + "learning_rate": 9.348373397407646e-05, + "loss": 2.8792, + "step": 41520 + }, + { + "epoch": 83.41, + "learning_rate": 9.346156260767809e-05, + "loss": 2.8821, + "step": 41540 + }, + { + "epoch": 83.45, + "learning_rate": 9.343935622508373e-05, + "loss": 2.8758, + "step": 41560 + }, + { + "epoch": 83.49, + "learning_rate": 9.341711484418468e-05, + "loss": 2.8779, + "step": 41580 + }, + { + "epoch": 83.53, + "learning_rate": 9.33948384829004e-05, + "loss": 2.8814, + "step": 41600 + }, + { + "epoch": 83.57, + "learning_rate": 9.337252715917857e-05, + "loss": 2.898, + "step": 41620 + }, + { + "epoch": 83.61, + "learning_rate": 9.3350180890995e-05, + "loss": 2.9043, + "step": 41640 + }, + { + "epoch": 83.65, + "learning_rate": 9.332779969635369e-05, + "loss": 2.9115, + "step": 41660 + }, + { + "epoch": 83.69, + "learning_rate": 9.330538359328675e-05, + "loss": 2.8964, + "step": 41680 + }, + { + "epoch": 83.73, + "learning_rate": 9.328293259985446e-05, + "loss": 2.9039, + "step": 41700 + }, + { + "epoch": 83.78, + "learning_rate": 9.326044673414513e-05, + "loss": 2.9035, + "step": 41720 + }, + { + "epoch": 83.82, + "learning_rate": 9.323792601427526e-05, + "loss": 2.9114, + "step": 41740 + }, + { + "epoch": 83.86, + "learning_rate": 9.321537045838936e-05, + "loss": 2.9285, + "step": 41760 + }, + { + "epoch": 83.9, + "learning_rate": 9.319278008466006e-05, + "loss": 2.9306, + "step": 41780 + }, + { + "epoch": 83.94, + "learning_rate": 9.317015491128797e-05, + "loss": 2.9257, + "step": 41800 + }, + { + "epoch": 83.98, + "learning_rate": 9.314749495650186e-05, + "loss": 2.9111, + "step": 41820 + }, + { + "epoch": 84.02, + "learning_rate": 9.312480023855836e-05, + "loss": 2.8821, + "step": 41840 + }, + { + "epoch": 84.06, + "learning_rate": 9.310207077574227e-05, + "loss": 2.8004, + "step": 41860 + }, + { + "epoch": 84.1, + "learning_rate": 9.307930658636626e-05, + "loss": 2.8312, + "step": 41880 + }, + { + "epoch": 84.14, + "learning_rate": 9.305650768877104e-05, + "loss": 2.8343, + "step": 41900 + }, + { + "epoch": 84.18, + "learning_rate": 9.303367410132526e-05, + "loss": 2.8554, + "step": 41920 + }, + { + "epoch": 84.22, + "learning_rate": 9.301080584242554e-05, + "loss": 2.8513, + "step": 41940 + }, + { + "epoch": 84.26, + "learning_rate": 9.298790293049641e-05, + "loss": 2.8587, + "step": 41960 + }, + { + "epoch": 84.3, + "learning_rate": 9.296496538399036e-05, + "loss": 2.8671, + "step": 41980 + }, + { + "epoch": 84.34, + "learning_rate": 9.294199322138773e-05, + "loss": 2.8853, + "step": 42000 + }, + { + "epoch": 84.34, + "eval_loss": 3.1971466541290283, + "eval_runtime": 34.2904, + "eval_samples_per_second": 22.339, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.003334910563650027, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0002901073397156948, + "step": 42000 + }, + { + "epoch": 84.38, + "learning_rate": 9.29189864611968e-05, + "loss": 2.8784, + "step": 42020 + }, + { + "epoch": 84.42, + "learning_rate": 9.289594512195368e-05, + "loss": 2.8568, + "step": 42040 + }, + { + "epoch": 84.46, + "learning_rate": 9.287286922222239e-05, + "loss": 2.8898, + "step": 42060 + }, + { + "epoch": 84.5, + "learning_rate": 9.28497587805947e-05, + "loss": 2.8704, + "step": 42080 + }, + { + "epoch": 84.54, + "learning_rate": 9.282661381569036e-05, + "loss": 2.8771, + "step": 42100 + }, + { + "epoch": 84.58, + "learning_rate": 9.280343434615679e-05, + "loss": 2.8842, + "step": 42120 + }, + { + "epoch": 84.62, + "learning_rate": 9.27802203906693e-05, + "loss": 2.8989, + "step": 42140 + }, + { + "epoch": 84.66, + "learning_rate": 9.275697196793094e-05, + "loss": 2.8898, + "step": 42160 + }, + { + "epoch": 84.7, + "learning_rate": 9.273368909667255e-05, + "loss": 2.8871, + "step": 42180 + }, + { + "epoch": 84.74, + "learning_rate": 9.271037179565272e-05, + "loss": 2.8987, + "step": 42200 + }, + { + "epoch": 84.78, + "learning_rate": 9.268702008365777e-05, + "loss": 2.8868, + "step": 42220 + }, + { + "epoch": 84.82, + "learning_rate": 9.266363397950178e-05, + "loss": 2.8867, + "step": 42240 + }, + { + "epoch": 84.86, + "learning_rate": 9.264021350202646e-05, + "loss": 2.9063, + "step": 42260 + }, + { + "epoch": 84.9, + "learning_rate": 9.26167586701013e-05, + "loss": 2.8965, + "step": 42280 + }, + { + "epoch": 84.94, + "learning_rate": 9.259326950262345e-05, + "loss": 2.909, + "step": 42300 + }, + { + "epoch": 84.98, + "learning_rate": 9.256974601851767e-05, + "loss": 2.9071, + "step": 42320 + }, + { + "epoch": 85.02, + "learning_rate": 9.254618823673644e-05, + "loss": 2.8382, + "step": 42340 + }, + { + "epoch": 85.06, + "learning_rate": 9.252259617625983e-05, + "loss": 2.7963, + "step": 42360 + }, + { + "epoch": 85.1, + "learning_rate": 9.249896985609551e-05, + "loss": 2.813, + "step": 42380 + }, + { + "epoch": 85.14, + "learning_rate": 9.247530929527881e-05, + "loss": 2.8221, + "step": 42400 + }, + { + "epoch": 85.18, + "learning_rate": 9.245161451287261e-05, + "loss": 2.8407, + "step": 42420 + }, + { + "epoch": 85.22, + "learning_rate": 9.242788552796735e-05, + "loss": 2.8443, + "step": 42440 + }, + { + "epoch": 85.26, + "learning_rate": 9.240412235968108e-05, + "loss": 2.8411, + "step": 42460 + }, + { + "epoch": 85.3, + "learning_rate": 9.238032502715931e-05, + "loss": 2.8383, + "step": 42480 + }, + { + "epoch": 85.34, + "learning_rate": 9.235649354957517e-05, + "loss": 2.8576, + "step": 42500 + }, + { + "epoch": 85.38, + "learning_rate": 9.233262794612923e-05, + "loss": 2.86, + "step": 42520 + }, + { + "epoch": 85.42, + "learning_rate": 9.230872823604956e-05, + "loss": 2.8618, + "step": 42540 + }, + { + "epoch": 85.46, + "learning_rate": 9.228479443859173e-05, + "loss": 2.8538, + "step": 42560 + }, + { + "epoch": 85.5, + "learning_rate": 9.226082657303879e-05, + "loss": 2.8592, + "step": 42580 + }, + { + "epoch": 85.54, + "learning_rate": 9.22368246587012e-05, + "loss": 2.8664, + "step": 42600 + }, + { + "epoch": 85.58, + "learning_rate": 9.221278871491688e-05, + "loss": 2.8777, + "step": 42620 + }, + { + "epoch": 85.62, + "learning_rate": 9.218871876105113e-05, + "loss": 2.8727, + "step": 42640 + }, + { + "epoch": 85.66, + "learning_rate": 9.21646148164967e-05, + "loss": 2.8686, + "step": 42660 + }, + { + "epoch": 85.7, + "learning_rate": 9.214047690067369e-05, + "loss": 2.8787, + "step": 42680 + }, + { + "epoch": 85.74, + "learning_rate": 9.21163050330296e-05, + "loss": 2.8827, + "step": 42700 + }, + { + "epoch": 85.78, + "learning_rate": 9.209209923303925e-05, + "loss": 2.8872, + "step": 42720 + }, + { + "epoch": 85.82, + "learning_rate": 9.206785952020482e-05, + "loss": 2.8845, + "step": 42740 + }, + { + "epoch": 85.86, + "learning_rate": 9.204358591405582e-05, + "loss": 2.8924, + "step": 42760 + }, + { + "epoch": 85.9, + "learning_rate": 9.201927843414904e-05, + "loss": 2.8866, + "step": 42780 + }, + { + "epoch": 85.94, + "learning_rate": 9.199493710006859e-05, + "loss": 2.8933, + "step": 42800 + }, + { + "epoch": 85.98, + "learning_rate": 9.197056193142583e-05, + "loss": 2.8773, + "step": 42820 + }, + { + "epoch": 86.02, + "learning_rate": 9.194615294785942e-05, + "loss": 2.8457, + "step": 42840 + }, + { + "epoch": 86.06, + "learning_rate": 9.192171016903521e-05, + "loss": 2.7884, + "step": 42860 + }, + { + "epoch": 86.1, + "learning_rate": 9.189723361464631e-05, + "loss": 2.7996, + "step": 42880 + }, + { + "epoch": 86.14, + "learning_rate": 9.187272330441303e-05, + "loss": 2.8105, + "step": 42900 + }, + { + "epoch": 86.18, + "learning_rate": 9.184817925808291e-05, + "loss": 2.8156, + "step": 42920 + }, + { + "epoch": 86.22, + "learning_rate": 9.182360149543063e-05, + "loss": 2.8231, + "step": 42940 + }, + { + "epoch": 86.27, + "learning_rate": 9.180022140920303e-05, + "loss": 2.8424, + "step": 42960 + }, + { + "epoch": 86.31, + "learning_rate": 9.177557795670238e-05, + "loss": 2.8124, + "step": 42980 + }, + { + "epoch": 86.35, + "learning_rate": 9.17509008463731e-05, + "loss": 2.8423, + "step": 43000 + }, + { + "epoch": 86.35, + "eval_loss": 3.1840007305145264, + "eval_runtime": 34.7666, + "eval_samples_per_second": 22.033, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.0037927957043078506, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.719756309834639e-05, + "step": 43000 + }, + { + "epoch": 86.39, + "learning_rate": 9.172619009809713e-05, + "loss": 2.836, + "step": 43020 + }, + { + "epoch": 86.43, + "learning_rate": 9.170144573178348e-05, + "loss": 2.8482, + "step": 43040 + }, + { + "epoch": 86.47, + "learning_rate": 9.16766677673682e-05, + "loss": 2.8524, + "step": 43060 + }, + { + "epoch": 86.51, + "learning_rate": 9.165185622481447e-05, + "loss": 2.842, + "step": 43080 + }, + { + "epoch": 86.55, + "learning_rate": 9.162701112411252e-05, + "loss": 2.8382, + "step": 43100 + }, + { + "epoch": 86.59, + "learning_rate": 9.160213248527955e-05, + "loss": 2.849, + "step": 43120 + }, + { + "epoch": 86.63, + "learning_rate": 9.15772203283599e-05, + "loss": 2.8786, + "step": 43140 + }, + { + "epoch": 86.67, + "learning_rate": 9.155227467342477e-05, + "loss": 2.8683, + "step": 43160 + }, + { + "epoch": 86.71, + "learning_rate": 9.152729554057249e-05, + "loss": 2.8468, + "step": 43180 + }, + { + "epoch": 86.75, + "learning_rate": 9.150228294992826e-05, + "loss": 2.8879, + "step": 43200 + }, + { + "epoch": 86.79, + "learning_rate": 9.147723692164427e-05, + "loss": 2.8745, + "step": 43220 + }, + { + "epoch": 86.83, + "learning_rate": 9.145215747589966e-05, + "loss": 2.8715, + "step": 43240 + }, + { + "epoch": 86.87, + "learning_rate": 9.142704463290049e-05, + "loss": 2.8719, + "step": 43260 + }, + { + "epoch": 86.91, + "learning_rate": 9.140189841287972e-05, + "loss": 2.8809, + "step": 43280 + }, + { + "epoch": 86.95, + "learning_rate": 9.137671883609721e-05, + "loss": 2.8833, + "step": 43300 + }, + { + "epoch": 86.99, + "learning_rate": 9.13515059228397e-05, + "loss": 2.8894, + "step": 43320 + }, + { + "epoch": 87.03, + "learning_rate": 9.132625969342075e-05, + "loss": 2.8038, + "step": 43340 + }, + { + "epoch": 87.07, + "learning_rate": 9.13009801681808e-05, + "loss": 2.7811, + "step": 43360 + }, + { + "epoch": 87.11, + "learning_rate": 9.127566736748714e-05, + "loss": 2.7949, + "step": 43380 + }, + { + "epoch": 87.15, + "learning_rate": 9.12503213117338e-05, + "loss": 2.8054, + "step": 43400 + }, + { + "epoch": 87.19, + "learning_rate": 9.122494202134162e-05, + "loss": 2.7804, + "step": 43420 + }, + { + "epoch": 87.23, + "learning_rate": 9.11995295167583e-05, + "loss": 2.8087, + "step": 43440 + }, + { + "epoch": 87.27, + "learning_rate": 9.117408381845817e-05, + "loss": 2.8007, + "step": 43460 + }, + { + "epoch": 87.31, + "learning_rate": 9.114860494694244e-05, + "loss": 2.8347, + "step": 43480 + }, + { + "epoch": 87.35, + "learning_rate": 9.112309292273891e-05, + "loss": 2.8312, + "step": 43500 + }, + { + "epoch": 87.39, + "learning_rate": 9.10975477664022e-05, + "loss": 2.8377, + "step": 43520 + }, + { + "epoch": 87.43, + "learning_rate": 9.107196949851356e-05, + "loss": 2.8334, + "step": 43540 + }, + { + "epoch": 87.47, + "learning_rate": 9.104635813968093e-05, + "loss": 2.8396, + "step": 43560 + }, + { + "epoch": 87.51, + "learning_rate": 9.102071371053896e-05, + "loss": 2.8407, + "step": 43580 + }, + { + "epoch": 87.55, + "learning_rate": 9.099503623174885e-05, + "loss": 2.8599, + "step": 43600 + }, + { + "epoch": 87.59, + "learning_rate": 9.096932572399852e-05, + "loss": 2.845, + "step": 43620 + }, + { + "epoch": 87.63, + "learning_rate": 9.094358220800243e-05, + "loss": 2.8397, + "step": 43640 + }, + { + "epoch": 87.67, + "learning_rate": 9.09178057045017e-05, + "loss": 2.8388, + "step": 43660 + }, + { + "epoch": 87.71, + "learning_rate": 9.089199623426398e-05, + "loss": 2.8566, + "step": 43680 + }, + { + "epoch": 87.75, + "learning_rate": 9.08661538180835e-05, + "loss": 2.854, + "step": 43700 + }, + { + "epoch": 87.79, + "learning_rate": 9.0840278476781e-05, + "loss": 2.846, + "step": 43720 + }, + { + "epoch": 87.83, + "learning_rate": 9.081437023120381e-05, + "loss": 2.8713, + "step": 43740 + }, + { + "epoch": 87.87, + "learning_rate": 9.078842910222573e-05, + "loss": 2.848, + "step": 43760 + }, + { + "epoch": 87.91, + "learning_rate": 9.076245511074703e-05, + "loss": 2.8607, + "step": 43780 + }, + { + "epoch": 87.95, + "learning_rate": 9.073644827769451e-05, + "loss": 2.8752, + "step": 43800 + }, + { + "epoch": 87.99, + "learning_rate": 9.071040862402141e-05, + "loss": 2.8715, + "step": 43820 + }, + { + "epoch": 88.03, + "learning_rate": 9.06843361707074e-05, + "loss": 2.7882, + "step": 43840 + }, + { + "epoch": 88.07, + "learning_rate": 9.065823093875858e-05, + "loss": 2.7729, + "step": 43860 + }, + { + "epoch": 88.11, + "learning_rate": 9.063209294920746e-05, + "loss": 2.7647, + "step": 43880 + }, + { + "epoch": 88.15, + "learning_rate": 9.060592222311296e-05, + "loss": 2.779, + "step": 43900 + }, + { + "epoch": 88.19, + "learning_rate": 9.057971878156036e-05, + "loss": 2.7953, + "step": 43920 + }, + { + "epoch": 88.23, + "learning_rate": 9.05534826456613e-05, + "loss": 2.8093, + "step": 43940 + }, + { + "epoch": 88.27, + "learning_rate": 9.052721383655374e-05, + "loss": 2.8082, + "step": 43960 + }, + { + "epoch": 88.31, + "learning_rate": 9.050091237540201e-05, + "loss": 2.7955, + "step": 43980 + }, + { + "epoch": 88.35, + "learning_rate": 9.047457828339672e-05, + "loss": 2.8217, + "step": 44000 + }, + { + "epoch": 88.35, + "eval_loss": 3.1697580814361572, + "eval_runtime": 32.5126, + "eval_samples_per_second": 23.56, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004088690310030291, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.799390774586597e-05, + "step": 44000 + }, + { + "epoch": 88.39, + "learning_rate": 9.044821158175476e-05, + "loss": 2.8138, + "step": 44020 + }, + { + "epoch": 88.43, + "learning_rate": 9.042181229171931e-05, + "loss": 2.8121, + "step": 44040 + }, + { + "epoch": 88.47, + "learning_rate": 9.03953804345598e-05, + "loss": 2.8212, + "step": 44060 + }, + { + "epoch": 88.51, + "learning_rate": 9.03689160315719e-05, + "loss": 2.8291, + "step": 44080 + }, + { + "epoch": 88.55, + "learning_rate": 9.03424191040775e-05, + "loss": 2.8411, + "step": 44100 + }, + { + "epoch": 88.59, + "learning_rate": 9.031588967342473e-05, + "loss": 2.8159, + "step": 44120 + }, + { + "epoch": 88.63, + "learning_rate": 9.028932776098781e-05, + "loss": 2.8491, + "step": 44140 + }, + { + "epoch": 88.67, + "learning_rate": 9.026273338816724e-05, + "loss": 2.8319, + "step": 44160 + }, + { + "epoch": 88.71, + "learning_rate": 9.02361065763896e-05, + "loss": 2.8401, + "step": 44180 + }, + { + "epoch": 88.76, + "learning_rate": 9.020944734710766e-05, + "loss": 2.8291, + "step": 44200 + }, + { + "epoch": 88.8, + "learning_rate": 9.018275572180024e-05, + "loss": 2.839, + "step": 44220 + }, + { + "epoch": 88.84, + "learning_rate": 9.015603172197233e-05, + "loss": 2.8578, + "step": 44240 + }, + { + "epoch": 88.88, + "learning_rate": 9.012927536915497e-05, + "loss": 2.8442, + "step": 44260 + }, + { + "epoch": 88.92, + "learning_rate": 9.010248668490528e-05, + "loss": 2.8596, + "step": 44280 + }, + { + "epoch": 88.96, + "learning_rate": 9.007566569080639e-05, + "loss": 2.8458, + "step": 44300 + }, + { + "epoch": 89.0, + "learning_rate": 9.004881240846751e-05, + "loss": 2.8605, + "step": 44320 + }, + { + "epoch": 89.04, + "learning_rate": 9.002192685952385e-05, + "loss": 2.7546, + "step": 44340 + }, + { + "epoch": 89.08, + "learning_rate": 8.999500906563658e-05, + "loss": 2.7547, + "step": 44360 + }, + { + "epoch": 89.12, + "learning_rate": 8.996805904849292e-05, + "loss": 2.7628, + "step": 44380 + }, + { + "epoch": 89.16, + "learning_rate": 8.9941076829806e-05, + "loss": 2.7622, + "step": 44400 + }, + { + "epoch": 89.2, + "learning_rate": 8.991406243131489e-05, + "loss": 2.7857, + "step": 44420 + }, + { + "epoch": 89.24, + "learning_rate": 8.98870158747846e-05, + "loss": 2.7764, + "step": 44440 + }, + { + "epoch": 89.28, + "learning_rate": 8.985993718200608e-05, + "loss": 2.782, + "step": 44460 + }, + { + "epoch": 89.32, + "learning_rate": 8.983282637479614e-05, + "loss": 2.7946, + "step": 44480 + }, + { + "epoch": 89.36, + "learning_rate": 8.980568347499744e-05, + "loss": 2.8034, + "step": 44500 + }, + { + "epoch": 89.4, + "learning_rate": 8.977850850447855e-05, + "loss": 2.7923, + "step": 44520 + }, + { + "epoch": 89.44, + "learning_rate": 8.975130148513384e-05, + "loss": 2.8333, + "step": 44540 + }, + { + "epoch": 89.48, + "learning_rate": 8.972406243888351e-05, + "loss": 2.8207, + "step": 44560 + }, + { + "epoch": 89.52, + "learning_rate": 8.969679138767359e-05, + "loss": 2.8094, + "step": 44580 + }, + { + "epoch": 89.56, + "learning_rate": 8.966948835347586e-05, + "loss": 2.8085, + "step": 44600 + }, + { + "epoch": 89.6, + "learning_rate": 8.964215335828787e-05, + "loss": 2.824, + "step": 44620 + }, + { + "epoch": 89.64, + "learning_rate": 8.961478642413294e-05, + "loss": 2.8275, + "step": 44640 + }, + { + "epoch": 89.68, + "learning_rate": 8.958738757306013e-05, + "loss": 2.8035, + "step": 44660 + }, + { + "epoch": 89.72, + "learning_rate": 8.955995682714417e-05, + "loss": 2.8249, + "step": 44680 + }, + { + "epoch": 89.76, + "learning_rate": 8.953249420848555e-05, + "loss": 2.8245, + "step": 44700 + }, + { + "epoch": 89.8, + "learning_rate": 8.950499973921038e-05, + "loss": 2.8207, + "step": 44720 + }, + { + "epoch": 89.84, + "learning_rate": 8.947747344147049e-05, + "loss": 2.8302, + "step": 44740 + }, + { + "epoch": 89.88, + "learning_rate": 8.944991533744327e-05, + "loss": 2.8478, + "step": 44760 + }, + { + "epoch": 89.92, + "learning_rate": 8.942232544933182e-05, + "loss": 2.855, + "step": 44780 + }, + { + "epoch": 89.96, + "learning_rate": 8.93947037993648e-05, + "loss": 2.8628, + "step": 44800 + }, + { + "epoch": 90.0, + "learning_rate": 8.93670504097965e-05, + "loss": 2.846, + "step": 44820 + }, + { + "epoch": 90.04, + "learning_rate": 8.933936530290672e-05, + "loss": 2.7397, + "step": 44840 + }, + { + "epoch": 90.08, + "learning_rate": 8.931164850100086e-05, + "loss": 2.7602, + "step": 44860 + }, + { + "epoch": 90.12, + "learning_rate": 8.928390002640989e-05, + "loss": 2.7513, + "step": 44880 + }, + { + "epoch": 90.16, + "learning_rate": 8.925611990149021e-05, + "loss": 2.7634, + "step": 44900 + }, + { + "epoch": 90.2, + "learning_rate": 8.922830814862377e-05, + "loss": 2.7759, + "step": 44920 + }, + { + "epoch": 90.24, + "learning_rate": 8.920046479021799e-05, + "loss": 2.7657, + "step": 44940 + }, + { + "epoch": 90.28, + "learning_rate": 8.91725898487058e-05, + "loss": 2.7819, + "step": 44960 + }, + { + "epoch": 90.32, + "learning_rate": 8.91446833465455e-05, + "loss": 2.7722, + "step": 44980 + }, + { + "epoch": 90.36, + "learning_rate": 8.911674530622088e-05, + "loss": 2.7826, + "step": 45000 + }, + { + "epoch": 90.36, + "eval_loss": 3.151148557662964, + "eval_runtime": 32.5176, + "eval_samples_per_second": 23.556, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0036359128498092146, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 2.2664635915288657e-05, + "step": 45000 + }, + { + "epoch": 90.4, + "learning_rate": 8.908877575024107e-05, + "loss": 2.7854, + "step": 45020 + }, + { + "epoch": 90.44, + "learning_rate": 8.906077470114069e-05, + "loss": 2.7936, + "step": 45040 + }, + { + "epoch": 90.48, + "learning_rate": 8.903274218147963e-05, + "loss": 2.7995, + "step": 45060 + }, + { + "epoch": 90.52, + "learning_rate": 8.900467821384324e-05, + "loss": 2.811, + "step": 45080 + }, + { + "epoch": 90.56, + "learning_rate": 8.897658282084211e-05, + "loss": 2.793, + "step": 45100 + }, + { + "epoch": 90.6, + "learning_rate": 8.894845602511219e-05, + "loss": 2.8101, + "step": 45120 + }, + { + "epoch": 90.64, + "learning_rate": 8.892029784931476e-05, + "loss": 2.8208, + "step": 45140 + }, + { + "epoch": 90.68, + "learning_rate": 8.889210831613634e-05, + "loss": 2.7957, + "step": 45160 + }, + { + "epoch": 90.72, + "learning_rate": 8.886388744828872e-05, + "loss": 2.8232, + "step": 45180 + }, + { + "epoch": 90.76, + "learning_rate": 8.883563526850894e-05, + "loss": 2.8242, + "step": 45200 + }, + { + "epoch": 90.8, + "learning_rate": 8.88073517995593e-05, + "loss": 2.8364, + "step": 45220 + }, + { + "epoch": 90.84, + "learning_rate": 8.877903706422724e-05, + "loss": 2.8216, + "step": 45240 + }, + { + "epoch": 90.88, + "learning_rate": 8.875069108532546e-05, + "loss": 2.8087, + "step": 45260 + }, + { + "epoch": 90.92, + "learning_rate": 8.87223138856918e-05, + "loss": 2.8371, + "step": 45280 + }, + { + "epoch": 90.96, + "learning_rate": 8.869390548818923e-05, + "loss": 2.8169, + "step": 45300 + }, + { + "epoch": 91.0, + "learning_rate": 8.866546591570592e-05, + "loss": 2.8234, + "step": 45320 + }, + { + "epoch": 91.04, + "learning_rate": 8.863699519115511e-05, + "loss": 2.7411, + "step": 45340 + }, + { + "epoch": 91.08, + "learning_rate": 8.860849333747514e-05, + "loss": 2.7439, + "step": 45360 + }, + { + "epoch": 91.12, + "learning_rate": 8.857996037762944e-05, + "loss": 2.7479, + "step": 45380 + }, + { + "epoch": 91.16, + "learning_rate": 8.855139633460649e-05, + "loss": 2.7497, + "step": 45400 + }, + { + "epoch": 91.2, + "learning_rate": 8.852280123141984e-05, + "loss": 2.758, + "step": 45420 + }, + { + "epoch": 91.24, + "learning_rate": 8.849417509110805e-05, + "loss": 2.7595, + "step": 45440 + }, + { + "epoch": 91.29, + "learning_rate": 8.846551793673467e-05, + "loss": 2.7707, + "step": 45460 + }, + { + "epoch": 91.33, + "learning_rate": 8.843682979138825e-05, + "loss": 2.7653, + "step": 45480 + }, + { + "epoch": 91.37, + "learning_rate": 8.840811067818233e-05, + "loss": 2.767, + "step": 45500 + }, + { + "epoch": 91.41, + "learning_rate": 8.837936062025538e-05, + "loss": 2.7821, + "step": 45520 + }, + { + "epoch": 91.45, + "learning_rate": 8.835057964077079e-05, + "loss": 2.7701, + "step": 45540 + }, + { + "epoch": 91.49, + "learning_rate": 8.832176776291688e-05, + "loss": 2.7874, + "step": 45560 + }, + { + "epoch": 91.53, + "learning_rate": 8.829292500990683e-05, + "loss": 2.7821, + "step": 45580 + }, + { + "epoch": 91.57, + "learning_rate": 8.826405140497878e-05, + "loss": 2.785, + "step": 45600 + }, + { + "epoch": 91.61, + "learning_rate": 8.823514697139564e-05, + "loss": 2.7862, + "step": 45620 + }, + { + "epoch": 91.65, + "learning_rate": 8.820621173244519e-05, + "loss": 2.8041, + "step": 45640 + }, + { + "epoch": 91.69, + "learning_rate": 8.817724571144004e-05, + "loss": 2.8083, + "step": 45660 + }, + { + "epoch": 91.73, + "learning_rate": 8.814824893171758e-05, + "loss": 2.7926, + "step": 45680 + }, + { + "epoch": 91.77, + "learning_rate": 8.811922141664e-05, + "loss": 2.8057, + "step": 45700 + }, + { + "epoch": 91.81, + "learning_rate": 8.809016318959424e-05, + "loss": 2.8142, + "step": 45720 + }, + { + "epoch": 91.85, + "learning_rate": 8.806107427399197e-05, + "loss": 2.8213, + "step": 45740 + }, + { + "epoch": 91.89, + "learning_rate": 8.803195469326964e-05, + "loss": 2.8062, + "step": 45760 + }, + { + "epoch": 91.93, + "learning_rate": 8.800280447088836e-05, + "loss": 2.8231, + "step": 45780 + }, + { + "epoch": 91.97, + "learning_rate": 8.797362363033392e-05, + "loss": 2.8138, + "step": 45800 + }, + { + "epoch": 92.01, + "learning_rate": 8.794441219511681e-05, + "loss": 2.7902, + "step": 45820 + }, + { + "epoch": 92.05, + "learning_rate": 8.791517018877216e-05, + "loss": 2.7038, + "step": 45840 + }, + { + "epoch": 92.09, + "learning_rate": 8.78858976348597e-05, + "loss": 2.7284, + "step": 45860 + }, + { + "epoch": 92.13, + "learning_rate": 8.785659455696384e-05, + "loss": 2.7328, + "step": 45880 + }, + { + "epoch": 92.17, + "learning_rate": 8.782726097869349e-05, + "loss": 2.7271, + "step": 45900 + }, + { + "epoch": 92.21, + "learning_rate": 8.779789692368223e-05, + "loss": 2.7463, + "step": 45920 + }, + { + "epoch": 92.25, + "learning_rate": 8.776850241558814e-05, + "loss": 2.7765, + "step": 45940 + }, + { + "epoch": 92.29, + "learning_rate": 8.773907747809383e-05, + "loss": 2.7623, + "step": 45960 + }, + { + "epoch": 92.33, + "learning_rate": 8.770962213490643e-05, + "loss": 2.7572, + "step": 45980 + }, + { + "epoch": 92.37, + "learning_rate": 8.768013640975761e-05, + "loss": 2.7663, + "step": 46000 + }, + { + "epoch": 92.37, + "eval_loss": 3.1335716247558594, + "eval_runtime": 34.2186, + "eval_samples_per_second": 22.386, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.0033108536023324113, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00021304757760371338, + "step": 46000 + }, + { + "epoch": 92.41, + "learning_rate": 8.765062032640346e-05, + "loss": 2.765, + "step": 46020 + }, + { + "epoch": 92.45, + "learning_rate": 8.762107390862455e-05, + "loss": 2.7924, + "step": 46040 + }, + { + "epoch": 92.49, + "learning_rate": 8.759149718022594e-05, + "loss": 2.7628, + "step": 46060 + }, + { + "epoch": 92.53, + "learning_rate": 8.756189016503702e-05, + "loss": 2.7737, + "step": 46080 + }, + { + "epoch": 92.57, + "learning_rate": 8.753225288691165e-05, + "loss": 2.7923, + "step": 46100 + }, + { + "epoch": 92.61, + "learning_rate": 8.750258536972804e-05, + "loss": 2.7757, + "step": 46120 + }, + { + "epoch": 92.65, + "learning_rate": 8.747288763738877e-05, + "loss": 2.793, + "step": 46140 + }, + { + "epoch": 92.69, + "learning_rate": 8.744315971382078e-05, + "loss": 2.8009, + "step": 46160 + }, + { + "epoch": 92.73, + "learning_rate": 8.741340162297531e-05, + "loss": 2.7896, + "step": 46180 + }, + { + "epoch": 92.77, + "learning_rate": 8.738361338882792e-05, + "loss": 2.7783, + "step": 46200 + }, + { + "epoch": 92.81, + "learning_rate": 8.735379503537844e-05, + "loss": 2.8142, + "step": 46220 + }, + { + "epoch": 92.85, + "learning_rate": 8.732394658665101e-05, + "loss": 2.8033, + "step": 46240 + }, + { + "epoch": 92.89, + "learning_rate": 8.729406806669396e-05, + "loss": 2.8038, + "step": 46260 + }, + { + "epoch": 92.93, + "learning_rate": 8.726415949957987e-05, + "loss": 2.798, + "step": 46280 + }, + { + "epoch": 92.97, + "learning_rate": 8.723422090940555e-05, + "loss": 2.8052, + "step": 46300 + }, + { + "epoch": 93.01, + "learning_rate": 8.720425232029198e-05, + "loss": 2.7697, + "step": 46320 + }, + { + "epoch": 93.05, + "learning_rate": 8.717425375638429e-05, + "loss": 2.7166, + "step": 46340 + }, + { + "epoch": 93.09, + "learning_rate": 8.714422524185181e-05, + "loss": 2.7142, + "step": 46360 + }, + { + "epoch": 93.13, + "learning_rate": 8.711416680088795e-05, + "loss": 2.7344, + "step": 46380 + }, + { + "epoch": 93.17, + "learning_rate": 8.708407845771027e-05, + "loss": 2.725, + "step": 46400 + }, + { + "epoch": 93.21, + "learning_rate": 8.70539602365604e-05, + "loss": 2.7226, + "step": 46420 + }, + { + "epoch": 93.25, + "learning_rate": 8.702381216170403e-05, + "loss": 2.7398, + "step": 46440 + }, + { + "epoch": 93.29, + "learning_rate": 8.699363425743093e-05, + "loss": 2.7418, + "step": 46460 + }, + { + "epoch": 93.33, + "learning_rate": 8.69634265480549e-05, + "loss": 2.7336, + "step": 46480 + }, + { + "epoch": 93.37, + "learning_rate": 8.693318905791375e-05, + "loss": 2.753, + "step": 46500 + }, + { + "epoch": 93.41, + "learning_rate": 8.690292181136924e-05, + "loss": 2.7499, + "step": 46520 + }, + { + "epoch": 93.45, + "learning_rate": 8.687262483280719e-05, + "loss": 2.7674, + "step": 46540 + }, + { + "epoch": 93.49, + "learning_rate": 8.684229814663731e-05, + "loss": 2.767, + "step": 46560 + }, + { + "epoch": 93.53, + "learning_rate": 8.681194177729328e-05, + "loss": 2.7813, + "step": 46580 + }, + { + "epoch": 93.57, + "learning_rate": 8.678155574923265e-05, + "loss": 2.7579, + "step": 46600 + }, + { + "epoch": 93.61, + "learning_rate": 8.675114008693689e-05, + "loss": 2.7643, + "step": 46620 + }, + { + "epoch": 93.65, + "learning_rate": 8.672069481491141e-05, + "loss": 2.765, + "step": 46640 + }, + { + "epoch": 93.69, + "learning_rate": 8.669021995768534e-05, + "loss": 2.7782, + "step": 46660 + }, + { + "epoch": 93.73, + "learning_rate": 8.665971553981175e-05, + "loss": 2.7661, + "step": 46680 + }, + { + "epoch": 93.78, + "learning_rate": 8.662918158586753e-05, + "loss": 2.7837, + "step": 46700 + }, + { + "epoch": 93.82, + "learning_rate": 8.65986181204533e-05, + "loss": 2.8053, + "step": 46720 + }, + { + "epoch": 93.86, + "learning_rate": 8.656802516819349e-05, + "loss": 2.7843, + "step": 46740 + }, + { + "epoch": 93.9, + "learning_rate": 8.653740275373631e-05, + "loss": 2.8004, + "step": 46760 + }, + { + "epoch": 93.94, + "learning_rate": 8.650675090175366e-05, + "loss": 2.7905, + "step": 46780 + }, + { + "epoch": 93.98, + "learning_rate": 8.647606963694122e-05, + "loss": 2.8014, + "step": 46800 + }, + { + "epoch": 94.02, + "learning_rate": 8.644535898401831e-05, + "loss": 2.7384, + "step": 46820 + }, + { + "epoch": 94.06, + "learning_rate": 8.641461896772793e-05, + "loss": 2.7017, + "step": 46840 + }, + { + "epoch": 94.1, + "learning_rate": 8.638384961283679e-05, + "loss": 2.7107, + "step": 46860 + }, + { + "epoch": 94.14, + "learning_rate": 8.63530509441352e-05, + "loss": 2.6978, + "step": 46880 + }, + { + "epoch": 94.18, + "learning_rate": 8.632222298643706e-05, + "loss": 2.7008, + "step": 46900 + }, + { + "epoch": 94.22, + "learning_rate": 8.629136576457991e-05, + "loss": 2.7217, + "step": 46920 + }, + { + "epoch": 94.26, + "learning_rate": 8.626047930342488e-05, + "loss": 2.7346, + "step": 46940 + }, + { + "epoch": 94.3, + "learning_rate": 8.622956362785662e-05, + "loss": 2.7401, + "step": 46960 + }, + { + "epoch": 94.34, + "learning_rate": 8.619861876278332e-05, + "loss": 2.7394, + "step": 46980 + }, + { + "epoch": 94.38, + "learning_rate": 8.616764473313671e-05, + "loss": 2.7397, + "step": 47000 + }, + { + "epoch": 94.38, + "eval_loss": 3.1263296604156494, + "eval_runtime": 34.2789, + "eval_samples_per_second": 22.346, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.003961882572332823, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 1.5865245140702057e-05, + "step": 47000 + }, + { + "epoch": 94.42, + "learning_rate": 8.6136641563872e-05, + "loss": 2.737, + "step": 47020 + }, + { + "epoch": 94.46, + "learning_rate": 8.610716158524992e-05, + "loss": 2.7411, + "step": 47040 + }, + { + "epoch": 94.5, + "learning_rate": 8.607610166559625e-05, + "loss": 2.7648, + "step": 47060 + }, + { + "epoch": 94.54, + "learning_rate": 8.60450126800791e-05, + "loss": 2.7632, + "step": 47080 + }, + { + "epoch": 94.58, + "learning_rate": 8.601389465374627e-05, + "loss": 2.7616, + "step": 47100 + }, + { + "epoch": 94.62, + "learning_rate": 8.598274761166903e-05, + "loss": 2.7732, + "step": 47120 + }, + { + "epoch": 94.66, + "learning_rate": 8.595157157894194e-05, + "loss": 2.7699, + "step": 47140 + }, + { + "epoch": 94.7, + "learning_rate": 8.592036658068301e-05, + "loss": 2.7514, + "step": 47160 + }, + { + "epoch": 94.74, + "learning_rate": 8.588913264203351e-05, + "loss": 2.7769, + "step": 47180 + }, + { + "epoch": 94.78, + "learning_rate": 8.585786978815808e-05, + "loss": 2.7746, + "step": 47200 + }, + { + "epoch": 94.82, + "learning_rate": 8.582657804424465e-05, + "loss": 2.771, + "step": 47220 + }, + { + "epoch": 94.86, + "learning_rate": 8.579525743550438e-05, + "loss": 2.7709, + "step": 47240 + }, + { + "epoch": 94.9, + "learning_rate": 8.576390798717174e-05, + "loss": 2.7646, + "step": 47260 + }, + { + "epoch": 94.94, + "learning_rate": 8.57325297245044e-05, + "loss": 2.7823, + "step": 47280 + }, + { + "epoch": 94.98, + "learning_rate": 8.570112267278328e-05, + "loss": 2.8006, + "step": 47300 + }, + { + "epoch": 95.02, + "learning_rate": 8.566968685731245e-05, + "loss": 2.7315, + "step": 47320 + }, + { + "epoch": 95.06, + "learning_rate": 8.563822230341919e-05, + "loss": 2.6939, + "step": 47340 + }, + { + "epoch": 95.1, + "learning_rate": 8.560672903645392e-05, + "loss": 2.6784, + "step": 47360 + }, + { + "epoch": 95.14, + "learning_rate": 8.557520708179021e-05, + "loss": 2.7214, + "step": 47380 + }, + { + "epoch": 95.18, + "learning_rate": 8.554365646482472e-05, + "loss": 2.7121, + "step": 47400 + }, + { + "epoch": 95.22, + "learning_rate": 8.551207721097721e-05, + "loss": 2.7172, + "step": 47420 + }, + { + "epoch": 95.26, + "learning_rate": 8.548046934569051e-05, + "loss": 2.7139, + "step": 47440 + }, + { + "epoch": 95.3, + "learning_rate": 8.544883289443053e-05, + "loss": 2.7152, + "step": 47460 + }, + { + "epoch": 95.34, + "learning_rate": 8.541716788268617e-05, + "loss": 2.7334, + "step": 47480 + }, + { + "epoch": 95.38, + "learning_rate": 8.538547433596933e-05, + "loss": 2.756, + "step": 47500 + }, + { + "epoch": 95.42, + "learning_rate": 8.535375227981497e-05, + "loss": 2.7178, + "step": 47520 + }, + { + "epoch": 95.46, + "learning_rate": 8.532200173978097e-05, + "loss": 2.7398, + "step": 47540 + }, + { + "epoch": 95.5, + "learning_rate": 8.529022274144816e-05, + "loss": 2.7445, + "step": 47560 + }, + { + "epoch": 95.54, + "learning_rate": 8.525841531042031e-05, + "loss": 2.7412, + "step": 47580 + }, + { + "epoch": 95.58, + "learning_rate": 8.522657947232407e-05, + "loss": 2.7432, + "step": 47600 + }, + { + "epoch": 95.62, + "learning_rate": 8.519471525280903e-05, + "loss": 2.7324, + "step": 47620 + }, + { + "epoch": 95.66, + "learning_rate": 8.516282267754761e-05, + "loss": 2.7537, + "step": 47640 + }, + { + "epoch": 95.7, + "learning_rate": 8.513090177223506e-05, + "loss": 2.7551, + "step": 47660 + }, + { + "epoch": 95.74, + "learning_rate": 8.509895256258948e-05, + "loss": 2.7591, + "step": 47680 + }, + { + "epoch": 95.78, + "learning_rate": 8.506697507435182e-05, + "loss": 2.7758, + "step": 47700 + }, + { + "epoch": 95.82, + "learning_rate": 8.50349693332857e-05, + "loss": 2.7382, + "step": 47720 + }, + { + "epoch": 95.86, + "learning_rate": 8.50029353651776e-05, + "loss": 2.7769, + "step": 47740 + }, + { + "epoch": 95.9, + "learning_rate": 8.497087319583672e-05, + "loss": 2.7683, + "step": 47760 + }, + { + "epoch": 95.94, + "learning_rate": 8.493878285109495e-05, + "loss": 2.7681, + "step": 47780 + }, + { + "epoch": 95.98, + "learning_rate": 8.49066643568069e-05, + "loss": 2.7677, + "step": 47800 + }, + { + "epoch": 96.02, + "learning_rate": 8.487451773884987e-05, + "loss": 2.725, + "step": 47820 + }, + { + "epoch": 96.06, + "learning_rate": 8.484234302312382e-05, + "loss": 2.69, + "step": 47840 + }, + { + "epoch": 96.1, + "learning_rate": 8.48101402355513e-05, + "loss": 2.6816, + "step": 47860 + }, + { + "epoch": 96.14, + "learning_rate": 8.477790940207756e-05, + "loss": 2.6851, + "step": 47880 + }, + { + "epoch": 96.18, + "learning_rate": 8.474565054867037e-05, + "loss": 2.6931, + "step": 47900 + }, + { + "epoch": 96.22, + "learning_rate": 8.471336370132012e-05, + "loss": 2.6884, + "step": 47920 + }, + { + "epoch": 96.27, + "learning_rate": 8.468104888603973e-05, + "loss": 2.71, + "step": 47940 + }, + { + "epoch": 96.31, + "learning_rate": 8.464870612886467e-05, + "loss": 2.7119, + "step": 47960 + }, + { + "epoch": 96.35, + "learning_rate": 8.46163354558529e-05, + "loss": 2.7139, + "step": 47980 + }, + { + "epoch": 96.39, + "learning_rate": 8.458393689308491e-05, + "loss": 2.7128, + "step": 48000 + }, + { + "epoch": 96.39, + "eval_loss": 3.1050965785980225, + "eval_runtime": 32.3765, + "eval_samples_per_second": 23.659, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.004855416596791393, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0003513791687304913, + "step": 48000 + }, + { + "epoch": 96.43, + "learning_rate": 8.455151046666363e-05, + "loss": 2.7272, + "step": 48020 + }, + { + "epoch": 96.47, + "learning_rate": 8.451905620271443e-05, + "loss": 2.7292, + "step": 48040 + }, + { + "epoch": 96.51, + "learning_rate": 8.448657412738515e-05, + "loss": 2.7327, + "step": 48060 + }, + { + "epoch": 96.55, + "learning_rate": 8.445406426684598e-05, + "loss": 2.7497, + "step": 48080 + }, + { + "epoch": 96.59, + "learning_rate": 8.442152664728958e-05, + "loss": 2.7236, + "step": 48100 + }, + { + "epoch": 96.63, + "learning_rate": 8.438896129493086e-05, + "loss": 2.7378, + "step": 48120 + }, + { + "epoch": 96.67, + "learning_rate": 8.435636823600716e-05, + "loss": 2.7424, + "step": 48140 + }, + { + "epoch": 96.71, + "learning_rate": 8.432374749677814e-05, + "loss": 2.7365, + "step": 48160 + }, + { + "epoch": 96.75, + "learning_rate": 8.42910991035257e-05, + "loss": 2.7649, + "step": 48180 + }, + { + "epoch": 96.79, + "learning_rate": 8.425842308255412e-05, + "loss": 2.7542, + "step": 48200 + }, + { + "epoch": 96.83, + "learning_rate": 8.422571946018982e-05, + "loss": 2.7557, + "step": 48220 + }, + { + "epoch": 96.87, + "learning_rate": 8.419298826278154e-05, + "loss": 2.7459, + "step": 48240 + }, + { + "epoch": 96.91, + "learning_rate": 8.416022951670022e-05, + "loss": 2.7498, + "step": 48260 + }, + { + "epoch": 96.95, + "learning_rate": 8.412744324833898e-05, + "loss": 2.7668, + "step": 48280 + }, + { + "epoch": 96.99, + "learning_rate": 8.409462948411315e-05, + "loss": 2.7652, + "step": 48300 + }, + { + "epoch": 97.03, + "learning_rate": 8.406178825046015e-05, + "loss": 2.6939, + "step": 48320 + }, + { + "epoch": 97.07, + "learning_rate": 8.402891957383959e-05, + "loss": 2.6769, + "step": 48340 + }, + { + "epoch": 97.11, + "learning_rate": 8.399602348073316e-05, + "loss": 2.6848, + "step": 48360 + }, + { + "epoch": 97.15, + "learning_rate": 8.396309999764467e-05, + "loss": 2.6706, + "step": 48380 + }, + { + "epoch": 97.19, + "learning_rate": 8.393014915109995e-05, + "loss": 2.6847, + "step": 48400 + }, + { + "epoch": 97.23, + "learning_rate": 8.389717096764691e-05, + "loss": 2.6892, + "step": 48420 + }, + { + "epoch": 97.27, + "learning_rate": 8.386416547385547e-05, + "loss": 2.6884, + "step": 48440 + }, + { + "epoch": 97.31, + "learning_rate": 8.383113269631757e-05, + "loss": 2.7204, + "step": 48460 + }, + { + "epoch": 97.35, + "learning_rate": 8.379807266164714e-05, + "loss": 2.7159, + "step": 48480 + }, + { + "epoch": 97.39, + "learning_rate": 8.376498539648001e-05, + "loss": 2.7202, + "step": 48500 + }, + { + "epoch": 97.43, + "learning_rate": 8.373187092747403e-05, + "loss": 2.7029, + "step": 48520 + }, + { + "epoch": 97.47, + "learning_rate": 8.369872928130891e-05, + "loss": 2.7289, + "step": 48540 + }, + { + "epoch": 97.51, + "learning_rate": 8.366556048468628e-05, + "loss": 2.7209, + "step": 48560 + }, + { + "epoch": 97.55, + "learning_rate": 8.363236456432964e-05, + "loss": 2.7224, + "step": 48580 + }, + { + "epoch": 97.59, + "learning_rate": 8.359914154698434e-05, + "loss": 2.7278, + "step": 48600 + }, + { + "epoch": 97.63, + "learning_rate": 8.356589145941757e-05, + "loss": 2.7332, + "step": 48620 + }, + { + "epoch": 97.67, + "learning_rate": 8.353261432841832e-05, + "loss": 2.7295, + "step": 48640 + }, + { + "epoch": 97.71, + "learning_rate": 8.34993101807974e-05, + "loss": 2.7137, + "step": 48660 + }, + { + "epoch": 97.75, + "learning_rate": 8.346597904338731e-05, + "loss": 2.7288, + "step": 48680 + }, + { + "epoch": 97.79, + "learning_rate": 8.343262094304238e-05, + "loss": 2.7512, + "step": 48700 + }, + { + "epoch": 97.83, + "learning_rate": 8.339923590663863e-05, + "loss": 2.7368, + "step": 48720 + }, + { + "epoch": 97.87, + "learning_rate": 8.336582396107378e-05, + "loss": 2.7528, + "step": 48740 + }, + { + "epoch": 97.91, + "learning_rate": 8.33323851332672e-05, + "loss": 2.7398, + "step": 48760 + }, + { + "epoch": 97.95, + "learning_rate": 8.329891945015998e-05, + "loss": 2.7404, + "step": 48780 + }, + { + "epoch": 97.99, + "learning_rate": 8.326542693871482e-05, + "loss": 2.7475, + "step": 48800 + }, + { + "epoch": 98.03, + "learning_rate": 8.323190762591601e-05, + "loss": 2.6761, + "step": 48820 + }, + { + "epoch": 98.07, + "learning_rate": 8.319836153876947e-05, + "loss": 2.6581, + "step": 48840 + }, + { + "epoch": 98.11, + "learning_rate": 8.316478870430269e-05, + "loss": 2.6823, + "step": 48860 + }, + { + "epoch": 98.15, + "learning_rate": 8.313118914956466e-05, + "loss": 2.6526, + "step": 48880 + }, + { + "epoch": 98.19, + "learning_rate": 8.309756290162595e-05, + "loss": 2.6744, + "step": 48900 + }, + { + "epoch": 98.23, + "learning_rate": 8.306390998757863e-05, + "loss": 2.6912, + "step": 48920 + }, + { + "epoch": 98.27, + "learning_rate": 8.303023043453624e-05, + "loss": 2.7027, + "step": 48940 + }, + { + "epoch": 98.31, + "learning_rate": 8.299652426963379e-05, + "loss": 2.6867, + "step": 48960 + }, + { + "epoch": 98.35, + "learning_rate": 8.296279152002771e-05, + "loss": 2.6898, + "step": 48980 + }, + { + "epoch": 98.39, + "learning_rate": 8.29290322128959e-05, + "loss": 2.7087, + "step": 49000 + }, + { + "epoch": 98.39, + "eval_loss": 3.093707323074341, + "eval_runtime": 32.5378, + "eval_samples_per_second": 23.542, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004440345970997653, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 9.519147084421235e-05, + "step": 49000 + }, + { + "epoch": 98.43, + "learning_rate": 8.28952463754376e-05, + "loss": 2.7053, + "step": 49020 + }, + { + "epoch": 98.47, + "learning_rate": 8.286312528092997e-05, + "loss": 2.7229, + "step": 49040 + }, + { + "epoch": 98.51, + "learning_rate": 8.282928778764783e-05, + "loss": 2.7235, + "step": 49060 + }, + { + "epoch": 98.55, + "learning_rate": 8.27954238444015e-05, + "loss": 2.7244, + "step": 49080 + }, + { + "epoch": 98.59, + "learning_rate": 8.276153347847452e-05, + "loss": 2.7025, + "step": 49100 + }, + { + "epoch": 98.63, + "learning_rate": 8.272761671717178e-05, + "loss": 2.7124, + "step": 49120 + }, + { + "epoch": 98.67, + "learning_rate": 8.269367358781937e-05, + "loss": 2.7185, + "step": 49140 + }, + { + "epoch": 98.71, + "learning_rate": 8.265970411776467e-05, + "loss": 2.7199, + "step": 49160 + }, + { + "epoch": 98.76, + "learning_rate": 8.262570833437626e-05, + "loss": 2.7375, + "step": 49180 + }, + { + "epoch": 98.8, + "learning_rate": 8.259168626504395e-05, + "loss": 2.7188, + "step": 49200 + }, + { + "epoch": 98.84, + "learning_rate": 8.255763793717868e-05, + "loss": 2.7194, + "step": 49220 + }, + { + "epoch": 98.88, + "learning_rate": 8.25235633782126e-05, + "loss": 2.7347, + "step": 49240 + }, + { + "epoch": 98.92, + "learning_rate": 8.248946261559893e-05, + "loss": 2.7359, + "step": 49260 + }, + { + "epoch": 98.96, + "learning_rate": 8.245533567681208e-05, + "loss": 2.7177, + "step": 49280 + }, + { + "epoch": 99.0, + "learning_rate": 8.24211825893475e-05, + "loss": 2.736, + "step": 49300 + }, + { + "epoch": 99.04, + "learning_rate": 8.238700338072167e-05, + "loss": 2.66, + "step": 49320 + }, + { + "epoch": 99.08, + "learning_rate": 8.235279807847223e-05, + "loss": 2.6525, + "step": 49340 + }, + { + "epoch": 99.12, + "learning_rate": 8.231856671015772e-05, + "loss": 2.6511, + "step": 49360 + }, + { + "epoch": 99.16, + "learning_rate": 8.228430930335775e-05, + "loss": 2.6553, + "step": 49380 + }, + { + "epoch": 99.2, + "learning_rate": 8.225002588567291e-05, + "loss": 2.6576, + "step": 49400 + }, + { + "epoch": 99.24, + "learning_rate": 8.221571648472472e-05, + "loss": 2.6755, + "step": 49420 + }, + { + "epoch": 99.28, + "learning_rate": 8.218138112815564e-05, + "loss": 2.6818, + "step": 49440 + }, + { + "epoch": 99.32, + "learning_rate": 8.214701984362906e-05, + "loss": 2.6823, + "step": 49460 + }, + { + "epoch": 99.36, + "learning_rate": 8.211263265882923e-05, + "loss": 2.6776, + "step": 49480 + }, + { + "epoch": 99.4, + "learning_rate": 8.207821960146128e-05, + "loss": 2.6877, + "step": 49500 + }, + { + "epoch": 99.44, + "learning_rate": 8.20437806992512e-05, + "loss": 2.6976, + "step": 49520 + }, + { + "epoch": 99.48, + "learning_rate": 8.200931597994582e-05, + "loss": 2.6912, + "step": 49540 + }, + { + "epoch": 99.52, + "learning_rate": 8.197482547131268e-05, + "loss": 2.7145, + "step": 49560 + }, + { + "epoch": 99.56, + "learning_rate": 8.194030920114021e-05, + "loss": 2.7181, + "step": 49580 + }, + { + "epoch": 99.6, + "learning_rate": 8.190576719723752e-05, + "loss": 2.7224, + "step": 49600 + }, + { + "epoch": 99.64, + "learning_rate": 8.18711994874345e-05, + "loss": 2.7121, + "step": 49620 + }, + { + "epoch": 99.68, + "learning_rate": 8.183660609958169e-05, + "loss": 2.7029, + "step": 49640 + }, + { + "epoch": 99.72, + "learning_rate": 8.18019870615504e-05, + "loss": 2.7157, + "step": 49660 + }, + { + "epoch": 99.76, + "learning_rate": 8.176734240123255e-05, + "loss": 2.7122, + "step": 49680 + }, + { + "epoch": 99.8, + "learning_rate": 8.17326721465407e-05, + "loss": 2.7088, + "step": 49700 + }, + { + "epoch": 99.84, + "learning_rate": 8.16979763254081e-05, + "loss": 2.7301, + "step": 49720 + }, + { + "epoch": 99.88, + "learning_rate": 8.166325496578847e-05, + "loss": 2.7209, + "step": 49740 + }, + { + "epoch": 99.92, + "learning_rate": 8.162850809565623e-05, + "loss": 2.7201, + "step": 49760 + }, + { + "epoch": 99.96, + "learning_rate": 8.159373574300629e-05, + "loss": 2.7294, + "step": 49780 + }, + { + "epoch": 100.0, + "learning_rate": 8.155893793585413e-05, + "loss": 2.7294, + "step": 49800 + }, + { + "epoch": 100.04, + "learning_rate": 8.152411470223569e-05, + "loss": 2.6203, + "step": 49820 + }, + { + "epoch": 100.08, + "learning_rate": 8.148926607020743e-05, + "loss": 2.649, + "step": 49840 + }, + { + "epoch": 100.12, + "learning_rate": 8.145439206784626e-05, + "loss": 2.6498, + "step": 49860 + }, + { + "epoch": 100.16, + "learning_rate": 8.141949272324953e-05, + "loss": 2.6425, + "step": 49880 + }, + { + "epoch": 100.2, + "learning_rate": 8.138456806453503e-05, + "loss": 2.6598, + "step": 49900 + }, + { + "epoch": 100.24, + "learning_rate": 8.13496181198409e-05, + "loss": 2.6693, + "step": 49920 + }, + { + "epoch": 100.28, + "learning_rate": 8.131464291732572e-05, + "loss": 2.6607, + "step": 49940 + }, + { + "epoch": 100.32, + "learning_rate": 8.127964248516832e-05, + "loss": 2.6618, + "step": 49960 + }, + { + "epoch": 100.36, + "learning_rate": 8.124461685156795e-05, + "loss": 2.69, + "step": 49980 + }, + { + "epoch": 100.4, + "learning_rate": 8.120956604474415e-05, + "loss": 2.678, + "step": 50000 + }, + { + "epoch": 100.4, + "eval_loss": 3.079261541366577, + "eval_runtime": 34.7109, + "eval_samples_per_second": 22.068, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.004271116348383213, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.799390774586597e-05, + "step": 50000 + }, + { + "epoch": 100.44, + "learning_rate": 8.117449009293668e-05, + "loss": 2.6639, + "step": 50020 + }, + { + "epoch": 100.48, + "learning_rate": 8.113938902440564e-05, + "loss": 2.6976, + "step": 50040 + }, + { + "epoch": 100.52, + "learning_rate": 8.110426286743129e-05, + "loss": 2.7093, + "step": 50060 + }, + { + "epoch": 100.56, + "learning_rate": 8.106911165031415e-05, + "loss": 2.6917, + "step": 50080 + }, + { + "epoch": 100.6, + "learning_rate": 8.103393540137496e-05, + "loss": 2.7193, + "step": 50100 + }, + { + "epoch": 100.64, + "learning_rate": 8.099873414895453e-05, + "loss": 2.7003, + "step": 50120 + }, + { + "epoch": 100.68, + "learning_rate": 8.096350792141392e-05, + "loss": 2.7088, + "step": 50140 + }, + { + "epoch": 100.72, + "learning_rate": 8.092825674713425e-05, + "loss": 2.7041, + "step": 50160 + }, + { + "epoch": 100.76, + "learning_rate": 8.089298065451672e-05, + "loss": 2.7071, + "step": 50180 + }, + { + "epoch": 100.8, + "learning_rate": 8.085767967198269e-05, + "loss": 2.7234, + "step": 50200 + }, + { + "epoch": 100.84, + "learning_rate": 8.082235382797349e-05, + "loss": 2.697, + "step": 50220 + }, + { + "epoch": 100.88, + "learning_rate": 8.078700315095055e-05, + "loss": 2.711, + "step": 50240 + }, + { + "epoch": 100.92, + "learning_rate": 8.075162766939526e-05, + "loss": 2.7079, + "step": 50260 + }, + { + "epoch": 100.96, + "learning_rate": 8.071622741180898e-05, + "loss": 2.7085, + "step": 50280 + }, + { + "epoch": 101.0, + "learning_rate": 8.068080240671308e-05, + "loss": 2.7087, + "step": 50300 + }, + { + "epoch": 101.04, + "learning_rate": 8.064535268264883e-05, + "loss": 2.6268, + "step": 50320 + }, + { + "epoch": 101.08, + "learning_rate": 8.060987826817745e-05, + "loss": 2.6246, + "step": 50340 + }, + { + "epoch": 101.12, + "learning_rate": 8.057437919188005e-05, + "loss": 2.6285, + "step": 50360 + }, + { + "epoch": 101.16, + "learning_rate": 8.053885548235755e-05, + "loss": 2.6463, + "step": 50380 + }, + { + "epoch": 101.2, + "learning_rate": 8.05033071682308e-05, + "loss": 2.6408, + "step": 50400 + }, + { + "epoch": 101.24, + "learning_rate": 8.046773427814042e-05, + "loss": 2.6415, + "step": 50420 + }, + { + "epoch": 101.29, + "learning_rate": 8.043213684074684e-05, + "loss": 2.6658, + "step": 50440 + }, + { + "epoch": 101.33, + "learning_rate": 8.039651488473028e-05, + "loss": 2.668, + "step": 50460 + }, + { + "epoch": 101.37, + "learning_rate": 8.03608684387907e-05, + "loss": 2.6669, + "step": 50480 + }, + { + "epoch": 101.41, + "learning_rate": 8.03251975316478e-05, + "loss": 2.6785, + "step": 50500 + }, + { + "epoch": 101.45, + "learning_rate": 8.0289502192041e-05, + "loss": 2.6668, + "step": 50520 + }, + { + "epoch": 101.49, + "learning_rate": 8.025378244872936e-05, + "loss": 2.6893, + "step": 50540 + }, + { + "epoch": 101.53, + "learning_rate": 8.021803833049166e-05, + "loss": 2.6778, + "step": 50560 + }, + { + "epoch": 101.57, + "learning_rate": 8.01822698661263e-05, + "loss": 2.6786, + "step": 50580 + }, + { + "epoch": 101.61, + "learning_rate": 8.014647708445124e-05, + "loss": 2.6797, + "step": 50600 + }, + { + "epoch": 101.65, + "learning_rate": 8.011066001430412e-05, + "loss": 2.6881, + "step": 50620 + }, + { + "epoch": 101.69, + "learning_rate": 8.007481868454208e-05, + "loss": 2.711, + "step": 50640 + }, + { + "epoch": 101.73, + "learning_rate": 8.003895312404183e-05, + "loss": 2.6926, + "step": 50660 + }, + { + "epoch": 101.77, + "learning_rate": 8.000306336169963e-05, + "loss": 2.6868, + "step": 50680 + }, + { + "epoch": 101.81, + "learning_rate": 7.99671494264312e-05, + "loss": 2.7059, + "step": 50700 + }, + { + "epoch": 101.85, + "learning_rate": 7.993121134717177e-05, + "loss": 2.692, + "step": 50720 + }, + { + "epoch": 101.89, + "learning_rate": 7.989524915287595e-05, + "loss": 2.7106, + "step": 50740 + }, + { + "epoch": 101.93, + "learning_rate": 7.985926287251787e-05, + "loss": 2.7176, + "step": 50760 + }, + { + "epoch": 101.97, + "learning_rate": 7.982325253509102e-05, + "loss": 2.7039, + "step": 50780 + }, + { + "epoch": 102.01, + "learning_rate": 7.978721816960826e-05, + "loss": 2.7078, + "step": 50800 + }, + { + "epoch": 102.05, + "learning_rate": 7.975115980510187e-05, + "loss": 2.6123, + "step": 50820 + }, + { + "epoch": 102.09, + "learning_rate": 7.971507747062337e-05, + "loss": 2.6222, + "step": 50840 + }, + { + "epoch": 102.13, + "learning_rate": 7.967897119524368e-05, + "loss": 2.6287, + "step": 50860 + }, + { + "epoch": 102.17, + "learning_rate": 7.964284100805297e-05, + "loss": 2.6512, + "step": 50880 + }, + { + "epoch": 102.21, + "learning_rate": 7.960668693816067e-05, + "loss": 2.6374, + "step": 50900 + }, + { + "epoch": 102.25, + "learning_rate": 7.957050901469545e-05, + "loss": 2.6407, + "step": 50920 + }, + { + "epoch": 102.29, + "learning_rate": 7.953430726680524e-05, + "loss": 2.6399, + "step": 50940 + }, + { + "epoch": 102.33, + "learning_rate": 7.949808172365713e-05, + "loss": 2.6497, + "step": 50960 + }, + { + "epoch": 102.37, + "learning_rate": 7.946183241443736e-05, + "loss": 2.6546, + "step": 50980 + }, + { + "epoch": 102.41, + "learning_rate": 7.942555936835135e-05, + "loss": 2.6771, + "step": 51000 + }, + { + "epoch": 102.41, + "eval_loss": 3.0642032623291016, + "eval_runtime": 32.4308, + "eval_samples_per_second": 23.62, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004621541562042848, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.57274441543371e-05, + "step": 51000 + }, + { + "epoch": 102.45, + "learning_rate": 7.938926261462366e-05, + "loss": 2.6729, + "step": 51020 + }, + { + "epoch": 102.49, + "learning_rate": 7.935294218249791e-05, + "loss": 2.6744, + "step": 51040 + }, + { + "epoch": 102.53, + "learning_rate": 7.931659810123683e-05, + "loss": 2.6603, + "step": 51060 + }, + { + "epoch": 102.57, + "learning_rate": 7.928023040012216e-05, + "loss": 2.6606, + "step": 51080 + }, + { + "epoch": 102.61, + "learning_rate": 7.92456592328612e-05, + "loss": 2.6634, + "step": 51100 + }, + { + "epoch": 102.65, + "learning_rate": 7.920924555732582e-05, + "loss": 2.657, + "step": 51120 + }, + { + "epoch": 102.69, + "learning_rate": 7.917280834842888e-05, + "loss": 2.6818, + "step": 51140 + }, + { + "epoch": 102.73, + "learning_rate": 7.913634763552717e-05, + "loss": 2.6863, + "step": 51160 + }, + { + "epoch": 102.77, + "learning_rate": 7.909986344799645e-05, + "loss": 2.6974, + "step": 51180 + }, + { + "epoch": 102.81, + "learning_rate": 7.906335581523135e-05, + "loss": 2.6819, + "step": 51200 + }, + { + "epoch": 102.85, + "learning_rate": 7.90268247666454e-05, + "loss": 2.7014, + "step": 51220 + }, + { + "epoch": 102.89, + "learning_rate": 7.899027033167104e-05, + "loss": 2.7026, + "step": 51240 + }, + { + "epoch": 102.93, + "learning_rate": 7.895369253975951e-05, + "loss": 2.6965, + "step": 51260 + }, + { + "epoch": 102.97, + "learning_rate": 7.891709142038084e-05, + "loss": 2.6928, + "step": 51280 + }, + { + "epoch": 103.01, + "learning_rate": 7.888046700302394e-05, + "loss": 2.6734, + "step": 51300 + }, + { + "epoch": 103.05, + "learning_rate": 7.884381931719642e-05, + "loss": 2.6133, + "step": 51320 + }, + { + "epoch": 103.09, + "learning_rate": 7.880714839242464e-05, + "loss": 2.6042, + "step": 51340 + }, + { + "epoch": 103.13, + "learning_rate": 7.877045425825374e-05, + "loss": 2.6168, + "step": 51360 + }, + { + "epoch": 103.17, + "learning_rate": 7.873373694424751e-05, + "loss": 2.6233, + "step": 51380 + }, + { + "epoch": 103.21, + "learning_rate": 7.869699647998839e-05, + "loss": 2.6457, + "step": 51400 + }, + { + "epoch": 103.25, + "learning_rate": 7.866023289507754e-05, + "loss": 2.6278, + "step": 51420 + }, + { + "epoch": 103.29, + "learning_rate": 7.862344621913472e-05, + "loss": 2.6469, + "step": 51440 + }, + { + "epoch": 103.33, + "learning_rate": 7.858663648179828e-05, + "loss": 2.6473, + "step": 51460 + }, + { + "epoch": 103.37, + "learning_rate": 7.854980371272517e-05, + "loss": 2.6416, + "step": 51480 + }, + { + "epoch": 103.41, + "learning_rate": 7.851294794159087e-05, + "loss": 2.6532, + "step": 51500 + }, + { + "epoch": 103.45, + "learning_rate": 7.847606919808944e-05, + "loss": 2.6419, + "step": 51520 + }, + { + "epoch": 103.49, + "learning_rate": 7.843916751193336e-05, + "loss": 2.64, + "step": 51540 + }, + { + "epoch": 103.53, + "learning_rate": 7.840224291285372e-05, + "loss": 2.6659, + "step": 51560 + }, + { + "epoch": 103.57, + "learning_rate": 7.836529543059995e-05, + "loss": 2.663, + "step": 51580 + }, + { + "epoch": 103.61, + "learning_rate": 7.832832509494001e-05, + "loss": 2.6696, + "step": 51600 + }, + { + "epoch": 103.65, + "learning_rate": 7.82913319356602e-05, + "loss": 2.6984, + "step": 51620 + }, + { + "epoch": 103.69, + "learning_rate": 7.825431598256525e-05, + "loss": 2.6915, + "step": 51640 + }, + { + "epoch": 103.73, + "learning_rate": 7.821727726547825e-05, + "loss": 2.6612, + "step": 51660 + }, + { + "epoch": 103.78, + "learning_rate": 7.818021581424064e-05, + "loss": 2.6782, + "step": 51680 + }, + { + "epoch": 103.82, + "learning_rate": 7.814313165871212e-05, + "loss": 2.6757, + "step": 51700 + }, + { + "epoch": 103.86, + "learning_rate": 7.810602482877075e-05, + "loss": 2.6952, + "step": 51720 + }, + { + "epoch": 103.9, + "learning_rate": 7.806889535431284e-05, + "loss": 2.6807, + "step": 51740 + }, + { + "epoch": 103.94, + "learning_rate": 7.803174326525293e-05, + "loss": 2.7006, + "step": 51760 + }, + { + "epoch": 103.98, + "learning_rate": 7.799456859152378e-05, + "loss": 2.6839, + "step": 51780 + }, + { + "epoch": 104.02, + "learning_rate": 7.795737136307636e-05, + "loss": 2.6529, + "step": 51800 + }, + { + "epoch": 104.06, + "learning_rate": 7.79201516098798e-05, + "loss": 2.5902, + "step": 51820 + }, + { + "epoch": 104.1, + "learning_rate": 7.78829093619214e-05, + "loss": 2.5988, + "step": 51840 + }, + { + "epoch": 104.14, + "learning_rate": 7.784564464920654e-05, + "loss": 2.6157, + "step": 51860 + }, + { + "epoch": 104.18, + "learning_rate": 7.780835750175874e-05, + "loss": 2.634, + "step": 51880 + }, + { + "epoch": 104.22, + "learning_rate": 7.777104794961957e-05, + "loss": 2.6197, + "step": 51900 + }, + { + "epoch": 104.26, + "learning_rate": 7.773371602284869e-05, + "loss": 2.6206, + "step": 51920 + }, + { + "epoch": 104.3, + "learning_rate": 7.769636175152374e-05, + "loss": 2.6333, + "step": 51940 + }, + { + "epoch": 104.34, + "learning_rate": 7.765898516574038e-05, + "loss": 2.6526, + "step": 51960 + }, + { + "epoch": 104.38, + "learning_rate": 7.762158629561225e-05, + "loss": 2.6331, + "step": 51980 + }, + { + "epoch": 104.42, + "learning_rate": 7.758416517127094e-05, + "loss": 2.634, + "step": 52000 + }, + { + "epoch": 104.42, + "eval_loss": 3.054405689239502, + "eval_runtime": 32.4234, + "eval_samples_per_second": 23.625, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004932724814860198, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00034676892950391645, + "step": 52000 + }, + { + "epoch": 104.46, + "learning_rate": 7.754859451764212e-05, + "loss": 2.631, + "step": 52020 + }, + { + "epoch": 104.5, + "learning_rate": 7.7511130084319e-05, + "loss": 2.6425, + "step": 52040 + }, + { + "epoch": 104.54, + "learning_rate": 7.747364348577531e-05, + "loss": 2.6574, + "step": 52060 + }, + { + "epoch": 104.58, + "learning_rate": 7.743613475221328e-05, + "loss": 2.6533, + "step": 52080 + }, + { + "epoch": 104.62, + "learning_rate": 7.739860391385303e-05, + "loss": 2.6766, + "step": 52100 + }, + { + "epoch": 104.66, + "learning_rate": 7.736105100093249e-05, + "loss": 2.6621, + "step": 52120 + }, + { + "epoch": 104.7, + "learning_rate": 7.732347604370736e-05, + "loss": 2.6519, + "step": 52140 + }, + { + "epoch": 104.74, + "learning_rate": 7.728587907245108e-05, + "loss": 2.6606, + "step": 52160 + }, + { + "epoch": 104.78, + "learning_rate": 7.724826011745491e-05, + "loss": 2.6816, + "step": 52180 + }, + { + "epoch": 104.82, + "learning_rate": 7.72106192090277e-05, + "loss": 2.6627, + "step": 52200 + }, + { + "epoch": 104.86, + "learning_rate": 7.717295637749609e-05, + "loss": 2.6645, + "step": 52220 + }, + { + "epoch": 104.9, + "learning_rate": 7.713527165320432e-05, + "loss": 2.6755, + "step": 52240 + }, + { + "epoch": 104.94, + "learning_rate": 7.709756506651432e-05, + "loss": 2.6821, + "step": 52260 + }, + { + "epoch": 104.98, + "learning_rate": 7.705983664780556e-05, + "loss": 2.6838, + "step": 52280 + }, + { + "epoch": 105.02, + "learning_rate": 7.70220864274752e-05, + "loss": 2.6388, + "step": 52300 + }, + { + "epoch": 105.06, + "learning_rate": 7.698431443593787e-05, + "loss": 2.5879, + "step": 52320 + }, + { + "epoch": 105.1, + "learning_rate": 7.694652070362578e-05, + "loss": 2.5858, + "step": 52340 + }, + { + "epoch": 105.14, + "learning_rate": 7.690870526098868e-05, + "loss": 2.6024, + "step": 52360 + }, + { + "epoch": 105.18, + "learning_rate": 7.687086813849378e-05, + "loss": 2.6119, + "step": 52380 + }, + { + "epoch": 105.22, + "learning_rate": 7.683300936662579e-05, + "loss": 2.6124, + "step": 52400 + }, + { + "epoch": 105.26, + "learning_rate": 7.679512897588679e-05, + "loss": 2.603, + "step": 52420 + }, + { + "epoch": 105.3, + "learning_rate": 7.675722699679636e-05, + "loss": 2.6278, + "step": 52440 + }, + { + "epoch": 105.34, + "learning_rate": 7.671930345989142e-05, + "loss": 2.6411, + "step": 52460 + }, + { + "epoch": 105.38, + "learning_rate": 7.668135839572631e-05, + "loss": 2.6421, + "step": 52480 + }, + { + "epoch": 105.42, + "learning_rate": 7.664339183487266e-05, + "loss": 2.6441, + "step": 52500 + }, + { + "epoch": 105.46, + "learning_rate": 7.660540380791942e-05, + "loss": 2.6444, + "step": 52520 + }, + { + "epoch": 105.5, + "learning_rate": 7.656739434547293e-05, + "loss": 2.655, + "step": 52540 + }, + { + "epoch": 105.54, + "learning_rate": 7.652936347815664e-05, + "loss": 2.6485, + "step": 52560 + }, + { + "epoch": 105.58, + "learning_rate": 7.649131123661139e-05, + "loss": 2.6337, + "step": 52580 + }, + { + "epoch": 105.62, + "learning_rate": 7.645323765149518e-05, + "loss": 2.6413, + "step": 52600 + }, + { + "epoch": 105.66, + "learning_rate": 7.641514275348318e-05, + "loss": 2.6385, + "step": 52620 + }, + { + "epoch": 105.7, + "learning_rate": 7.637702657326779e-05, + "loss": 2.6623, + "step": 52640 + }, + { + "epoch": 105.74, + "learning_rate": 7.63388891415585e-05, + "loss": 2.6607, + "step": 52660 + }, + { + "epoch": 105.78, + "learning_rate": 7.630073048908197e-05, + "loss": 2.6485, + "step": 52680 + }, + { + "epoch": 105.82, + "learning_rate": 7.626255064658194e-05, + "loss": 2.6582, + "step": 52700 + }, + { + "epoch": 105.86, + "learning_rate": 7.622434964481921e-05, + "loss": 2.6708, + "step": 52720 + }, + { + "epoch": 105.9, + "learning_rate": 7.618612751457163e-05, + "loss": 2.6595, + "step": 52740 + }, + { + "epoch": 105.94, + "learning_rate": 7.614788428663411e-05, + "loss": 2.6774, + "step": 52760 + }, + { + "epoch": 105.98, + "learning_rate": 7.610961999181849e-05, + "loss": 2.673, + "step": 52780 + }, + { + "epoch": 106.02, + "learning_rate": 7.607133466095365e-05, + "loss": 2.6105, + "step": 52800 + }, + { + "epoch": 106.06, + "learning_rate": 7.603302832488537e-05, + "loss": 2.5771, + "step": 52820 + }, + { + "epoch": 106.1, + "learning_rate": 7.59947010144764e-05, + "loss": 2.5918, + "step": 52840 + }, + { + "epoch": 106.14, + "learning_rate": 7.595635276060633e-05, + "loss": 2.5952, + "step": 52860 + }, + { + "epoch": 106.18, + "learning_rate": 7.591798359417167e-05, + "loss": 2.5947, + "step": 52880 + }, + { + "epoch": 106.22, + "learning_rate": 7.587959354608576e-05, + "loss": 2.6015, + "step": 52900 + }, + { + "epoch": 106.27, + "learning_rate": 7.584118264727878e-05, + "loss": 2.6106, + "step": 52920 + }, + { + "epoch": 106.31, + "learning_rate": 7.580275092869766e-05, + "loss": 2.6133, + "step": 52940 + }, + { + "epoch": 106.35, + "learning_rate": 7.576429842130621e-05, + "loss": 2.6292, + "step": 52960 + }, + { + "epoch": 106.39, + "learning_rate": 7.572582515608486e-05, + "loss": 2.6153, + "step": 52980 + }, + { + "epoch": 106.43, + "learning_rate": 7.568733116403085e-05, + "loss": 2.6366, + "step": 53000 + }, + { + "epoch": 106.43, + "eval_loss": 3.0405330657958984, + "eval_runtime": 34.7493, + "eval_samples_per_second": 22.044, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.004604863710063153, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 7.479329852045257e-05, + "step": 53000 + }, + { + "epoch": 106.47, + "learning_rate": 7.564881647615809e-05, + "loss": 2.6173, + "step": 53020 + }, + { + "epoch": 106.51, + "learning_rate": 7.561028112349716e-05, + "loss": 2.6326, + "step": 53040 + }, + { + "epoch": 106.55, + "learning_rate": 7.557172513709532e-05, + "loss": 2.6263, + "step": 53060 + }, + { + "epoch": 106.59, + "learning_rate": 7.553314854801641e-05, + "loss": 2.6316, + "step": 53080 + }, + { + "epoch": 106.63, + "learning_rate": 7.549455138734089e-05, + "loss": 2.6468, + "step": 53100 + }, + { + "epoch": 106.67, + "learning_rate": 7.545593368616582e-05, + "loss": 2.6613, + "step": 53120 + }, + { + "epoch": 106.71, + "learning_rate": 7.541729547560477e-05, + "loss": 2.637, + "step": 53140 + }, + { + "epoch": 106.75, + "learning_rate": 7.537863678678783e-05, + "loss": 2.6444, + "step": 53160 + }, + { + "epoch": 106.79, + "learning_rate": 7.533995765086165e-05, + "loss": 2.6497, + "step": 53180 + }, + { + "epoch": 106.83, + "learning_rate": 7.530125809898927e-05, + "loss": 2.6616, + "step": 53200 + }, + { + "epoch": 106.87, + "learning_rate": 7.526253816235023e-05, + "loss": 2.6492, + "step": 53220 + }, + { + "epoch": 106.91, + "learning_rate": 7.522379787214049e-05, + "loss": 2.6618, + "step": 53240 + }, + { + "epoch": 106.95, + "learning_rate": 7.51850372595724e-05, + "loss": 2.6551, + "step": 53260 + }, + { + "epoch": 106.99, + "learning_rate": 7.51462563558747e-05, + "loss": 2.6664, + "step": 53280 + }, + { + "epoch": 107.03, + "learning_rate": 7.510745519229243e-05, + "loss": 2.589, + "step": 53300 + }, + { + "epoch": 107.07, + "learning_rate": 7.506863380008702e-05, + "loss": 2.5837, + "step": 53320 + }, + { + "epoch": 107.11, + "learning_rate": 7.502979221053615e-05, + "loss": 2.5701, + "step": 53340 + }, + { + "epoch": 107.15, + "learning_rate": 7.499093045493379e-05, + "loss": 2.5885, + "step": 53360 + }, + { + "epoch": 107.19, + "learning_rate": 7.495204856459014e-05, + "loss": 2.5994, + "step": 53380 + }, + { + "epoch": 107.23, + "learning_rate": 7.491314657083166e-05, + "loss": 2.6078, + "step": 53400 + }, + { + "epoch": 107.27, + "learning_rate": 7.487422450500096e-05, + "loss": 2.6223, + "step": 53420 + }, + { + "epoch": 107.31, + "learning_rate": 7.483528239845688e-05, + "loss": 2.6108, + "step": 53440 + }, + { + "epoch": 107.35, + "learning_rate": 7.479632028257435e-05, + "loss": 2.6014, + "step": 53460 + }, + { + "epoch": 107.39, + "learning_rate": 7.475733818874443e-05, + "loss": 2.6154, + "step": 53480 + }, + { + "epoch": 107.43, + "learning_rate": 7.471833614837431e-05, + "loss": 2.6007, + "step": 53500 + }, + { + "epoch": 107.47, + "learning_rate": 7.467931419288721e-05, + "loss": 2.6417, + "step": 53520 + }, + { + "epoch": 107.51, + "learning_rate": 7.464027235372244e-05, + "loss": 2.6132, + "step": 53540 + }, + { + "epoch": 107.55, + "learning_rate": 7.460121066233529e-05, + "loss": 2.6468, + "step": 53560 + }, + { + "epoch": 107.59, + "learning_rate": 7.456212915019706e-05, + "loss": 2.6124, + "step": 53580 + }, + { + "epoch": 107.63, + "learning_rate": 7.452302784879501e-05, + "loss": 2.6501, + "step": 53600 + }, + { + "epoch": 107.67, + "learning_rate": 7.448390678963238e-05, + "loss": 2.6222, + "step": 53620 + }, + { + "epoch": 107.71, + "learning_rate": 7.444476600422828e-05, + "loss": 2.6469, + "step": 53640 + }, + { + "epoch": 107.75, + "learning_rate": 7.440560552411771e-05, + "loss": 2.632, + "step": 53660 + }, + { + "epoch": 107.79, + "learning_rate": 7.43664253808516e-05, + "loss": 2.6237, + "step": 53680 + }, + { + "epoch": 107.83, + "learning_rate": 7.432722560599669e-05, + "loss": 2.646, + "step": 53700 + }, + { + "epoch": 107.87, + "learning_rate": 7.428800623113548e-05, + "loss": 2.6457, + "step": 53720 + }, + { + "epoch": 107.91, + "learning_rate": 7.424876728786637e-05, + "loss": 2.6444, + "step": 53740 + }, + { + "epoch": 107.95, + "learning_rate": 7.420950880780343e-05, + "loss": 2.6407, + "step": 53760 + }, + { + "epoch": 107.99, + "learning_rate": 7.417023082257652e-05, + "loss": 2.6423, + "step": 53780 + }, + { + "epoch": 108.03, + "learning_rate": 7.413093336383121e-05, + "loss": 2.591, + "step": 53800 + }, + { + "epoch": 108.07, + "learning_rate": 7.409161646322874e-05, + "loss": 2.5483, + "step": 53820 + }, + { + "epoch": 108.11, + "learning_rate": 7.405228015244603e-05, + "loss": 2.5707, + "step": 53840 + }, + { + "epoch": 108.15, + "learning_rate": 7.401292446317564e-05, + "loss": 2.5727, + "step": 53860 + }, + { + "epoch": 108.19, + "learning_rate": 7.397354942712574e-05, + "loss": 2.5897, + "step": 53880 + }, + { + "epoch": 108.23, + "learning_rate": 7.393415507602007e-05, + "loss": 2.5836, + "step": 53900 + }, + { + "epoch": 108.27, + "learning_rate": 7.389474144159796e-05, + "loss": 2.5984, + "step": 53920 + }, + { + "epoch": 108.31, + "learning_rate": 7.385530855561426e-05, + "loss": 2.6078, + "step": 53940 + }, + { + "epoch": 108.35, + "learning_rate": 7.381585644983935e-05, + "loss": 2.6073, + "step": 53960 + }, + { + "epoch": 108.39, + "learning_rate": 7.377638515605904e-05, + "loss": 2.605, + "step": 53980 + }, + { + "epoch": 108.43, + "learning_rate": 7.373689470607469e-05, + "loss": 2.6252, + "step": 54000 + }, + { + "epoch": 108.43, + "eval_loss": 3.0286459922790527, + "eval_runtime": 32.4828, + "eval_samples_per_second": 23.582, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004195280710142059, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00012012257035102989, + "step": 54000 + }, + { + "epoch": 108.47, + "learning_rate": 7.369738513170301e-05, + "loss": 2.6132, + "step": 54020 + }, + { + "epoch": 108.51, + "learning_rate": 7.365785646477614e-05, + "loss": 2.6158, + "step": 54040 + }, + { + "epoch": 108.55, + "learning_rate": 7.361830873714165e-05, + "loss": 2.6331, + "step": 54060 + }, + { + "epoch": 108.59, + "learning_rate": 7.35787419806624e-05, + "loss": 2.6148, + "step": 54080 + }, + { + "epoch": 108.63, + "learning_rate": 7.35391562272166e-05, + "loss": 2.6228, + "step": 54100 + }, + { + "epoch": 108.67, + "learning_rate": 7.349955150869781e-05, + "loss": 2.6298, + "step": 54120 + }, + { + "epoch": 108.71, + "learning_rate": 7.345992785701483e-05, + "loss": 2.6271, + "step": 54140 + }, + { + "epoch": 108.76, + "learning_rate": 7.342028530409171e-05, + "loss": 2.6262, + "step": 54160 + }, + { + "epoch": 108.8, + "learning_rate": 7.338062388186772e-05, + "loss": 2.6168, + "step": 54180 + }, + { + "epoch": 108.84, + "learning_rate": 7.33409436222974e-05, + "loss": 2.6182, + "step": 54200 + }, + { + "epoch": 108.88, + "learning_rate": 7.330124455735039e-05, + "loss": 2.6285, + "step": 54220 + }, + { + "epoch": 108.92, + "learning_rate": 7.32615267190115e-05, + "loss": 2.6393, + "step": 54240 + }, + { + "epoch": 108.96, + "learning_rate": 7.322179013928067e-05, + "loss": 2.639, + "step": 54260 + }, + { + "epoch": 109.0, + "learning_rate": 7.318203485017299e-05, + "loss": 2.6563, + "step": 54280 + }, + { + "epoch": 109.04, + "learning_rate": 7.314226088371854e-05, + "loss": 2.5591, + "step": 54300 + }, + { + "epoch": 109.08, + "learning_rate": 7.31024682719625e-05, + "loss": 2.5753, + "step": 54320 + }, + { + "epoch": 109.12, + "learning_rate": 7.306265704696504e-05, + "loss": 2.5705, + "step": 54340 + }, + { + "epoch": 109.16, + "learning_rate": 7.302282724080138e-05, + "loss": 2.5712, + "step": 54360 + }, + { + "epoch": 109.2, + "learning_rate": 7.298297888556164e-05, + "loss": 2.5806, + "step": 54380 + }, + { + "epoch": 109.24, + "learning_rate": 7.294311201335093e-05, + "loss": 2.5879, + "step": 54400 + }, + { + "epoch": 109.28, + "learning_rate": 7.290322665628928e-05, + "loss": 2.5824, + "step": 54420 + }, + { + "epoch": 109.32, + "learning_rate": 7.286332284651159e-05, + "loss": 2.5968, + "step": 54440 + }, + { + "epoch": 109.36, + "learning_rate": 7.282340061616766e-05, + "loss": 2.6061, + "step": 54460 + }, + { + "epoch": 109.4, + "learning_rate": 7.278345999742208e-05, + "loss": 2.5984, + "step": 54480 + }, + { + "epoch": 109.44, + "learning_rate": 7.274350102245431e-05, + "loss": 2.5996, + "step": 54500 + }, + { + "epoch": 109.48, + "learning_rate": 7.270352372345855e-05, + "loss": 2.5868, + "step": 54520 + }, + { + "epoch": 109.52, + "learning_rate": 7.266352813264378e-05, + "loss": 2.5956, + "step": 54540 + }, + { + "epoch": 109.56, + "learning_rate": 7.262351428223378e-05, + "loss": 2.6124, + "step": 54560 + }, + { + "epoch": 109.6, + "learning_rate": 7.258348220446695e-05, + "loss": 2.6044, + "step": 54580 + }, + { + "epoch": 109.64, + "learning_rate": 7.25434319315964e-05, + "loss": 2.6117, + "step": 54600 + }, + { + "epoch": 109.68, + "learning_rate": 7.250336349588994e-05, + "loss": 2.6244, + "step": 54620 + }, + { + "epoch": 109.72, + "learning_rate": 7.246327692962996e-05, + "loss": 2.6128, + "step": 54640 + }, + { + "epoch": 109.76, + "learning_rate": 7.24231722651135e-05, + "loss": 2.6341, + "step": 54660 + }, + { + "epoch": 109.8, + "learning_rate": 7.238304953465217e-05, + "loss": 2.6272, + "step": 54680 + }, + { + "epoch": 109.84, + "learning_rate": 7.234290877057208e-05, + "loss": 2.6208, + "step": 54700 + }, + { + "epoch": 109.88, + "learning_rate": 7.230275000521398e-05, + "loss": 2.6243, + "step": 54720 + }, + { + "epoch": 109.92, + "learning_rate": 7.226257327093304e-05, + "loss": 2.6104, + "step": 54740 + }, + { + "epoch": 109.96, + "learning_rate": 7.222237860009892e-05, + "loss": 2.6289, + "step": 54760 + }, + { + "epoch": 110.0, + "learning_rate": 7.218216602509574e-05, + "loss": 2.6152, + "step": 54780 + }, + { + "epoch": 110.04, + "learning_rate": 7.214193557832206e-05, + "loss": 2.5496, + "step": 54800 + }, + { + "epoch": 110.08, + "learning_rate": 7.21016872921908e-05, + "loss": 2.5612, + "step": 54820 + }, + { + "epoch": 110.12, + "learning_rate": 7.206142119912931e-05, + "loss": 2.5478, + "step": 54840 + }, + { + "epoch": 110.16, + "learning_rate": 7.202113733157923e-05, + "loss": 2.5713, + "step": 54860 + }, + { + "epoch": 110.2, + "learning_rate": 7.198083572199657e-05, + "loss": 2.5561, + "step": 54880 + }, + { + "epoch": 110.24, + "learning_rate": 7.194051640285157e-05, + "loss": 2.5849, + "step": 54900 + }, + { + "epoch": 110.28, + "learning_rate": 7.190017940662878e-05, + "loss": 2.5678, + "step": 54920 + }, + { + "epoch": 110.32, + "learning_rate": 7.185982476582705e-05, + "loss": 2.5855, + "step": 54940 + }, + { + "epoch": 110.36, + "learning_rate": 7.181945251295931e-05, + "loss": 2.5725, + "step": 54960 + }, + { + "epoch": 110.4, + "learning_rate": 7.17790626805528e-05, + "loss": 2.589, + "step": 54980 + }, + { + "epoch": 110.44, + "learning_rate": 7.173865530114886e-05, + "loss": 2.6023, + "step": 55000 + }, + { + "epoch": 110.44, + "eval_loss": 3.0221402645111084, + "eval_runtime": 32.4674, + "eval_samples_per_second": 23.593, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0038154856301598756, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 6.799390774586597e-05, + "step": 55000 + }, + { + "epoch": 110.48, + "learning_rate": 7.1698230407303e-05, + "loss": 2.5987, + "step": 55020 + }, + { + "epoch": 110.52, + "learning_rate": 7.16577880315848e-05, + "loss": 2.6037, + "step": 55040 + }, + { + "epoch": 110.56, + "learning_rate": 7.161732820657799e-05, + "loss": 2.5935, + "step": 55060 + }, + { + "epoch": 110.6, + "learning_rate": 7.157685096488029e-05, + "loss": 2.6005, + "step": 55080 + }, + { + "epoch": 110.64, + "learning_rate": 7.153635633910349e-05, + "loss": 2.6083, + "step": 55100 + }, + { + "epoch": 110.68, + "learning_rate": 7.149584436187338e-05, + "loss": 2.6003, + "step": 55120 + }, + { + "epoch": 110.72, + "learning_rate": 7.145531506582975e-05, + "loss": 2.6072, + "step": 55140 + }, + { + "epoch": 110.76, + "learning_rate": 7.141476848362627e-05, + "loss": 2.6025, + "step": 55160 + }, + { + "epoch": 110.8, + "learning_rate": 7.137420464793063e-05, + "loss": 2.6184, + "step": 55180 + }, + { + "epoch": 110.84, + "learning_rate": 7.133362359142439e-05, + "loss": 2.618, + "step": 55200 + }, + { + "epoch": 110.88, + "learning_rate": 7.129302534680293e-05, + "loss": 2.6265, + "step": 55220 + }, + { + "epoch": 110.92, + "learning_rate": 7.125240994677557e-05, + "loss": 2.6284, + "step": 55240 + }, + { + "epoch": 110.96, + "learning_rate": 7.121177742406534e-05, + "loss": 2.6213, + "step": 55260 + }, + { + "epoch": 111.0, + "learning_rate": 7.11711278114092e-05, + "loss": 2.6197, + "step": 55280 + }, + { + "epoch": 111.04, + "learning_rate": 7.113046114155777e-05, + "loss": 2.534, + "step": 55300 + }, + { + "epoch": 111.08, + "learning_rate": 7.108977744727547e-05, + "loss": 2.5525, + "step": 55320 + }, + { + "epoch": 111.12, + "learning_rate": 7.104907676134041e-05, + "loss": 2.554, + "step": 55340 + }, + { + "epoch": 111.16, + "learning_rate": 7.100835911654437e-05, + "loss": 2.5613, + "step": 55360 + }, + { + "epoch": 111.2, + "learning_rate": 7.096762454569289e-05, + "loss": 2.5578, + "step": 55380 + }, + { + "epoch": 111.24, + "learning_rate": 7.0926873081605e-05, + "loss": 2.5584, + "step": 55400 + }, + { + "epoch": 111.29, + "learning_rate": 7.088610475711345e-05, + "loss": 2.564, + "step": 55420 + }, + { + "epoch": 111.33, + "learning_rate": 7.084531960506456e-05, + "loss": 2.5766, + "step": 55440 + }, + { + "epoch": 111.37, + "learning_rate": 7.080451765831817e-05, + "loss": 2.578, + "step": 55460 + }, + { + "epoch": 111.41, + "learning_rate": 7.076369894974768e-05, + "loss": 2.5801, + "step": 55480 + }, + { + "epoch": 111.45, + "learning_rate": 7.072286351223999e-05, + "loss": 2.5665, + "step": 55500 + }, + { + "epoch": 111.49, + "learning_rate": 7.068201137869546e-05, + "loss": 2.5951, + "step": 55520 + }, + { + "epoch": 111.53, + "learning_rate": 7.064114258202792e-05, + "loss": 2.6012, + "step": 55540 + }, + { + "epoch": 111.57, + "learning_rate": 7.060025715516463e-05, + "loss": 2.5858, + "step": 55560 + }, + { + "epoch": 111.61, + "learning_rate": 7.055935513104623e-05, + "loss": 2.5924, + "step": 55580 + }, + { + "epoch": 111.65, + "learning_rate": 7.051843654262676e-05, + "loss": 2.615, + "step": 55600 + }, + { + "epoch": 111.69, + "learning_rate": 7.047750142287356e-05, + "loss": 2.593, + "step": 55620 + }, + { + "epoch": 111.73, + "learning_rate": 7.043654980476735e-05, + "loss": 2.6138, + "step": 55640 + }, + { + "epoch": 111.77, + "learning_rate": 7.039558172130208e-05, + "loss": 2.6217, + "step": 55660 + }, + { + "epoch": 111.81, + "learning_rate": 7.0354597205485e-05, + "loss": 2.6048, + "step": 55680 + }, + { + "epoch": 111.85, + "learning_rate": 7.031359629033661e-05, + "loss": 2.6063, + "step": 55700 + }, + { + "epoch": 111.89, + "learning_rate": 7.027257900889059e-05, + "loss": 2.6088, + "step": 55720 + }, + { + "epoch": 111.93, + "learning_rate": 7.023154539419384e-05, + "loss": 2.5975, + "step": 55740 + }, + { + "epoch": 111.97, + "learning_rate": 7.019049547930638e-05, + "loss": 2.6106, + "step": 55760 + }, + { + "epoch": 112.01, + "learning_rate": 7.01494292973014e-05, + "loss": 2.5893, + "step": 55780 + }, + { + "epoch": 112.05, + "learning_rate": 7.010834688126518e-05, + "loss": 2.5097, + "step": 55800 + }, + { + "epoch": 112.09, + "learning_rate": 7.006724826429706e-05, + "loss": 2.5238, + "step": 55820 + }, + { + "epoch": 112.13, + "learning_rate": 7.002613347950949e-05, + "loss": 2.5307, + "step": 55840 + }, + { + "epoch": 112.17, + "learning_rate": 6.998500256002789e-05, + "loss": 2.5305, + "step": 55860 + }, + { + "epoch": 112.21, + "learning_rate": 6.994385553899069e-05, + "loss": 2.5612, + "step": 55880 + }, + { + "epoch": 112.25, + "learning_rate": 6.990269244954933e-05, + "loss": 2.5492, + "step": 55900 + }, + { + "epoch": 112.29, + "learning_rate": 6.986151332486813e-05, + "loss": 2.5401, + "step": 55920 + }, + { + "epoch": 112.33, + "learning_rate": 6.98203181981244e-05, + "loss": 2.5733, + "step": 55940 + }, + { + "epoch": 112.37, + "learning_rate": 6.97791071025083e-05, + "loss": 2.5725, + "step": 55960 + }, + { + "epoch": 112.41, + "learning_rate": 6.973788007122283e-05, + "loss": 2.5625, + "step": 55980 + }, + { + "epoch": 112.45, + "learning_rate": 6.969663713748392e-05, + "loss": 2.5941, + "step": 56000 + }, + { + "epoch": 112.45, + "eval_loss": 3.009904623031616, + "eval_runtime": 32.253, + "eval_samples_per_second": 23.75, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.0039947700798041805, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 9.745793443574124e-05, + "step": 56000 + }, + { + "epoch": 112.49, + "learning_rate": 6.965537833452024e-05, + "loss": 2.5722, + "step": 56020 + }, + { + "epoch": 112.53, + "learning_rate": 6.961616780311188e-05, + "loss": 2.5925, + "step": 56040 + }, + { + "epoch": 112.57, + "learning_rate": 6.957487815078227e-05, + "loss": 2.5961, + "step": 56060 + }, + { + "epoch": 112.61, + "learning_rate": 6.953357272732691e-05, + "loss": 2.587, + "step": 56080 + }, + { + "epoch": 112.65, + "learning_rate": 6.949225156602481e-05, + "loss": 2.5951, + "step": 56100 + }, + { + "epoch": 112.69, + "learning_rate": 6.945091470016771e-05, + "loss": 2.5933, + "step": 56120 + }, + { + "epoch": 112.73, + "learning_rate": 6.940956216305996e-05, + "loss": 2.5988, + "step": 56140 + }, + { + "epoch": 112.77, + "learning_rate": 6.936819398801856e-05, + "loss": 2.6035, + "step": 56160 + }, + { + "epoch": 112.81, + "learning_rate": 6.932681020837307e-05, + "loss": 2.594, + "step": 56180 + }, + { + "epoch": 112.85, + "learning_rate": 6.928541085746571e-05, + "loss": 2.6158, + "step": 56200 + }, + { + "epoch": 112.89, + "learning_rate": 6.924399596865113e-05, + "loss": 2.6003, + "step": 56220 + }, + { + "epoch": 112.93, + "learning_rate": 6.92025655752966e-05, + "loss": 2.6145, + "step": 56240 + }, + { + "epoch": 112.97, + "learning_rate": 6.916111971078183e-05, + "loss": 2.6099, + "step": 56260 + }, + { + "epoch": 113.01, + "learning_rate": 6.9119658408499e-05, + "loss": 2.5951, + "step": 56280 + }, + { + "epoch": 113.05, + "learning_rate": 6.907818170185273e-05, + "loss": 2.5065, + "step": 56300 + }, + { + "epoch": 113.09, + "learning_rate": 6.903668962426005e-05, + "loss": 2.5299, + "step": 56320 + }, + { + "epoch": 113.13, + "learning_rate": 6.899518220915038e-05, + "loss": 2.5414, + "step": 56340 + }, + { + "epoch": 113.17, + "learning_rate": 6.895365948996552e-05, + "loss": 2.5514, + "step": 56360 + }, + { + "epoch": 113.21, + "learning_rate": 6.891212150015955e-05, + "loss": 2.544, + "step": 56380 + }, + { + "epoch": 113.25, + "learning_rate": 6.887056827319885e-05, + "loss": 2.5589, + "step": 56400 + }, + { + "epoch": 113.29, + "learning_rate": 6.882899984256216e-05, + "loss": 2.5734, + "step": 56420 + }, + { + "epoch": 113.33, + "learning_rate": 6.878741624174039e-05, + "loss": 2.5618, + "step": 56440 + }, + { + "epoch": 113.37, + "learning_rate": 6.87458175042367e-05, + "loss": 2.5582, + "step": 56460 + }, + { + "epoch": 113.41, + "learning_rate": 6.870420366356642e-05, + "loss": 2.5478, + "step": 56480 + }, + { + "epoch": 113.45, + "learning_rate": 6.86625747532571e-05, + "loss": 2.5668, + "step": 56500 + }, + { + "epoch": 113.49, + "learning_rate": 6.862093080684838e-05, + "loss": 2.5627, + "step": 56520 + }, + { + "epoch": 113.53, + "learning_rate": 6.857927185789204e-05, + "loss": 2.5589, + "step": 56540 + }, + { + "epoch": 113.57, + "learning_rate": 6.853759793995196e-05, + "loss": 2.5899, + "step": 56560 + }, + { + "epoch": 113.61, + "learning_rate": 6.849590908660404e-05, + "loss": 2.5946, + "step": 56580 + }, + { + "epoch": 113.65, + "learning_rate": 6.845420533143627e-05, + "loss": 2.5814, + "step": 56600 + }, + { + "epoch": 113.69, + "learning_rate": 6.841248670804853e-05, + "loss": 2.5996, + "step": 56620 + }, + { + "epoch": 113.73, + "learning_rate": 6.837075325005286e-05, + "loss": 2.5927, + "step": 56640 + }, + { + "epoch": 113.78, + "learning_rate": 6.832900499107311e-05, + "loss": 2.5704, + "step": 56660 + }, + { + "epoch": 113.82, + "learning_rate": 6.82872419647451e-05, + "loss": 2.5842, + "step": 56680 + }, + { + "epoch": 113.86, + "learning_rate": 6.824546420471653e-05, + "loss": 2.5822, + "step": 56700 + }, + { + "epoch": 113.9, + "learning_rate": 6.820367174464703e-05, + "loss": 2.6015, + "step": 56720 + }, + { + "epoch": 113.94, + "learning_rate": 6.816186461820798e-05, + "loss": 2.6006, + "step": 56740 + }, + { + "epoch": 113.98, + "learning_rate": 6.812004285908266e-05, + "loss": 2.589, + "step": 56760 + }, + { + "epoch": 114.02, + "learning_rate": 6.807820650096609e-05, + "loss": 2.5718, + "step": 56780 + }, + { + "epoch": 114.06, + "learning_rate": 6.803635557756507e-05, + "loss": 2.5255, + "step": 56800 + }, + { + "epoch": 114.1, + "learning_rate": 6.799449012259816e-05, + "loss": 2.5288, + "step": 56820 + }, + { + "epoch": 114.14, + "learning_rate": 6.795261016979555e-05, + "loss": 2.5301, + "step": 56840 + }, + { + "epoch": 114.18, + "learning_rate": 6.791071575289922e-05, + "loss": 2.5238, + "step": 56860 + }, + { + "epoch": 114.22, + "learning_rate": 6.786880690566268e-05, + "loss": 2.5333, + "step": 56880 + }, + { + "epoch": 114.26, + "learning_rate": 6.78268836618512e-05, + "loss": 2.5421, + "step": 56900 + }, + { + "epoch": 114.3, + "learning_rate": 6.778494605524151e-05, + "loss": 2.5229, + "step": 56920 + }, + { + "epoch": 114.34, + "learning_rate": 6.774299411962203e-05, + "loss": 2.5641, + "step": 56940 + }, + { + "epoch": 114.38, + "learning_rate": 6.770102788879267e-05, + "loss": 2.5578, + "step": 56960 + }, + { + "epoch": 114.42, + "learning_rate": 6.765904739656486e-05, + "loss": 2.5723, + "step": 56980 + }, + { + "epoch": 114.46, + "learning_rate": 6.761705267676153e-05, + "loss": 2.5659, + "step": 57000 + }, + { + "epoch": 114.46, + "eval_loss": 2.9935545921325684, + "eval_runtime": 32.44, + "eval_samples_per_second": 23.613, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.0036283097003805493, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0004556364939408191, + "step": 57000 + }, + { + "epoch": 114.5, + "learning_rate": 6.757504376321704e-05, + "loss": 2.565, + "step": 57020 + }, + { + "epoch": 114.54, + "learning_rate": 6.753302068977725e-05, + "loss": 2.5688, + "step": 57040 + }, + { + "epoch": 114.58, + "learning_rate": 6.749098349029935e-05, + "loss": 2.5619, + "step": 57060 + }, + { + "epoch": 114.62, + "learning_rate": 6.7448932198652e-05, + "loss": 2.5811, + "step": 57080 + }, + { + "epoch": 114.66, + "learning_rate": 6.740686684871515e-05, + "loss": 2.5671, + "step": 57100 + }, + { + "epoch": 114.7, + "learning_rate": 6.736478747438007e-05, + "loss": 2.574, + "step": 57120 + }, + { + "epoch": 114.74, + "learning_rate": 6.732269410954938e-05, + "loss": 2.5673, + "step": 57140 + }, + { + "epoch": 114.78, + "learning_rate": 6.728058678813694e-05, + "loss": 2.5675, + "step": 57160 + }, + { + "epoch": 114.82, + "learning_rate": 6.723846554406782e-05, + "loss": 2.5864, + "step": 57180 + }, + { + "epoch": 114.86, + "learning_rate": 6.719633041127839e-05, + "loss": 2.5901, + "step": 57200 + }, + { + "epoch": 114.9, + "learning_rate": 6.715418142371614e-05, + "loss": 2.5878, + "step": 57220 + }, + { + "epoch": 114.94, + "learning_rate": 6.711201861533978e-05, + "loss": 2.5881, + "step": 57240 + }, + { + "epoch": 114.98, + "learning_rate": 6.70698420201191e-05, + "loss": 2.5954, + "step": 57260 + }, + { + "epoch": 115.02, + "learning_rate": 6.7027651672035e-05, + "loss": 2.5349, + "step": 57280 + }, + { + "epoch": 115.06, + "learning_rate": 6.698544760507952e-05, + "loss": 2.5162, + "step": 57300 + }, + { + "epoch": 115.1, + "learning_rate": 6.694322985325569e-05, + "loss": 2.5178, + "step": 57320 + }, + { + "epoch": 115.14, + "learning_rate": 6.69009984505776e-05, + "loss": 2.525, + "step": 57340 + }, + { + "epoch": 115.18, + "learning_rate": 6.685875343107033e-05, + "loss": 2.5293, + "step": 57360 + }, + { + "epoch": 115.22, + "learning_rate": 6.681649482876994e-05, + "loss": 2.5162, + "step": 57380 + }, + { + "epoch": 115.26, + "learning_rate": 6.677422267772338e-05, + "loss": 2.5461, + "step": 57400 + }, + { + "epoch": 115.3, + "learning_rate": 6.673193701198862e-05, + "loss": 2.5549, + "step": 57420 + }, + { + "epoch": 115.34, + "learning_rate": 6.66896378656344e-05, + "loss": 2.5307, + "step": 57440 + }, + { + "epoch": 115.38, + "learning_rate": 6.664732527274041e-05, + "loss": 2.5618, + "step": 57460 + }, + { + "epoch": 115.42, + "learning_rate": 6.660499926739714e-05, + "loss": 2.5577, + "step": 57480 + }, + { + "epoch": 115.46, + "learning_rate": 6.656265988370588e-05, + "loss": 2.5493, + "step": 57500 + }, + { + "epoch": 115.5, + "learning_rate": 6.652030715577871e-05, + "loss": 2.557, + "step": 57520 + }, + { + "epoch": 115.54, + "learning_rate": 6.647794111773843e-05, + "loss": 2.5737, + "step": 57540 + }, + { + "epoch": 115.58, + "learning_rate": 6.643556180371866e-05, + "loss": 2.5586, + "step": 57560 + }, + { + "epoch": 115.62, + "learning_rate": 6.63931692478636e-05, + "loss": 2.5556, + "step": 57580 + }, + { + "epoch": 115.66, + "learning_rate": 6.635076348432815e-05, + "loss": 2.5495, + "step": 57600 + }, + { + "epoch": 115.7, + "learning_rate": 6.630834454727792e-05, + "loss": 2.5716, + "step": 57620 + }, + { + "epoch": 115.74, + "learning_rate": 6.626591247088903e-05, + "loss": 2.5772, + "step": 57640 + }, + { + "epoch": 115.78, + "learning_rate": 6.622346728934827e-05, + "loss": 2.5658, + "step": 57660 + }, + { + "epoch": 115.82, + "learning_rate": 6.618100903685294e-05, + "loss": 2.5578, + "step": 57680 + }, + { + "epoch": 115.86, + "learning_rate": 6.61385377476109e-05, + "loss": 2.5878, + "step": 57700 + }, + { + "epoch": 115.9, + "learning_rate": 6.609605345584047e-05, + "loss": 2.5771, + "step": 57720 + }, + { + "epoch": 115.94, + "learning_rate": 6.605355619577054e-05, + "loss": 2.562, + "step": 57740 + }, + { + "epoch": 115.98, + "learning_rate": 6.601104600164032e-05, + "loss": 2.5757, + "step": 57760 + }, + { + "epoch": 116.02, + "learning_rate": 6.596852290769952e-05, + "loss": 2.5328, + "step": 57780 + }, + { + "epoch": 116.06, + "learning_rate": 6.592598694820826e-05, + "loss": 2.4876, + "step": 57800 + }, + { + "epoch": 116.1, + "learning_rate": 6.588343815743697e-05, + "loss": 2.5086, + "step": 57820 + }, + { + "epoch": 116.14, + "learning_rate": 6.584087656966644e-05, + "loss": 2.5124, + "step": 57840 + }, + { + "epoch": 116.18, + "learning_rate": 6.57983022191878e-05, + "loss": 2.5218, + "step": 57860 + }, + { + "epoch": 116.22, + "learning_rate": 6.575571514030239e-05, + "loss": 2.5238, + "step": 57880 + }, + { + "epoch": 116.27, + "learning_rate": 6.571311536732188e-05, + "loss": 2.5309, + "step": 57900 + }, + { + "epoch": 116.31, + "learning_rate": 6.567050293456812e-05, + "loss": 2.514, + "step": 57920 + }, + { + "epoch": 116.35, + "learning_rate": 6.562787787637321e-05, + "loss": 2.5411, + "step": 57940 + }, + { + "epoch": 116.39, + "learning_rate": 6.558524022707935e-05, + "loss": 2.5352, + "step": 57960 + }, + { + "epoch": 116.43, + "learning_rate": 6.554259002103895e-05, + "loss": 2.5363, + "step": 57980 + }, + { + "epoch": 116.47, + "learning_rate": 6.549992729261451e-05, + "loss": 2.5513, + "step": 58000 + }, + { + "epoch": 116.47, + "eval_loss": 2.9885976314544678, + "eval_runtime": 34.5765, + "eval_samples_per_second": 22.154, + "eval_steps_per_second": 0.116, + "eval_tse_ndup": 0.00370365409424138, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0001994487960545402, + "step": 58000 + }, + { + "epoch": 116.51, + "learning_rate": 6.54572520761786e-05, + "loss": 2.5457, + "step": 58020 + }, + { + "epoch": 116.55, + "learning_rate": 6.54145644061139e-05, + "loss": 2.5561, + "step": 58040 + }, + { + "epoch": 116.59, + "learning_rate": 6.537186431681306e-05, + "loss": 2.5516, + "step": 58060 + }, + { + "epoch": 116.63, + "learning_rate": 6.532915184267881e-05, + "loss": 2.5569, + "step": 58080 + }, + { + "epoch": 116.67, + "learning_rate": 6.528642701812378e-05, + "loss": 2.5603, + "step": 58100 + }, + { + "epoch": 116.71, + "learning_rate": 6.524368987757061e-05, + "loss": 2.5712, + "step": 58120 + }, + { + "epoch": 116.75, + "learning_rate": 6.520094045545184e-05, + "loss": 2.5621, + "step": 58140 + }, + { + "epoch": 116.79, + "learning_rate": 6.515817878620992e-05, + "loss": 2.5609, + "step": 58160 + }, + { + "epoch": 116.83, + "learning_rate": 6.511540490429713e-05, + "loss": 2.5712, + "step": 58180 + }, + { + "epoch": 116.87, + "learning_rate": 6.507261884417561e-05, + "loss": 2.575, + "step": 58200 + }, + { + "epoch": 116.91, + "learning_rate": 6.502982064031736e-05, + "loss": 2.5711, + "step": 58220 + }, + { + "epoch": 116.95, + "learning_rate": 6.498701032720406e-05, + "loss": 2.5685, + "step": 58240 + }, + { + "epoch": 116.99, + "learning_rate": 6.494632934496418e-05, + "loss": 2.5872, + "step": 58260 + }, + { + "epoch": 117.03, + "learning_rate": 6.490349551801864e-05, + "loss": 2.513, + "step": 58280 + }, + { + "epoch": 117.07, + "learning_rate": 6.486064968359596e-05, + "loss": 2.4979, + "step": 58300 + }, + { + "epoch": 117.11, + "learning_rate": 6.481779187621621e-05, + "loss": 2.4869, + "step": 58320 + }, + { + "epoch": 117.15, + "learning_rate": 6.477492213040924e-05, + "loss": 2.5071, + "step": 58340 + }, + { + "epoch": 117.19, + "learning_rate": 6.473204048071432e-05, + "loss": 2.5143, + "step": 58360 + }, + { + "epoch": 117.23, + "learning_rate": 6.468914696168053e-05, + "loss": 2.5129, + "step": 58380 + }, + { + "epoch": 117.27, + "learning_rate": 6.464624160786633e-05, + "loss": 2.514, + "step": 58400 + }, + { + "epoch": 117.31, + "learning_rate": 6.46033244538398e-05, + "loss": 2.5215, + "step": 58420 + }, + { + "epoch": 117.35, + "learning_rate": 6.456039553417852e-05, + "loss": 2.5299, + "step": 58440 + }, + { + "epoch": 117.39, + "learning_rate": 6.451745488346956e-05, + "loss": 2.5158, + "step": 58460 + }, + { + "epoch": 117.43, + "learning_rate": 6.447450253630942e-05, + "loss": 2.5254, + "step": 58480 + }, + { + "epoch": 117.47, + "learning_rate": 6.443153852730404e-05, + "loss": 2.5456, + "step": 58500 + }, + { + "epoch": 117.51, + "learning_rate": 6.438856289106871e-05, + "loss": 2.5523, + "step": 58520 + }, + { + "epoch": 117.55, + "learning_rate": 6.434557566222817e-05, + "loss": 2.5585, + "step": 58540 + }, + { + "epoch": 117.59, + "learning_rate": 6.430257687541641e-05, + "loss": 2.5347, + "step": 58560 + }, + { + "epoch": 117.63, + "learning_rate": 6.425956656527681e-05, + "loss": 2.5513, + "step": 58580 + }, + { + "epoch": 117.67, + "learning_rate": 6.421654476646199e-05, + "loss": 2.5718, + "step": 58600 + }, + { + "epoch": 117.71, + "learning_rate": 6.417351151363381e-05, + "loss": 2.5459, + "step": 58620 + }, + { + "epoch": 117.75, + "learning_rate": 6.413046684146343e-05, + "loss": 2.5403, + "step": 58640 + }, + { + "epoch": 117.79, + "learning_rate": 6.408741078463114e-05, + "loss": 2.5517, + "step": 58660 + }, + { + "epoch": 117.83, + "learning_rate": 6.404434337782642e-05, + "loss": 2.5629, + "step": 58680 + }, + { + "epoch": 117.87, + "learning_rate": 6.400126465574793e-05, + "loss": 2.5534, + "step": 58700 + }, + { + "epoch": 117.91, + "learning_rate": 6.395817465310338e-05, + "loss": 2.583, + "step": 58720 + }, + { + "epoch": 117.95, + "learning_rate": 6.391507340460963e-05, + "loss": 2.5778, + "step": 58740 + }, + { + "epoch": 117.99, + "learning_rate": 6.387196094499258e-05, + "loss": 2.5657, + "step": 58760 + }, + { + "epoch": 118.03, + "learning_rate": 6.382883730898717e-05, + "loss": 2.5176, + "step": 58780 + }, + { + "epoch": 118.07, + "learning_rate": 6.378570253133729e-05, + "loss": 2.4915, + "step": 58800 + }, + { + "epoch": 118.11, + "learning_rate": 6.37425566467959e-05, + "loss": 2.4959, + "step": 58820 + }, + { + "epoch": 118.15, + "learning_rate": 6.369939969012483e-05, + "loss": 2.4895, + "step": 58840 + }, + { + "epoch": 118.19, + "learning_rate": 6.365623169609487e-05, + "loss": 2.5042, + "step": 58860 + }, + { + "epoch": 118.23, + "learning_rate": 6.36130526994857e-05, + "loss": 2.5194, + "step": 58880 + }, + { + "epoch": 118.27, + "learning_rate": 6.356986273508585e-05, + "loss": 2.5007, + "step": 58900 + }, + { + "epoch": 118.31, + "learning_rate": 6.35266618376927e-05, + "loss": 2.4949, + "step": 58920 + }, + { + "epoch": 118.35, + "learning_rate": 6.34834500421124e-05, + "loss": 2.5166, + "step": 58940 + }, + { + "epoch": 118.39, + "learning_rate": 6.344022738315998e-05, + "loss": 2.5188, + "step": 58960 + }, + { + "epoch": 118.43, + "learning_rate": 6.33969938956591e-05, + "loss": 2.5408, + "step": 58980 + }, + { + "epoch": 118.47, + "learning_rate": 6.335374961444222e-05, + "loss": 2.5447, + "step": 59000 + }, + { + "epoch": 118.47, + "eval_loss": 2.9815523624420166, + "eval_runtime": 32.3949, + "eval_samples_per_second": 23.646, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004175785799212398, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0001790506237307804, + "step": 59000 + }, + { + "epoch": 118.51, + "learning_rate": 6.331049457435046e-05, + "loss": 2.55, + "step": 59020 + }, + { + "epoch": 118.55, + "learning_rate": 6.326722881023366e-05, + "loss": 2.5327, + "step": 59040 + }, + { + "epoch": 118.59, + "learning_rate": 6.322395235695022e-05, + "loss": 2.5437, + "step": 59060 + }, + { + "epoch": 118.63, + "learning_rate": 6.318066524936725e-05, + "loss": 2.5223, + "step": 59080 + }, + { + "epoch": 118.67, + "learning_rate": 6.313736752236033e-05, + "loss": 2.5381, + "step": 59100 + }, + { + "epoch": 118.71, + "learning_rate": 6.309405921081369e-05, + "loss": 2.555, + "step": 59120 + }, + { + "epoch": 118.76, + "learning_rate": 6.305074034962006e-05, + "loss": 2.5344, + "step": 59140 + }, + { + "epoch": 118.8, + "learning_rate": 6.300741097368065e-05, + "loss": 2.5587, + "step": 59160 + }, + { + "epoch": 118.84, + "learning_rate": 6.296407111790516e-05, + "loss": 2.5388, + "step": 59180 + }, + { + "epoch": 118.88, + "learning_rate": 6.292072081721173e-05, + "loss": 2.5499, + "step": 59200 + }, + { + "epoch": 118.92, + "learning_rate": 6.287736010652693e-05, + "loss": 2.5678, + "step": 59220 + }, + { + "epoch": 118.96, + "learning_rate": 6.283398902078567e-05, + "loss": 2.5462, + "step": 59240 + }, + { + "epoch": 119.0, + "learning_rate": 6.279060759493128e-05, + "loss": 2.5675, + "step": 59260 + }, + { + "epoch": 119.04, + "learning_rate": 6.274721586391535e-05, + "loss": 2.4834, + "step": 59280 + }, + { + "epoch": 119.08, + "learning_rate": 6.270381386269786e-05, + "loss": 2.4824, + "step": 59300 + }, + { + "epoch": 119.12, + "learning_rate": 6.266040162624699e-05, + "loss": 2.4747, + "step": 59320 + }, + { + "epoch": 119.16, + "learning_rate": 6.261697918953921e-05, + "loss": 2.4749, + "step": 59340 + }, + { + "epoch": 119.2, + "learning_rate": 6.257354658755919e-05, + "loss": 2.5095, + "step": 59360 + }, + { + "epoch": 119.24, + "learning_rate": 6.253010385529978e-05, + "loss": 2.5068, + "step": 59380 + }, + { + "epoch": 119.28, + "learning_rate": 6.248665102776203e-05, + "loss": 2.5109, + "step": 59400 + }, + { + "epoch": 119.32, + "learning_rate": 6.244318813995507e-05, + "loss": 2.5145, + "step": 59420 + }, + { + "epoch": 119.36, + "learning_rate": 6.239971522689616e-05, + "loss": 2.5225, + "step": 59440 + }, + { + "epoch": 119.4, + "learning_rate": 6.235623232361069e-05, + "loss": 2.5097, + "step": 59460 + }, + { + "epoch": 119.44, + "learning_rate": 6.231273946513201e-05, + "loss": 2.5326, + "step": 59480 + }, + { + "epoch": 119.48, + "learning_rate": 6.226923668650152e-05, + "loss": 2.5221, + "step": 59500 + }, + { + "epoch": 119.52, + "learning_rate": 6.222572402276867e-05, + "loss": 2.5205, + "step": 59520 + }, + { + "epoch": 119.56, + "learning_rate": 6.218220150899076e-05, + "loss": 2.5199, + "step": 59540 + }, + { + "epoch": 119.6, + "learning_rate": 6.213866918023317e-05, + "loss": 2.5261, + "step": 59560 + }, + { + "epoch": 119.64, + "learning_rate": 6.209512707156904e-05, + "loss": 2.5293, + "step": 59580 + }, + { + "epoch": 119.68, + "learning_rate": 6.205157521807951e-05, + "loss": 2.5366, + "step": 59600 + }, + { + "epoch": 119.72, + "learning_rate": 6.20080136548535e-05, + "loss": 2.539, + "step": 59620 + }, + { + "epoch": 119.76, + "learning_rate": 6.196444241698777e-05, + "loss": 2.5221, + "step": 59640 + }, + { + "epoch": 119.8, + "learning_rate": 6.192086153958692e-05, + "loss": 2.5518, + "step": 59660 + }, + { + "epoch": 119.84, + "learning_rate": 6.187727105776321e-05, + "loss": 2.5568, + "step": 59680 + }, + { + "epoch": 119.88, + "learning_rate": 6.183367100663676e-05, + "loss": 2.5508, + "step": 59700 + }, + { + "epoch": 119.92, + "learning_rate": 6.179006142133529e-05, + "loss": 2.5398, + "step": 59720 + }, + { + "epoch": 119.96, + "learning_rate": 6.174644233699433e-05, + "loss": 2.5498, + "step": 59740 + }, + { + "epoch": 120.0, + "learning_rate": 6.170281378875692e-05, + "loss": 2.5758, + "step": 59760 + }, + { + "epoch": 120.04, + "learning_rate": 6.165917581177383e-05, + "loss": 2.4766, + "step": 59780 + }, + { + "epoch": 120.08, + "learning_rate": 6.161552844120336e-05, + "loss": 2.4636, + "step": 59800 + }, + { + "epoch": 120.12, + "learning_rate": 6.157187171221144e-05, + "loss": 2.488, + "step": 59820 + }, + { + "epoch": 120.16, + "learning_rate": 6.15282056599715e-05, + "loss": 2.488, + "step": 59840 + }, + { + "epoch": 120.2, + "learning_rate": 6.148453031966447e-05, + "loss": 2.4918, + "step": 59860 + }, + { + "epoch": 120.24, + "learning_rate": 6.14408457264788e-05, + "loss": 2.5033, + "step": 59880 + }, + { + "epoch": 120.28, + "learning_rate": 6.139715191561038e-05, + "loss": 2.4908, + "step": 59900 + }, + { + "epoch": 120.32, + "learning_rate": 6.135344892226253e-05, + "loss": 2.4964, + "step": 59920 + }, + { + "epoch": 120.36, + "learning_rate": 6.130973678164593e-05, + "loss": 2.5143, + "step": 59940 + }, + { + "epoch": 120.4, + "learning_rate": 6.126601552897869e-05, + "loss": 2.5218, + "step": 59960 + }, + { + "epoch": 120.44, + "learning_rate": 6.122228519948622e-05, + "loss": 2.507, + "step": 59980 + }, + { + "epoch": 120.48, + "learning_rate": 6.117854582840129e-05, + "loss": 2.5177, + "step": 60000 + }, + { + "epoch": 120.48, + "eval_loss": 2.960603713989258, + "eval_runtime": 32.3983, + "eval_samples_per_second": 23.643, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004640649820180373, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 5.892805337975052e-05, + "step": 60000 + }, + { + "epoch": 120.52, + "learning_rate": 6.113479745096387e-05, + "loss": 2.5248, + "step": 60020 + }, + { + "epoch": 120.56, + "learning_rate": 6.109104010242128e-05, + "loss": 2.5389, + "step": 60040 + }, + { + "epoch": 120.6, + "learning_rate": 6.104727381802798e-05, + "loss": 2.5203, + "step": 60060 + }, + { + "epoch": 120.64, + "learning_rate": 6.100349863304573e-05, + "loss": 2.5121, + "step": 60080 + }, + { + "epoch": 120.68, + "learning_rate": 6.0959714582743364e-05, + "loss": 2.5009, + "step": 60100 + }, + { + "epoch": 120.72, + "learning_rate": 6.0915921702396916e-05, + "loss": 2.5362, + "step": 60120 + }, + { + "epoch": 120.76, + "learning_rate": 6.0872120027289536e-05, + "loss": 2.5326, + "step": 60140 + }, + { + "epoch": 120.8, + "learning_rate": 6.08283095927114e-05, + "loss": 2.5355, + "step": 60160 + }, + { + "epoch": 120.84, + "learning_rate": 6.078449043395982e-05, + "loss": 2.5444, + "step": 60180 + }, + { + "epoch": 120.88, + "learning_rate": 6.074066258633908e-05, + "loss": 2.5344, + "step": 60200 + }, + { + "epoch": 120.92, + "learning_rate": 6.0696826085160505e-05, + "loss": 2.5448, + "step": 60220 + }, + { + "epoch": 120.96, + "learning_rate": 6.065298096574235e-05, + "loss": 2.5528, + "step": 60240 + }, + { + "epoch": 121.0, + "learning_rate": 6.060912726340986e-05, + "loss": 2.5241, + "step": 60260 + }, + { + "epoch": 121.04, + "learning_rate": 6.0565265013495144e-05, + "loss": 2.475, + "step": 60280 + }, + { + "epoch": 121.08, + "learning_rate": 6.052358799106528e-05, + "loss": 2.4845, + "step": 60300 + }, + { + "epoch": 121.12, + "learning_rate": 6.047970917501538e-05, + "loss": 2.4635, + "step": 60320 + }, + { + "epoch": 121.16, + "learning_rate": 6.043582191565307e-05, + "loss": 2.4696, + "step": 60340 + }, + { + "epoch": 121.2, + "learning_rate": 6.0391926248337536e-05, + "loss": 2.5, + "step": 60360 + }, + { + "epoch": 121.24, + "learning_rate": 6.0348022208434716e-05, + "loss": 2.5008, + "step": 60380 + }, + { + "epoch": 121.29, + "learning_rate": 6.0304109831317324e-05, + "loss": 2.4893, + "step": 60400 + }, + { + "epoch": 121.33, + "learning_rate": 6.026018915236474e-05, + "loss": 2.5028, + "step": 60420 + }, + { + "epoch": 121.37, + "learning_rate": 6.021626020696311e-05, + "loss": 2.5018, + "step": 60440 + }, + { + "epoch": 121.41, + "learning_rate": 6.017232303050513e-05, + "loss": 2.5005, + "step": 60460 + }, + { + "epoch": 121.45, + "learning_rate": 6.012837765839022e-05, + "loss": 2.5076, + "step": 60480 + }, + { + "epoch": 121.49, + "learning_rate": 6.008442412602438e-05, + "loss": 2.5185, + "step": 60500 + }, + { + "epoch": 121.53, + "learning_rate": 6.004046246882016e-05, + "loss": 2.5083, + "step": 60520 + }, + { + "epoch": 121.57, + "learning_rate": 5.999649272219668e-05, + "loss": 2.5062, + "step": 60540 + }, + { + "epoch": 121.61, + "learning_rate": 5.995251492157958e-05, + "loss": 2.5068, + "step": 60560 + }, + { + "epoch": 121.65, + "learning_rate": 5.990852910240098e-05, + "loss": 2.5248, + "step": 60580 + }, + { + "epoch": 121.69, + "learning_rate": 5.9864535300099435e-05, + "loss": 2.5266, + "step": 60600 + }, + { + "epoch": 121.73, + "learning_rate": 5.9820533550119986e-05, + "loss": 2.5058, + "step": 60620 + }, + { + "epoch": 121.77, + "learning_rate": 5.9776523887914026e-05, + "loss": 2.5166, + "step": 60640 + }, + { + "epoch": 121.81, + "learning_rate": 5.973250634893936e-05, + "loss": 2.5238, + "step": 60660 + }, + { + "epoch": 121.85, + "learning_rate": 5.968848096866011e-05, + "loss": 2.5159, + "step": 60680 + }, + { + "epoch": 121.89, + "learning_rate": 5.964444778254675e-05, + "loss": 2.521, + "step": 60700 + }, + { + "epoch": 121.93, + "learning_rate": 5.9600406826076006e-05, + "loss": 2.5368, + "step": 60720 + }, + { + "epoch": 121.97, + "learning_rate": 5.955635813473089e-05, + "loss": 2.5318, + "step": 60740 + }, + { + "epoch": 122.01, + "learning_rate": 5.951230174400063e-05, + "loss": 2.5324, + "step": 60760 + }, + { + "epoch": 122.05, + "learning_rate": 5.946823768938067e-05, + "loss": 2.4583, + "step": 60780 + }, + { + "epoch": 122.09, + "learning_rate": 5.9424166006372615e-05, + "loss": 2.4532, + "step": 60800 + }, + { + "epoch": 122.13, + "learning_rate": 5.938008673048423e-05, + "loss": 2.4811, + "step": 60820 + }, + { + "epoch": 122.17, + "learning_rate": 5.93359998972294e-05, + "loss": 2.474, + "step": 60840 + }, + { + "epoch": 122.21, + "learning_rate": 5.929190554212807e-05, + "loss": 2.4841, + "step": 60860 + }, + { + "epoch": 122.25, + "learning_rate": 5.924780370070629e-05, + "loss": 2.4771, + "step": 60880 + }, + { + "epoch": 122.29, + "learning_rate": 5.920369440849609e-05, + "loss": 2.4714, + "step": 60900 + }, + { + "epoch": 122.33, + "learning_rate": 5.915957770103556e-05, + "loss": 2.5058, + "step": 60920 + }, + { + "epoch": 122.37, + "learning_rate": 5.91154536138687e-05, + "loss": 2.4943, + "step": 60940 + }, + { + "epoch": 122.41, + "learning_rate": 5.907132218254552e-05, + "loss": 2.4888, + "step": 60960 + }, + { + "epoch": 122.45, + "learning_rate": 5.90271834426219e-05, + "loss": 2.5121, + "step": 60980 + }, + { + "epoch": 122.49, + "learning_rate": 5.898303742965964e-05, + "loss": 2.4846, + "step": 61000 + }, + { + "epoch": 122.49, + "eval_loss": 2.9568583965301514, + "eval_runtime": 34.1589, + "eval_samples_per_second": 22.425, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.004200177034607833, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0002901073397156948, + "step": 61000 + }, + { + "epoch": 122.53, + "learning_rate": 5.893888417922636e-05, + "loss": 2.5009, + "step": 61020 + }, + { + "epoch": 122.57, + "learning_rate": 5.889472372689556e-05, + "loss": 2.4972, + "step": 61040 + }, + { + "epoch": 122.61, + "learning_rate": 5.885055610824652e-05, + "loss": 2.5171, + "step": 61060 + }, + { + "epoch": 122.65, + "learning_rate": 5.880638135886427e-05, + "loss": 2.5109, + "step": 61080 + }, + { + "epoch": 122.69, + "learning_rate": 5.8762199514339624e-05, + "loss": 2.5152, + "step": 61100 + }, + { + "epoch": 122.73, + "learning_rate": 5.8718010610269095e-05, + "loss": 2.5106, + "step": 61120 + }, + { + "epoch": 122.77, + "learning_rate": 5.867381468225489e-05, + "loss": 2.521, + "step": 61140 + }, + { + "epoch": 122.81, + "learning_rate": 5.862961176590486e-05, + "loss": 2.5061, + "step": 61160 + }, + { + "epoch": 122.85, + "learning_rate": 5.85854018968325e-05, + "loss": 2.536, + "step": 61180 + }, + { + "epoch": 122.89, + "learning_rate": 5.8541185110656906e-05, + "loss": 2.5416, + "step": 61200 + }, + { + "epoch": 122.93, + "learning_rate": 5.849696144300273e-05, + "loss": 2.5263, + "step": 61220 + }, + { + "epoch": 122.97, + "learning_rate": 5.845273092950019e-05, + "loss": 2.5269, + "step": 61240 + }, + { + "epoch": 123.01, + "learning_rate": 5.840849360578503e-05, + "loss": 2.5008, + "step": 61260 + }, + { + "epoch": 123.05, + "learning_rate": 5.8364249507498435e-05, + "loss": 2.4614, + "step": 61280 + }, + { + "epoch": 123.09, + "learning_rate": 5.83199986702871e-05, + "loss": 2.4572, + "step": 61300 + }, + { + "epoch": 123.13, + "learning_rate": 5.827574112980311e-05, + "loss": 2.4617, + "step": 61320 + }, + { + "epoch": 123.17, + "learning_rate": 5.8231476921703964e-05, + "loss": 2.4678, + "step": 61340 + }, + { + "epoch": 123.21, + "learning_rate": 5.8187206081652554e-05, + "loss": 2.4724, + "step": 61360 + }, + { + "epoch": 123.25, + "learning_rate": 5.814292864531705e-05, + "loss": 2.4653, + "step": 61380 + }, + { + "epoch": 123.29, + "learning_rate": 5.8098644648371046e-05, + "loss": 2.4765, + "step": 61400 + }, + { + "epoch": 123.33, + "learning_rate": 5.8054354126493324e-05, + "loss": 2.4781, + "step": 61420 + }, + { + "epoch": 123.37, + "learning_rate": 5.8010057115367966e-05, + "loss": 2.4844, + "step": 61440 + }, + { + "epoch": 123.41, + "learning_rate": 5.796575365068425e-05, + "loss": 2.4895, + "step": 61460 + }, + { + "epoch": 123.45, + "learning_rate": 5.792144376813671e-05, + "loss": 2.4932, + "step": 61480 + }, + { + "epoch": 123.49, + "learning_rate": 5.787712750342501e-05, + "loss": 2.4866, + "step": 61500 + }, + { + "epoch": 123.53, + "learning_rate": 5.7832804892253946e-05, + "loss": 2.4936, + "step": 61520 + }, + { + "epoch": 123.57, + "learning_rate": 5.7788475970333455e-05, + "loss": 2.4947, + "step": 61540 + }, + { + "epoch": 123.61, + "learning_rate": 5.774414077337855e-05, + "loss": 2.4937, + "step": 61560 + }, + { + "epoch": 123.65, + "learning_rate": 5.769979933710932e-05, + "loss": 2.5189, + "step": 61580 + }, + { + "epoch": 123.69, + "learning_rate": 5.765545169725083e-05, + "loss": 2.5022, + "step": 61600 + }, + { + "epoch": 123.73, + "learning_rate": 5.7611097889533184e-05, + "loss": 2.5097, + "step": 61620 + }, + { + "epoch": 123.78, + "learning_rate": 5.756673794969143e-05, + "loss": 2.5088, + "step": 61640 + }, + { + "epoch": 123.82, + "learning_rate": 5.752237191346559e-05, + "loss": 2.5255, + "step": 61660 + }, + { + "epoch": 123.86, + "learning_rate": 5.7477999816600556e-05, + "loss": 2.5244, + "step": 61680 + }, + { + "epoch": 123.9, + "learning_rate": 5.743362169484616e-05, + "loss": 2.5254, + "step": 61700 + }, + { + "epoch": 123.94, + "learning_rate": 5.738923758395701e-05, + "loss": 2.5185, + "step": 61720 + }, + { + "epoch": 123.98, + "learning_rate": 5.734484751969262e-05, + "loss": 2.5263, + "step": 61740 + }, + { + "epoch": 124.02, + "learning_rate": 5.7300451537817245e-05, + "loss": 2.482, + "step": 61760 + }, + { + "epoch": 124.06, + "learning_rate": 5.725604967409992e-05, + "loss": 2.458, + "step": 61780 + }, + { + "epoch": 124.1, + "learning_rate": 5.7211641964314456e-05, + "loss": 2.4602, + "step": 61800 + }, + { + "epoch": 124.14, + "learning_rate": 5.7167228444239284e-05, + "loss": 2.4581, + "step": 61820 + }, + { + "epoch": 124.18, + "learning_rate": 5.712280914965764e-05, + "loss": 2.4812, + "step": 61840 + }, + { + "epoch": 124.22, + "learning_rate": 5.70783841163573e-05, + "loss": 2.4524, + "step": 61860 + }, + { + "epoch": 124.26, + "learning_rate": 5.7033953380130736e-05, + "loss": 2.4718, + "step": 61880 + }, + { + "epoch": 124.3, + "learning_rate": 5.698951697677498e-05, + "loss": 2.463, + "step": 61900 + }, + { + "epoch": 124.34, + "learning_rate": 5.694507494209165e-05, + "loss": 2.4523, + "step": 61920 + }, + { + "epoch": 124.38, + "learning_rate": 5.6900627311886865e-05, + "loss": 2.487, + "step": 61940 + }, + { + "epoch": 124.42, + "learning_rate": 5.68561741219713e-05, + "loss": 2.483, + "step": 61960 + }, + { + "epoch": 124.46, + "learning_rate": 5.681171540816008e-05, + "loss": 2.499, + "step": 61980 + }, + { + "epoch": 124.5, + "learning_rate": 5.67672512062728e-05, + "loss": 2.4804, + "step": 62000 + }, + { + "epoch": 124.5, + "eval_loss": 2.9488375186920166, + "eval_runtime": 34.0535, + "eval_samples_per_second": 22.494, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.0039627047898703254, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00016556285565963018, + "step": 62000 + }, + { + "epoch": 124.54, + "learning_rate": 5.6722781552133466e-05, + "loss": 2.5065, + "step": 62020 + }, + { + "epoch": 124.58, + "learning_rate": 5.6678306481570444e-05, + "loss": 2.5035, + "step": 62040 + }, + { + "epoch": 124.62, + "learning_rate": 5.663382603041652e-05, + "loss": 2.4815, + "step": 62060 + }, + { + "epoch": 124.66, + "learning_rate": 5.658934023450878e-05, + "loss": 2.4871, + "step": 62080 + }, + { + "epoch": 124.7, + "learning_rate": 5.654484912968863e-05, + "loss": 2.4975, + "step": 62100 + }, + { + "epoch": 124.74, + "learning_rate": 5.650035275180175e-05, + "loss": 2.511, + "step": 62120 + }, + { + "epoch": 124.78, + "learning_rate": 5.645585113669808e-05, + "loss": 2.5124, + "step": 62140 + }, + { + "epoch": 124.82, + "learning_rate": 5.641134432023174e-05, + "loss": 2.5035, + "step": 62160 + }, + { + "epoch": 124.86, + "learning_rate": 5.636683233826111e-05, + "loss": 2.4982, + "step": 62180 + }, + { + "epoch": 124.9, + "learning_rate": 5.632231522664864e-05, + "loss": 2.5121, + "step": 62200 + }, + { + "epoch": 124.94, + "learning_rate": 5.6277793021260994e-05, + "loss": 2.5043, + "step": 62220 + }, + { + "epoch": 124.98, + "learning_rate": 5.62332657579689e-05, + "loss": 2.5194, + "step": 62240 + }, + { + "epoch": 125.02, + "learning_rate": 5.618873347264716e-05, + "loss": 2.4592, + "step": 62260 + }, + { + "epoch": 125.06, + "learning_rate": 5.614419620117466e-05, + "loss": 2.4231, + "step": 62280 + }, + { + "epoch": 125.1, + "learning_rate": 5.609965397943425e-05, + "loss": 2.4355, + "step": 62300 + }, + { + "epoch": 125.14, + "learning_rate": 5.605733431628139e-05, + "loss": 2.4585, + "step": 62320 + }, + { + "epoch": 125.18, + "learning_rate": 5.601278254474176e-05, + "loss": 2.4406, + "step": 62340 + }, + { + "epoch": 125.22, + "learning_rate": 5.5968225928811856e-05, + "loss": 2.4616, + "step": 62360 + }, + { + "epoch": 125.26, + "learning_rate": 5.592366450439012e-05, + "loss": 2.4619, + "step": 62380 + }, + { + "epoch": 125.3, + "learning_rate": 5.587909830737887e-05, + "loss": 2.4694, + "step": 62400 + }, + { + "epoch": 125.34, + "learning_rate": 5.583452737368432e-05, + "loss": 2.4778, + "step": 62420 + }, + { + "epoch": 125.38, + "learning_rate": 5.578995173921644e-05, + "loss": 2.4568, + "step": 62440 + }, + { + "epoch": 125.42, + "learning_rate": 5.574537143988903e-05, + "loss": 2.474, + "step": 62460 + }, + { + "epoch": 125.46, + "learning_rate": 5.5700786511619605e-05, + "loss": 2.4869, + "step": 62480 + }, + { + "epoch": 125.5, + "learning_rate": 5.565619699032949e-05, + "loss": 2.4865, + "step": 62500 + }, + { + "epoch": 125.54, + "learning_rate": 5.5611602911943596e-05, + "loss": 2.4928, + "step": 62520 + }, + { + "epoch": 125.58, + "learning_rate": 5.556700431239061e-05, + "loss": 2.4874, + "step": 62540 + }, + { + "epoch": 125.62, + "learning_rate": 5.552240122760279e-05, + "loss": 2.4936, + "step": 62560 + }, + { + "epoch": 125.66, + "learning_rate": 5.547779369351606e-05, + "loss": 2.4949, + "step": 62580 + }, + { + "epoch": 125.7, + "learning_rate": 5.543318174606987e-05, + "loss": 2.4952, + "step": 62600 + }, + { + "epoch": 125.74, + "learning_rate": 5.5388565421207304e-05, + "loss": 2.4955, + "step": 62620 + }, + { + "epoch": 125.78, + "learning_rate": 5.53439447548749e-05, + "loss": 2.5114, + "step": 62640 + }, + { + "epoch": 125.82, + "learning_rate": 5.529931978302272e-05, + "loss": 2.4929, + "step": 62660 + }, + { + "epoch": 125.86, + "learning_rate": 5.525469054160433e-05, + "loss": 2.5028, + "step": 62680 + }, + { + "epoch": 125.9, + "learning_rate": 5.521005706657666e-05, + "loss": 2.5052, + "step": 62700 + }, + { + "epoch": 125.94, + "learning_rate": 5.516541939390013e-05, + "loss": 2.5068, + "step": 62720 + }, + { + "epoch": 125.98, + "learning_rate": 5.5120777559538486e-05, + "loss": 2.4928, + "step": 62740 + }, + { + "epoch": 126.02, + "learning_rate": 5.5076131599458855e-05, + "loss": 2.4559, + "step": 62760 + }, + { + "epoch": 126.06, + "learning_rate": 5.503148154963168e-05, + "loss": 2.4301, + "step": 62780 + }, + { + "epoch": 126.1, + "learning_rate": 5.498682744603071e-05, + "loss": 2.4323, + "step": 62800 + }, + { + "epoch": 126.14, + "learning_rate": 5.494216932463293e-05, + "loss": 2.4373, + "step": 62820 + }, + { + "epoch": 126.18, + "learning_rate": 5.48975072214186e-05, + "loss": 2.457, + "step": 62840 + }, + { + "epoch": 126.22, + "learning_rate": 5.485284117237115e-05, + "loss": 2.4498, + "step": 62860 + }, + { + "epoch": 126.27, + "learning_rate": 5.480817121347722e-05, + "loss": 2.4557, + "step": 62880 + }, + { + "epoch": 126.31, + "learning_rate": 5.476349738072659e-05, + "loss": 2.4688, + "step": 62900 + }, + { + "epoch": 126.35, + "learning_rate": 5.4718819710112166e-05, + "loss": 2.4631, + "step": 62920 + }, + { + "epoch": 126.39, + "learning_rate": 5.467413823762993e-05, + "loss": 2.4615, + "step": 62940 + }, + { + "epoch": 126.43, + "learning_rate": 5.462945299927894e-05, + "loss": 2.4795, + "step": 62960 + }, + { + "epoch": 126.47, + "learning_rate": 5.4584764031061296e-05, + "loss": 2.4542, + "step": 62980 + }, + { + "epoch": 126.51, + "learning_rate": 5.454007136898207e-05, + "loss": 2.4898, + "step": 63000 + }, + { + "epoch": 126.51, + "eval_loss": 2.939746618270874, + "eval_runtime": 34.2651, + "eval_samples_per_second": 22.355, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.0033324132567144477, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0001745176965477227, + "step": 63000 + }, + { + "epoch": 126.55, + "learning_rate": 5.4495375049049366e-05, + "loss": 2.4848, + "step": 63020 + }, + { + "epoch": 126.59, + "learning_rate": 5.445067510727415e-05, + "loss": 2.4898, + "step": 63040 + }, + { + "epoch": 126.63, + "learning_rate": 5.4405971579670414e-05, + "loss": 2.4891, + "step": 63060 + }, + { + "epoch": 126.67, + "learning_rate": 5.436126450225495e-05, + "loss": 2.4742, + "step": 63080 + }, + { + "epoch": 126.71, + "learning_rate": 5.4316553911047455e-05, + "loss": 2.4766, + "step": 63100 + }, + { + "epoch": 126.75, + "learning_rate": 5.427183984207043e-05, + "loss": 2.4827, + "step": 63120 + }, + { + "epoch": 126.79, + "learning_rate": 5.42271223313492e-05, + "loss": 2.5016, + "step": 63140 + }, + { + "epoch": 126.83, + "learning_rate": 5.4182401414911873e-05, + "loss": 2.4881, + "step": 63160 + }, + { + "epoch": 126.87, + "learning_rate": 5.413767712878926e-05, + "loss": 2.492, + "step": 63180 + }, + { + "epoch": 126.91, + "learning_rate": 5.409294950901492e-05, + "loss": 2.5071, + "step": 63200 + }, + { + "epoch": 126.95, + "learning_rate": 5.404821859162509e-05, + "loss": 2.4969, + "step": 63220 + }, + { + "epoch": 126.99, + "learning_rate": 5.4003484412658665e-05, + "loss": 2.5037, + "step": 63240 + }, + { + "epoch": 127.03, + "learning_rate": 5.395874700815715e-05, + "loss": 2.4401, + "step": 63260 + }, + { + "epoch": 127.07, + "learning_rate": 5.39140064141647e-05, + "loss": 2.4364, + "step": 63280 + }, + { + "epoch": 127.11, + "learning_rate": 5.386926266672795e-05, + "loss": 2.4551, + "step": 63300 + }, + { + "epoch": 127.15, + "learning_rate": 5.382451580189616e-05, + "loss": 2.4204, + "step": 63320 + }, + { + "epoch": 127.19, + "learning_rate": 5.3779765855721066e-05, + "loss": 2.4476, + "step": 63340 + }, + { + "epoch": 127.23, + "learning_rate": 5.373501286425691e-05, + "loss": 2.4449, + "step": 63360 + }, + { + "epoch": 127.27, + "learning_rate": 5.369025686356033e-05, + "loss": 2.4396, + "step": 63380 + }, + { + "epoch": 127.31, + "learning_rate": 5.364549788969045e-05, + "loss": 2.4501, + "step": 63400 + }, + { + "epoch": 127.35, + "learning_rate": 5.360073597870879e-05, + "loss": 2.4574, + "step": 63420 + }, + { + "epoch": 127.39, + "learning_rate": 5.355597116667918e-05, + "loss": 2.4554, + "step": 63440 + }, + { + "epoch": 127.43, + "learning_rate": 5.351120348966783e-05, + "loss": 2.4614, + "step": 63460 + }, + { + "epoch": 127.47, + "learning_rate": 5.346643298374322e-05, + "loss": 2.4655, + "step": 63480 + }, + { + "epoch": 127.51, + "learning_rate": 5.3421659684976197e-05, + "loss": 2.4607, + "step": 63500 + }, + { + "epoch": 127.55, + "learning_rate": 5.337688362943976e-05, + "loss": 2.462, + "step": 63520 + }, + { + "epoch": 127.59, + "learning_rate": 5.333210485320919e-05, + "loss": 2.4748, + "step": 63540 + }, + { + "epoch": 127.63, + "learning_rate": 5.32873233923619e-05, + "loss": 2.4913, + "step": 63560 + }, + { + "epoch": 127.67, + "learning_rate": 5.3242539282977543e-05, + "loss": 2.4698, + "step": 63580 + }, + { + "epoch": 127.71, + "learning_rate": 5.319775256113785e-05, + "loss": 2.4961, + "step": 63600 + }, + { + "epoch": 127.75, + "learning_rate": 5.315296326292667e-05, + "loss": 2.4782, + "step": 63620 + }, + { + "epoch": 127.79, + "learning_rate": 5.310817142442994e-05, + "loss": 2.475, + "step": 63640 + }, + { + "epoch": 127.83, + "learning_rate": 5.3063377081735645e-05, + "loss": 2.5018, + "step": 63660 + }, + { + "epoch": 127.87, + "learning_rate": 5.301858027093376e-05, + "loss": 2.4948, + "step": 63680 + }, + { + "epoch": 127.91, + "learning_rate": 5.2973781028116275e-05, + "loss": 2.492, + "step": 63700 + }, + { + "epoch": 127.95, + "learning_rate": 5.292897938937713e-05, + "loss": 2.501, + "step": 63720 + }, + { + "epoch": 127.99, + "learning_rate": 5.2884175390812195e-05, + "loss": 2.4868, + "step": 63740 + }, + { + "epoch": 128.03, + "learning_rate": 5.283936906851923e-05, + "loss": 2.4378, + "step": 63760 + }, + { + "epoch": 128.07, + "learning_rate": 5.2794560458597897e-05, + "loss": 2.4196, + "step": 63780 + }, + { + "epoch": 128.11, + "learning_rate": 5.274974959714969e-05, + "loss": 2.4219, + "step": 63800 + }, + { + "epoch": 128.15, + "learning_rate": 5.2704936520277895e-05, + "loss": 2.4452, + "step": 63820 + }, + { + "epoch": 128.19, + "learning_rate": 5.2660121264087605e-05, + "loss": 2.4384, + "step": 63840 + }, + { + "epoch": 128.23, + "learning_rate": 5.261530386468565e-05, + "loss": 2.4419, + "step": 63860 + }, + { + "epoch": 128.27, + "learning_rate": 5.2570484358180605e-05, + "loss": 2.454, + "step": 63880 + }, + { + "epoch": 128.31, + "learning_rate": 5.252566278068275e-05, + "loss": 2.4621, + "step": 63900 + }, + { + "epoch": 128.35, + "learning_rate": 5.2480839168304e-05, + "loss": 2.4417, + "step": 63920 + }, + { + "epoch": 128.39, + "learning_rate": 5.243601355715796e-05, + "loss": 2.4519, + "step": 63940 + }, + { + "epoch": 128.43, + "learning_rate": 5.239118598335978e-05, + "loss": 2.455, + "step": 63960 + }, + { + "epoch": 128.47, + "learning_rate": 5.234635648302626e-05, + "loss": 2.4521, + "step": 63980 + }, + { + "epoch": 128.51, + "learning_rate": 5.23015250922757e-05, + "loss": 2.4621, + "step": 64000 + }, + { + "epoch": 128.51, + "eval_loss": 2.9333345890045166, + "eval_runtime": 32.4696, + "eval_samples_per_second": 23.591, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004982978891266681, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0004011640557006092, + "step": 64000 + }, + { + "epoch": 128.55, + "learning_rate": 5.225669184722795e-05, + "loss": 2.4677, + "step": 64020 + }, + { + "epoch": 128.59, + "learning_rate": 5.221185678400431e-05, + "loss": 2.4684, + "step": 64040 + }, + { + "epoch": 128.63, + "learning_rate": 5.216701993872762e-05, + "loss": 2.4749, + "step": 64060 + }, + { + "epoch": 128.67, + "learning_rate": 5.2122181347522116e-05, + "loss": 2.4668, + "step": 64080 + }, + { + "epoch": 128.71, + "learning_rate": 5.207734104651344e-05, + "loss": 2.4601, + "step": 64100 + }, + { + "epoch": 128.76, + "learning_rate": 5.203249907182859e-05, + "loss": 2.468, + "step": 64120 + }, + { + "epoch": 128.8, + "learning_rate": 5.1987655459595944e-05, + "loss": 2.4876, + "step": 64140 + }, + { + "epoch": 128.84, + "learning_rate": 5.194281024594522e-05, + "loss": 2.4785, + "step": 64160 + }, + { + "epoch": 128.88, + "learning_rate": 5.189796346700735e-05, + "loss": 2.4734, + "step": 64180 + }, + { + "epoch": 128.92, + "learning_rate": 5.185311515891459e-05, + "loss": 2.4854, + "step": 64200 + }, + { + "epoch": 128.96, + "learning_rate": 5.180826535780043e-05, + "loss": 2.4752, + "step": 64220 + }, + { + "epoch": 129.0, + "learning_rate": 5.1763414099799534e-05, + "loss": 2.4888, + "step": 64240 + }, + { + "epoch": 129.04, + "learning_rate": 5.171856142104775e-05, + "loss": 2.4187, + "step": 64260 + }, + { + "epoch": 129.08, + "learning_rate": 5.1673707357682077e-05, + "loss": 2.4198, + "step": 64280 + }, + { + "epoch": 129.12, + "learning_rate": 5.1628851945840596e-05, + "loss": 2.4197, + "step": 64300 + }, + { + "epoch": 129.16, + "learning_rate": 5.158399522166253e-05, + "loss": 2.43, + "step": 64320 + }, + { + "epoch": 129.2, + "learning_rate": 5.1541380151058595e-05, + "loss": 2.4324, + "step": 64340 + }, + { + "epoch": 129.24, + "learning_rate": 5.149652097177351e-05, + "loss": 2.4284, + "step": 64360 + }, + { + "epoch": 129.28, + "learning_rate": 5.1451660586768526e-05, + "loss": 2.4449, + "step": 64380 + }, + { + "epoch": 129.32, + "learning_rate": 5.140679903218679e-05, + "loss": 2.4595, + "step": 64400 + }, + { + "epoch": 129.36, + "learning_rate": 5.1361936344172525e-05, + "loss": 2.45, + "step": 64420 + }, + { + "epoch": 129.4, + "learning_rate": 5.1317072558870726e-05, + "loss": 2.4548, + "step": 64440 + }, + { + "epoch": 129.44, + "learning_rate": 5.127220771242737e-05, + "loss": 2.4474, + "step": 64460 + }, + { + "epoch": 129.48, + "learning_rate": 5.122734184098922e-05, + "loss": 2.4434, + "step": 64480 + }, + { + "epoch": 129.52, + "learning_rate": 5.118247498070393e-05, + "loss": 2.4658, + "step": 64500 + }, + { + "epoch": 129.56, + "learning_rate": 5.1137607167719905e-05, + "loss": 2.4688, + "step": 64520 + }, + { + "epoch": 129.6, + "learning_rate": 5.109273843818633e-05, + "loss": 2.4647, + "step": 64540 + }, + { + "epoch": 129.64, + "learning_rate": 5.104786882825313e-05, + "loss": 2.4517, + "step": 64560 + }, + { + "epoch": 129.68, + "learning_rate": 5.100299837407094e-05, + "loss": 2.4707, + "step": 64580 + }, + { + "epoch": 129.72, + "learning_rate": 5.0958127111791086e-05, + "loss": 2.4644, + "step": 64600 + }, + { + "epoch": 129.76, + "learning_rate": 5.0913255077565514e-05, + "loss": 2.4684, + "step": 64620 + }, + { + "epoch": 129.8, + "learning_rate": 5.0868382307546834e-05, + "loss": 2.4831, + "step": 64640 + }, + { + "epoch": 129.84, + "learning_rate": 5.08235088378882e-05, + "loss": 2.4532, + "step": 64660 + }, + { + "epoch": 129.88, + "learning_rate": 5.077863470474339e-05, + "loss": 2.464, + "step": 64680 + }, + { + "epoch": 129.92, + "learning_rate": 5.073375994426664e-05, + "loss": 2.4771, + "step": 64700 + }, + { + "epoch": 129.96, + "learning_rate": 5.068888459261276e-05, + "loss": 2.4671, + "step": 64720 + }, + { + "epoch": 130.0, + "learning_rate": 5.0644008685937005e-05, + "loss": 2.4662, + "step": 64740 + }, + { + "epoch": 130.04, + "learning_rate": 5.05991322603951e-05, + "loss": 2.4079, + "step": 64760 + }, + { + "epoch": 130.08, + "learning_rate": 5.055425535214314e-05, + "loss": 2.4095, + "step": 64780 + }, + { + "epoch": 130.12, + "learning_rate": 5.050937799733766e-05, + "loss": 2.4182, + "step": 64800 + }, + { + "epoch": 130.16, + "learning_rate": 5.0464500232135537e-05, + "loss": 2.4207, + "step": 64820 + }, + { + "epoch": 130.2, + "learning_rate": 5.041962209269393e-05, + "loss": 2.4299, + "step": 64840 + }, + { + "epoch": 130.24, + "learning_rate": 5.037474361517042e-05, + "loss": 2.4211, + "step": 64860 + }, + { + "epoch": 130.28, + "learning_rate": 5.032986483572273e-05, + "loss": 2.4383, + "step": 64880 + }, + { + "epoch": 130.32, + "learning_rate": 5.0284985790508924e-05, + "loss": 2.4168, + "step": 64900 + }, + { + "epoch": 130.36, + "learning_rate": 5.0240106515687194e-05, + "loss": 2.4418, + "step": 64920 + }, + { + "epoch": 130.4, + "learning_rate": 5.019522704741598e-05, + "loss": 2.4422, + "step": 64940 + }, + { + "epoch": 130.44, + "learning_rate": 5.015034742185386e-05, + "loss": 2.4476, + "step": 64960 + }, + { + "epoch": 130.48, + "learning_rate": 5.010546767515953e-05, + "loss": 2.4446, + "step": 64980 + }, + { + "epoch": 130.52, + "learning_rate": 5.006058784349181e-05, + "loss": 2.4503, + "step": 65000 + }, + { + "epoch": 130.52, + "eval_loss": 2.9224929809570312, + "eval_runtime": 32.4743, + "eval_samples_per_second": 23.588, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.00449143992339401, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00010433463725384498, + "step": 65000 + }, + { + "epoch": 130.56, + "learning_rate": 5.0015707963009575e-05, + "loss": 2.4619, + "step": 65020 + }, + { + "epoch": 130.6, + "learning_rate": 4.99708280698717e-05, + "loss": 2.454, + "step": 65040 + }, + { + "epoch": 130.64, + "learning_rate": 4.992594820023712e-05, + "loss": 2.4489, + "step": 65060 + }, + { + "epoch": 130.68, + "learning_rate": 4.988106839026474e-05, + "loss": 2.4638, + "step": 65080 + }, + { + "epoch": 130.72, + "learning_rate": 4.983618867611342e-05, + "loss": 2.456, + "step": 65100 + }, + { + "epoch": 130.76, + "learning_rate": 4.979130909394191e-05, + "loss": 2.4627, + "step": 65120 + }, + { + "epoch": 130.8, + "learning_rate": 4.97464296799089e-05, + "loss": 2.4689, + "step": 65140 + }, + { + "epoch": 130.84, + "learning_rate": 4.9701550470172914e-05, + "loss": 2.4707, + "step": 65160 + }, + { + "epoch": 130.88, + "learning_rate": 4.965667150089234e-05, + "loss": 2.4762, + "step": 65180 + }, + { + "epoch": 130.92, + "learning_rate": 4.9611792808225316e-05, + "loss": 2.4661, + "step": 65200 + }, + { + "epoch": 130.96, + "learning_rate": 4.956691442832981e-05, + "loss": 2.47, + "step": 65220 + }, + { + "epoch": 131.0, + "learning_rate": 4.952203639736354e-05, + "loss": 2.4674, + "step": 65240 + }, + { + "epoch": 131.04, + "learning_rate": 4.947715875148391e-05, + "loss": 2.4054, + "step": 65260 + }, + { + "epoch": 131.08, + "learning_rate": 4.9432281526848034e-05, + "loss": 2.4021, + "step": 65280 + }, + { + "epoch": 131.12, + "learning_rate": 4.9387404759612674e-05, + "loss": 2.4199, + "step": 65300 + }, + { + "epoch": 131.16, + "learning_rate": 4.934252848593424e-05, + "loss": 2.413, + "step": 65320 + }, + { + "epoch": 131.2, + "learning_rate": 4.929765274196872e-05, + "loss": 2.4068, + "step": 65340 + }, + { + "epoch": 131.24, + "learning_rate": 4.925277756387169e-05, + "loss": 2.4171, + "step": 65360 + }, + { + "epoch": 131.29, + "learning_rate": 4.920790298779831e-05, + "loss": 2.4108, + "step": 65380 + }, + { + "epoch": 131.33, + "learning_rate": 4.916302904990315e-05, + "loss": 2.445, + "step": 65400 + }, + { + "epoch": 131.37, + "learning_rate": 4.911815578634041e-05, + "loss": 2.4278, + "step": 65420 + }, + { + "epoch": 131.41, + "learning_rate": 4.907328323326359e-05, + "loss": 2.4474, + "step": 65440 + }, + { + "epoch": 131.45, + "learning_rate": 4.9028411426825746e-05, + "loss": 2.448, + "step": 65460 + }, + { + "epoch": 131.49, + "learning_rate": 4.898354040317926e-05, + "loss": 2.4544, + "step": 65480 + }, + { + "epoch": 131.53, + "learning_rate": 4.8938670198475904e-05, + "loss": 2.4343, + "step": 65500 + }, + { + "epoch": 131.57, + "learning_rate": 4.8893800848866824e-05, + "loss": 2.4455, + "step": 65520 + }, + { + "epoch": 131.61, + "learning_rate": 4.8848932390502396e-05, + "loss": 2.4433, + "step": 65540 + }, + { + "epoch": 131.65, + "learning_rate": 4.8804064859532366e-05, + "loss": 2.46, + "step": 65560 + }, + { + "epoch": 131.69, + "learning_rate": 4.8759198292105655e-05, + "loss": 2.4444, + "step": 65580 + }, + { + "epoch": 131.73, + "learning_rate": 4.8714332724370504e-05, + "loss": 2.4575, + "step": 65600 + }, + { + "epoch": 131.77, + "learning_rate": 4.866946819247422e-05, + "loss": 2.456, + "step": 65620 + }, + { + "epoch": 131.81, + "learning_rate": 4.8624604732563396e-05, + "loss": 2.4453, + "step": 65640 + }, + { + "epoch": 131.85, + "learning_rate": 4.85797423807837e-05, + "loss": 2.4541, + "step": 65660 + }, + { + "epoch": 131.89, + "learning_rate": 4.8534881173279905e-05, + "loss": 2.4645, + "step": 65680 + }, + { + "epoch": 131.93, + "learning_rate": 4.8490021146195904e-05, + "loss": 2.4625, + "step": 65700 + }, + { + "epoch": 131.97, + "learning_rate": 4.8445162335674574e-05, + "loss": 2.4662, + "step": 65720 + }, + { + "epoch": 132.01, + "learning_rate": 4.840030477785788e-05, + "loss": 2.4617, + "step": 65740 + }, + { + "epoch": 132.05, + "learning_rate": 4.835544850888671e-05, + "loss": 2.4039, + "step": 65760 + }, + { + "epoch": 132.09, + "learning_rate": 4.831059356490098e-05, + "loss": 2.4006, + "step": 65780 + }, + { + "epoch": 132.13, + "learning_rate": 4.826573998203952e-05, + "loss": 2.4069, + "step": 65800 + }, + { + "epoch": 132.17, + "learning_rate": 4.822088779644001e-05, + "loss": 2.4043, + "step": 65820 + }, + { + "epoch": 132.21, + "learning_rate": 4.817603704423909e-05, + "loss": 2.4249, + "step": 65840 + }, + { + "epoch": 132.25, + "learning_rate": 4.813118776157216e-05, + "loss": 2.4145, + "step": 65860 + }, + { + "epoch": 132.29, + "learning_rate": 4.808633998457351e-05, + "loss": 2.4363, + "step": 65880 + }, + { + "epoch": 132.33, + "learning_rate": 4.804149374937615e-05, + "loss": 2.4112, + "step": 65900 + }, + { + "epoch": 132.37, + "learning_rate": 4.79966490921119e-05, + "loss": 2.4311, + "step": 65920 + }, + { + "epoch": 132.41, + "learning_rate": 4.795180604891132e-05, + "loss": 2.4274, + "step": 65940 + }, + { + "epoch": 132.45, + "learning_rate": 4.790696465590358e-05, + "loss": 2.4308, + "step": 65960 + }, + { + "epoch": 132.49, + "learning_rate": 4.786212494921664e-05, + "loss": 2.4263, + "step": 65980 + }, + { + "epoch": 132.53, + "learning_rate": 4.7817286964976995e-05, + "loss": 2.4322, + "step": 66000 + }, + { + "epoch": 132.53, + "eval_loss": 2.9167988300323486, + "eval_runtime": 34.6199, + "eval_samples_per_second": 22.126, + "eval_steps_per_second": 0.116, + "eval_tse_ndup": 0.0040002918995163264, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0001314548883086742, + "step": 66000 + }, + { + "epoch": 132.57, + "learning_rate": 4.777245073930984e-05, + "loss": 2.4385, + "step": 66020 + }, + { + "epoch": 132.61, + "learning_rate": 4.772761630833886e-05, + "loss": 2.4258, + "step": 66040 + }, + { + "epoch": 132.65, + "learning_rate": 4.76827837081864e-05, + "loss": 2.4465, + "step": 66060 + }, + { + "epoch": 132.69, + "learning_rate": 4.7637952974973244e-05, + "loss": 2.4357, + "step": 66080 + }, + { + "epoch": 132.73, + "learning_rate": 4.7593124144818696e-05, + "loss": 2.4689, + "step": 66100 + }, + { + "epoch": 132.77, + "learning_rate": 4.754829725384059e-05, + "loss": 2.4448, + "step": 66120 + }, + { + "epoch": 132.81, + "learning_rate": 4.750347233815506e-05, + "loss": 2.4499, + "step": 66140 + }, + { + "epoch": 132.85, + "learning_rate": 4.74586494338768e-05, + "loss": 2.4504, + "step": 66160 + }, + { + "epoch": 132.89, + "learning_rate": 4.741382857711878e-05, + "loss": 2.4595, + "step": 66180 + }, + { + "epoch": 132.93, + "learning_rate": 4.736900980399237e-05, + "loss": 2.4606, + "step": 66200 + }, + { + "epoch": 132.97, + "learning_rate": 4.732419315060723e-05, + "loss": 2.4544, + "step": 66220 + }, + { + "epoch": 133.01, + "learning_rate": 4.727937865307134e-05, + "loss": 2.4545, + "step": 66240 + }, + { + "epoch": 133.05, + "learning_rate": 4.723456634749095e-05, + "loss": 2.4015, + "step": 66260 + }, + { + "epoch": 133.09, + "learning_rate": 4.718975626997047e-05, + "loss": 2.3997, + "step": 66280 + }, + { + "epoch": 133.13, + "learning_rate": 4.7144948456612634e-05, + "loss": 2.3937, + "step": 66300 + }, + { + "epoch": 133.17, + "learning_rate": 4.7100142943518235e-05, + "loss": 2.3982, + "step": 66320 + }, + { + "epoch": 133.21, + "learning_rate": 4.705533976678631e-05, + "loss": 2.4123, + "step": 66340 + }, + { + "epoch": 133.25, + "learning_rate": 4.701053896251392e-05, + "loss": 2.4, + "step": 66360 + }, + { + "epoch": 133.29, + "learning_rate": 4.696798042882182e-05, + "loss": 2.4211, + "step": 66380 + }, + { + "epoch": 133.33, + "learning_rate": 4.692318435466265e-05, + "loss": 2.4186, + "step": 66400 + }, + { + "epoch": 133.37, + "learning_rate": 4.687839075943825e-05, + "loss": 2.4208, + "step": 66420 + }, + { + "epoch": 133.41, + "learning_rate": 4.683359967923805e-05, + "loss": 2.4071, + "step": 66440 + }, + { + "epoch": 133.45, + "learning_rate": 4.6788811150149395e-05, + "loss": 2.4275, + "step": 66460 + }, + { + "epoch": 133.49, + "learning_rate": 4.674402520825759e-05, + "loss": 2.4245, + "step": 66480 + }, + { + "epoch": 133.53, + "learning_rate": 4.6699241889645874e-05, + "loss": 2.451, + "step": 66500 + }, + { + "epoch": 133.57, + "learning_rate": 4.665446123039534e-05, + "loss": 2.4336, + "step": 66520 + }, + { + "epoch": 133.61, + "learning_rate": 4.6609683266584974e-05, + "loss": 2.4374, + "step": 66540 + }, + { + "epoch": 133.65, + "learning_rate": 4.6564908034291553e-05, + "loss": 2.4392, + "step": 66560 + }, + { + "epoch": 133.69, + "learning_rate": 4.6520135569589675e-05, + "loss": 2.4459, + "step": 66580 + }, + { + "epoch": 133.73, + "learning_rate": 4.647536590855174e-05, + "loss": 2.434, + "step": 66600 + }, + { + "epoch": 133.78, + "learning_rate": 4.643059908724781e-05, + "loss": 2.4322, + "step": 66620 + }, + { + "epoch": 133.82, + "learning_rate": 4.6385835141745734e-05, + "loss": 2.4513, + "step": 66640 + }, + { + "epoch": 133.86, + "learning_rate": 4.6341074108111e-05, + "loss": 2.4373, + "step": 66660 + }, + { + "epoch": 133.9, + "learning_rate": 4.6296316022406776e-05, + "loss": 2.4493, + "step": 66680 + }, + { + "epoch": 133.94, + "learning_rate": 4.625156092069383e-05, + "loss": 2.4523, + "step": 66700 + }, + { + "epoch": 133.98, + "learning_rate": 4.620680883903054e-05, + "loss": 2.4569, + "step": 66720 + }, + { + "epoch": 134.02, + "learning_rate": 4.6162059813472876e-05, + "loss": 2.4319, + "step": 66740 + }, + { + "epoch": 134.06, + "learning_rate": 4.611731388007426e-05, + "loss": 2.3838, + "step": 66760 + }, + { + "epoch": 134.1, + "learning_rate": 4.6072571074885774e-05, + "loss": 2.3803, + "step": 66780 + }, + { + "epoch": 134.14, + "learning_rate": 4.602783143395579e-05, + "loss": 2.3874, + "step": 66800 + }, + { + "epoch": 134.18, + "learning_rate": 4.5983094993330296e-05, + "loss": 2.4049, + "step": 66820 + }, + { + "epoch": 134.22, + "learning_rate": 4.59383617890526e-05, + "loss": 2.4099, + "step": 66840 + }, + { + "epoch": 134.26, + "learning_rate": 4.5893631857163457e-05, + "loss": 2.3983, + "step": 66860 + }, + { + "epoch": 134.3, + "learning_rate": 4.584890523370096e-05, + "loss": 2.4111, + "step": 66880 + }, + { + "epoch": 134.34, + "learning_rate": 4.5804181954700524e-05, + "loss": 2.4121, + "step": 66900 + }, + { + "epoch": 134.38, + "learning_rate": 4.575946205619492e-05, + "loss": 2.4238, + "step": 66920 + }, + { + "epoch": 134.42, + "learning_rate": 4.5714745574214126e-05, + "loss": 2.3962, + "step": 66940 + }, + { + "epoch": 134.46, + "learning_rate": 4.5670032544785444e-05, + "loss": 2.4171, + "step": 66960 + }, + { + "epoch": 134.5, + "learning_rate": 4.5625323003933304e-05, + "loss": 2.4139, + "step": 66980 + }, + { + "epoch": 134.54, + "learning_rate": 4.558061698767941e-05, + "loss": 2.4239, + "step": 67000 + }, + { + "epoch": 134.54, + "eval_loss": 2.9044148921966553, + "eval_runtime": 32.4443, + "eval_samples_per_second": 23.61, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.005567236604094664, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0003309809964067315, + "step": 67000 + }, + { + "epoch": 134.58, + "learning_rate": 4.553591453204261e-05, + "loss": 2.4427, + "step": 67020 + }, + { + "epoch": 134.62, + "learning_rate": 4.549121567303881e-05, + "loss": 2.4326, + "step": 67040 + }, + { + "epoch": 134.66, + "learning_rate": 4.544652044668113e-05, + "loss": 2.4264, + "step": 67060 + }, + { + "epoch": 134.7, + "learning_rate": 4.5401828888979665e-05, + "loss": 2.445, + "step": 67080 + }, + { + "epoch": 134.74, + "learning_rate": 4.535714103594162e-05, + "loss": 2.4525, + "step": 67100 + }, + { + "epoch": 134.78, + "learning_rate": 4.531245692357117e-05, + "loss": 2.4244, + "step": 67120 + }, + { + "epoch": 134.82, + "learning_rate": 4.5267776587869503e-05, + "loss": 2.4433, + "step": 67140 + }, + { + "epoch": 134.86, + "learning_rate": 4.5223100064834786e-05, + "loss": 2.4547, + "step": 67160 + }, + { + "epoch": 134.9, + "learning_rate": 4.517842739046205e-05, + "loss": 2.4427, + "step": 67180 + }, + { + "epoch": 134.94, + "learning_rate": 4.5133758600743295e-05, + "loss": 2.4483, + "step": 67200 + }, + { + "epoch": 134.98, + "learning_rate": 4.5089093731667333e-05, + "loss": 2.4406, + "step": 67220 + }, + { + "epoch": 135.02, + "learning_rate": 4.504443281921985e-05, + "loss": 2.4247, + "step": 67240 + }, + { + "epoch": 135.06, + "learning_rate": 4.499977589938333e-05, + "loss": 2.3828, + "step": 67260 + }, + { + "epoch": 135.1, + "learning_rate": 4.495512300813707e-05, + "loss": 2.3818, + "step": 67280 + }, + { + "epoch": 135.14, + "learning_rate": 4.4910474181457055e-05, + "loss": 2.3649, + "step": 67300 + }, + { + "epoch": 135.18, + "learning_rate": 4.486582945531607e-05, + "loss": 2.3905, + "step": 67320 + }, + { + "epoch": 135.22, + "learning_rate": 4.482118886568357e-05, + "loss": 2.4007, + "step": 67340 + }, + { + "epoch": 135.26, + "learning_rate": 4.4776552448525646e-05, + "loss": 2.3917, + "step": 67360 + }, + { + "epoch": 135.3, + "learning_rate": 4.473192023980509e-05, + "loss": 2.4, + "step": 67380 + }, + { + "epoch": 135.34, + "learning_rate": 4.468729227548122e-05, + "loss": 2.4194, + "step": 67400 + }, + { + "epoch": 135.38, + "learning_rate": 4.464266859151003e-05, + "loss": 2.4167, + "step": 67420 + }, + { + "epoch": 135.42, + "learning_rate": 4.459804922384398e-05, + "loss": 2.4251, + "step": 67440 + }, + { + "epoch": 135.46, + "learning_rate": 4.455343420843207e-05, + "loss": 2.4038, + "step": 67460 + }, + { + "epoch": 135.5, + "learning_rate": 4.4508823581219886e-05, + "loss": 2.422, + "step": 67480 + }, + { + "epoch": 135.54, + "learning_rate": 4.446421737814933e-05, + "loss": 2.4177, + "step": 67500 + }, + { + "epoch": 135.58, + "learning_rate": 4.4419615635158875e-05, + "loss": 2.4096, + "step": 67520 + }, + { + "epoch": 135.62, + "learning_rate": 4.437501838818327e-05, + "loss": 2.4126, + "step": 67540 + }, + { + "epoch": 135.66, + "learning_rate": 4.433042567315378e-05, + "loss": 2.4195, + "step": 67560 + }, + { + "epoch": 135.7, + "learning_rate": 4.42858375259979e-05, + "loss": 2.424, + "step": 67580 + }, + { + "epoch": 135.74, + "learning_rate": 4.4241253982639525e-05, + "loss": 2.44, + "step": 67600 + }, + { + "epoch": 135.78, + "learning_rate": 4.4196675078998814e-05, + "loss": 2.4285, + "step": 67620 + }, + { + "epoch": 135.82, + "learning_rate": 4.415210085099215e-05, + "loss": 2.4408, + "step": 67640 + }, + { + "epoch": 135.86, + "learning_rate": 4.410753133453222e-05, + "loss": 2.4364, + "step": 67660 + }, + { + "epoch": 135.9, + "learning_rate": 4.4062966565527844e-05, + "loss": 2.4516, + "step": 67680 + }, + { + "epoch": 135.94, + "learning_rate": 4.40184065798841e-05, + "loss": 2.4469, + "step": 67700 + }, + { + "epoch": 135.98, + "learning_rate": 4.39738514135021e-05, + "loss": 2.4466, + "step": 67720 + }, + { + "epoch": 136.02, + "learning_rate": 4.3929301102279183e-05, + "loss": 2.3994, + "step": 67740 + }, + { + "epoch": 136.06, + "learning_rate": 4.388475568210873e-05, + "loss": 2.3792, + "step": 67760 + }, + { + "epoch": 136.1, + "learning_rate": 4.3840215188880157e-05, + "loss": 2.3846, + "step": 67780 + }, + { + "epoch": 136.14, + "learning_rate": 4.379567965847896e-05, + "loss": 2.4027, + "step": 67800 + }, + { + "epoch": 136.18, + "learning_rate": 4.3751149126786584e-05, + "loss": 2.395, + "step": 67820 + }, + { + "epoch": 136.22, + "learning_rate": 4.370662362968051e-05, + "loss": 2.407, + "step": 67840 + }, + { + "epoch": 136.27, + "learning_rate": 4.366210320303407e-05, + "loss": 2.4021, + "step": 67860 + }, + { + "epoch": 136.31, + "learning_rate": 4.361758788271662e-05, + "loss": 2.4014, + "step": 67880 + }, + { + "epoch": 136.35, + "learning_rate": 4.357307770459335e-05, + "loss": 2.3923, + "step": 67900 + }, + { + "epoch": 136.39, + "learning_rate": 4.352857270452527e-05, + "loss": 2.4016, + "step": 67920 + }, + { + "epoch": 136.43, + "learning_rate": 4.348407291836928e-05, + "loss": 2.4044, + "step": 67940 + }, + { + "epoch": 136.47, + "learning_rate": 4.343957838197805e-05, + "loss": 2.4098, + "step": 67960 + }, + { + "epoch": 136.51, + "learning_rate": 4.339508913120002e-05, + "loss": 2.4079, + "step": 67980 + }, + { + "epoch": 136.55, + "learning_rate": 4.335060520187938e-05, + "loss": 2.414, + "step": 68000 + }, + { + "epoch": 136.55, + "eval_loss": 2.897592782974243, + "eval_runtime": 34.661, + "eval_samples_per_second": 22.1, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.00402719277773755, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00016751453198233807, + "step": 68000 + }, + { + "epoch": 136.59, + "learning_rate": 4.3306126629856e-05, + "loss": 2.4271, + "step": 68020 + }, + { + "epoch": 136.63, + "learning_rate": 4.3261653450965536e-05, + "loss": 2.414, + "step": 68040 + }, + { + "epoch": 136.67, + "learning_rate": 4.3217185701039144e-05, + "loss": 2.4226, + "step": 68060 + }, + { + "epoch": 136.71, + "learning_rate": 4.317272341590373e-05, + "loss": 2.4232, + "step": 68080 + }, + { + "epoch": 136.75, + "learning_rate": 4.312826663138174e-05, + "loss": 2.3941, + "step": 68100 + }, + { + "epoch": 136.79, + "learning_rate": 4.308381538329123e-05, + "loss": 2.4166, + "step": 68120 + }, + { + "epoch": 136.83, + "learning_rate": 4.3039369707445714e-05, + "loss": 2.4189, + "step": 68140 + }, + { + "epoch": 136.87, + "learning_rate": 4.2994929639654315e-05, + "loss": 2.4313, + "step": 68160 + }, + { + "epoch": 136.91, + "learning_rate": 4.295049521572157e-05, + "loss": 2.4384, + "step": 68180 + }, + { + "epoch": 136.95, + "learning_rate": 4.29060664714475e-05, + "loss": 2.4313, + "step": 68200 + }, + { + "epoch": 136.99, + "learning_rate": 4.2861643442627564e-05, + "loss": 2.4227, + "step": 68220 + }, + { + "epoch": 137.03, + "learning_rate": 4.281722616505254e-05, + "loss": 2.3861, + "step": 68240 + }, + { + "epoch": 137.07, + "learning_rate": 4.277281467450869e-05, + "loss": 2.3626, + "step": 68260 + }, + { + "epoch": 137.11, + "learning_rate": 4.27284090067775e-05, + "loss": 2.3773, + "step": 68280 + }, + { + "epoch": 137.15, + "learning_rate": 4.2684009197635824e-05, + "loss": 2.3566, + "step": 68300 + }, + { + "epoch": 137.19, + "learning_rate": 4.2639615282855816e-05, + "loss": 2.3858, + "step": 68320 + }, + { + "epoch": 137.23, + "learning_rate": 4.259522729820482e-05, + "loss": 2.3906, + "step": 68340 + }, + { + "epoch": 137.27, + "learning_rate": 4.255084527944545e-05, + "loss": 2.3967, + "step": 68360 + }, + { + "epoch": 137.31, + "learning_rate": 4.250646926233547e-05, + "loss": 2.3795, + "step": 68380 + }, + { + "epoch": 137.35, + "learning_rate": 4.246209928262789e-05, + "loss": 2.4031, + "step": 68400 + }, + { + "epoch": 137.39, + "learning_rate": 4.241773537607073e-05, + "loss": 2.3821, + "step": 68420 + }, + { + "epoch": 137.43, + "learning_rate": 4.237337757840726e-05, + "loss": 2.3963, + "step": 68440 + }, + { + "epoch": 137.47, + "learning_rate": 4.232902592537569e-05, + "loss": 2.4102, + "step": 68460 + }, + { + "epoch": 137.51, + "learning_rate": 4.2284680452709394e-05, + "loss": 2.4089, + "step": 68480 + }, + { + "epoch": 137.55, + "learning_rate": 4.22403411961367e-05, + "loss": 2.4086, + "step": 68500 + }, + { + "epoch": 137.59, + "learning_rate": 4.219822469258667e-05, + "loss": 2.4102, + "step": 68520 + }, + { + "epoch": 137.63, + "learning_rate": 4.215389766014115e-05, + "loss": 2.4142, + "step": 68540 + }, + { + "epoch": 137.67, + "learning_rate": 4.2109576949158593e-05, + "loss": 2.4178, + "step": 68560 + }, + { + "epoch": 137.71, + "learning_rate": 4.206526259534736e-05, + "loss": 2.4187, + "step": 68580 + }, + { + "epoch": 137.75, + "learning_rate": 4.202095463441075e-05, + "loss": 2.4304, + "step": 68600 + }, + { + "epoch": 137.79, + "learning_rate": 4.197665310204685e-05, + "loss": 2.4157, + "step": 68620 + }, + { + "epoch": 137.83, + "learning_rate": 4.1932358033948644e-05, + "loss": 2.429, + "step": 68640 + }, + { + "epoch": 137.87, + "learning_rate": 4.188806946580384e-05, + "loss": 2.4315, + "step": 68660 + }, + { + "epoch": 137.91, + "learning_rate": 4.184378743329493e-05, + "loss": 2.4313, + "step": 68680 + }, + { + "epoch": 137.95, + "learning_rate": 4.1799511972099214e-05, + "loss": 2.4321, + "step": 68700 + }, + { + "epoch": 137.99, + "learning_rate": 4.175524311788854e-05, + "loss": 2.4333, + "step": 68720 + }, + { + "epoch": 138.03, + "learning_rate": 4.17109809063296e-05, + "loss": 2.3698, + "step": 68740 + }, + { + "epoch": 138.07, + "learning_rate": 4.1666725373083604e-05, + "loss": 2.3635, + "step": 68760 + }, + { + "epoch": 138.11, + "learning_rate": 4.162247655380647e-05, + "loss": 2.3718, + "step": 68780 + }, + { + "epoch": 138.15, + "learning_rate": 4.157823448414865e-05, + "loss": 2.3774, + "step": 68800 + }, + { + "epoch": 138.19, + "learning_rate": 4.1533999199755194e-05, + "loss": 2.3692, + "step": 68820 + }, + { + "epoch": 138.23, + "learning_rate": 4.148977073626568e-05, + "loss": 2.3888, + "step": 68840 + }, + { + "epoch": 138.27, + "learning_rate": 4.1445549129314144e-05, + "loss": 2.3857, + "step": 68860 + }, + { + "epoch": 138.31, + "learning_rate": 4.1401334414529196e-05, + "loss": 2.3789, + "step": 68880 + }, + { + "epoch": 138.35, + "learning_rate": 4.135712662753377e-05, + "loss": 2.3811, + "step": 68900 + }, + { + "epoch": 138.39, + "learning_rate": 4.131292580394534e-05, + "loss": 2.4066, + "step": 68920 + }, + { + "epoch": 138.43, + "learning_rate": 4.126873197937565e-05, + "loss": 2.387, + "step": 68940 + }, + { + "epoch": 138.47, + "learning_rate": 4.122454518943092e-05, + "loss": 2.3983, + "step": 68960 + }, + { + "epoch": 138.51, + "learning_rate": 4.1180365469711625e-05, + "loss": 2.395, + "step": 68980 + }, + { + "epoch": 138.55, + "learning_rate": 4.113619285581256e-05, + "loss": 2.4056, + "step": 69000 + }, + { + "epoch": 138.55, + "eval_loss": 2.8908073902130127, + "eval_runtime": 34.7957, + "eval_samples_per_second": 22.014, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.005084230267726611, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0003037834333083851, + "step": 69000 + }, + { + "epoch": 138.59, + "learning_rate": 4.109202738332283e-05, + "loss": 2.4024, + "step": 69020 + }, + { + "epoch": 138.63, + "learning_rate": 4.104786908782572e-05, + "loss": 2.4028, + "step": 69040 + }, + { + "epoch": 138.67, + "learning_rate": 4.100371800489881e-05, + "loss": 2.4063, + "step": 69060 + }, + { + "epoch": 138.71, + "learning_rate": 4.095957417011379e-05, + "loss": 2.4107, + "step": 69080 + }, + { + "epoch": 138.76, + "learning_rate": 4.091543761903658e-05, + "loss": 2.4094, + "step": 69100 + }, + { + "epoch": 138.8, + "learning_rate": 4.0871308387227195e-05, + "loss": 2.409, + "step": 69120 + }, + { + "epoch": 138.84, + "learning_rate": 4.0827186510239755e-05, + "loss": 2.4261, + "step": 69140 + }, + { + "epoch": 138.88, + "learning_rate": 4.078307202362248e-05, + "loss": 2.4191, + "step": 69160 + }, + { + "epoch": 138.92, + "learning_rate": 4.0738964962917584e-05, + "loss": 2.4248, + "step": 69180 + }, + { + "epoch": 138.96, + "learning_rate": 4.0694865363661344e-05, + "loss": 2.4314, + "step": 69200 + }, + { + "epoch": 139.0, + "learning_rate": 4.0650773261384007e-05, + "loss": 2.4334, + "step": 69220 + }, + { + "epoch": 139.04, + "learning_rate": 4.060668869160977e-05, + "loss": 2.3552, + "step": 69240 + }, + { + "epoch": 139.08, + "learning_rate": 4.056261168985682e-05, + "loss": 2.3652, + "step": 69260 + }, + { + "epoch": 139.12, + "learning_rate": 4.0518542291637154e-05, + "loss": 2.3601, + "step": 69280 + }, + { + "epoch": 139.16, + "learning_rate": 4.047448053245673e-05, + "loss": 2.3678, + "step": 69300 + }, + { + "epoch": 139.2, + "learning_rate": 4.043042644781526e-05, + "loss": 2.3743, + "step": 69320 + }, + { + "epoch": 139.24, + "learning_rate": 4.0386380073206376e-05, + "loss": 2.3851, + "step": 69340 + }, + { + "epoch": 139.28, + "learning_rate": 4.034234144411741e-05, + "loss": 2.3824, + "step": 69360 + }, + { + "epoch": 139.32, + "learning_rate": 4.02983105960295e-05, + "loss": 2.3827, + "step": 69380 + }, + { + "epoch": 139.36, + "learning_rate": 4.0254287564417534e-05, + "loss": 2.3939, + "step": 69400 + }, + { + "epoch": 139.4, + "learning_rate": 4.0210272384750026e-05, + "loss": 2.3988, + "step": 69420 + }, + { + "epoch": 139.44, + "learning_rate": 4.0166265092489255e-05, + "loss": 2.405, + "step": 69440 + }, + { + "epoch": 139.48, + "learning_rate": 4.012226572309106e-05, + "loss": 2.3923, + "step": 69460 + }, + { + "epoch": 139.52, + "learning_rate": 4.0078274312004973e-05, + "loss": 2.3854, + "step": 69480 + }, + { + "epoch": 139.56, + "learning_rate": 4.003429089467405e-05, + "loss": 2.3944, + "step": 69500 + }, + { + "epoch": 139.6, + "learning_rate": 3.9990315506534955e-05, + "loss": 2.3983, + "step": 69520 + }, + { + "epoch": 139.64, + "learning_rate": 3.9946348183017865e-05, + "loss": 2.3957, + "step": 69540 + }, + { + "epoch": 139.68, + "learning_rate": 3.9902388959546454e-05, + "loss": 2.3992, + "step": 69560 + }, + { + "epoch": 139.72, + "learning_rate": 3.985843787153788e-05, + "loss": 2.4076, + "step": 69580 + }, + { + "epoch": 139.76, + "learning_rate": 3.981449495440271e-05, + "loss": 2.4088, + "step": 69600 + }, + { + "epoch": 139.8, + "learning_rate": 3.977056024354502e-05, + "loss": 2.4012, + "step": 69620 + }, + { + "epoch": 139.84, + "learning_rate": 3.972663377436214e-05, + "loss": 2.4073, + "step": 69640 + }, + { + "epoch": 139.88, + "learning_rate": 3.968271558224489e-05, + "loss": 2.3936, + "step": 69660 + }, + { + "epoch": 139.92, + "learning_rate": 3.963880570257733e-05, + "loss": 2.4103, + "step": 69680 + }, + { + "epoch": 139.96, + "learning_rate": 3.9594904170736865e-05, + "loss": 2.4322, + "step": 69700 + }, + { + "epoch": 140.0, + "learning_rate": 3.9551011022094195e-05, + "loss": 2.4189, + "step": 69720 + }, + { + "epoch": 140.04, + "learning_rate": 3.95071262920132e-05, + "loss": 2.3542, + "step": 69740 + }, + { + "epoch": 140.08, + "learning_rate": 3.946325001585103e-05, + "loss": 2.3689, + "step": 69760 + }, + { + "epoch": 140.12, + "learning_rate": 3.941938222895799e-05, + "loss": 2.3536, + "step": 69780 + }, + { + "epoch": 140.16, + "learning_rate": 3.937552296667761e-05, + "loss": 2.3777, + "step": 69800 + }, + { + "epoch": 140.2, + "learning_rate": 3.9331672264346445e-05, + "loss": 2.3683, + "step": 69820 + }, + { + "epoch": 140.24, + "learning_rate": 3.9287830157294244e-05, + "loss": 2.3716, + "step": 69840 + }, + { + "epoch": 140.28, + "learning_rate": 3.9243996680843824e-05, + "loss": 2.3841, + "step": 69860 + }, + { + "epoch": 140.32, + "learning_rate": 3.920017187031098e-05, + "loss": 2.3848, + "step": 69880 + }, + { + "epoch": 140.36, + "learning_rate": 3.9156355761004615e-05, + "loss": 2.3594, + "step": 69900 + }, + { + "epoch": 140.4, + "learning_rate": 3.911254838822653e-05, + "loss": 2.3847, + "step": 69920 + }, + { + "epoch": 140.44, + "learning_rate": 3.906874978727158e-05, + "loss": 2.3949, + "step": 69940 + }, + { + "epoch": 140.48, + "learning_rate": 3.902495999342747e-05, + "loss": 2.3905, + "step": 69960 + }, + { + "epoch": 140.52, + "learning_rate": 3.898117904197485e-05, + "loss": 2.3905, + "step": 69980 + }, + { + "epoch": 140.56, + "learning_rate": 3.893740696818727e-05, + "loss": 2.381, + "step": 70000 + }, + { + "epoch": 140.56, + "eval_loss": 2.885134696960449, + "eval_runtime": 32.4825, + "eval_samples_per_second": 23.582, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.005313852572959812, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00019718233246301134, + "step": 70000 + }, + { + "epoch": 140.6, + "learning_rate": 3.889364380733107e-05, + "loss": 2.3849, + "step": 70020 + }, + { + "epoch": 140.64, + "learning_rate": 3.884988959466546e-05, + "loss": 2.4015, + "step": 70040 + }, + { + "epoch": 140.68, + "learning_rate": 3.88061443654424e-05, + "loss": 2.4043, + "step": 70060 + }, + { + "epoch": 140.72, + "learning_rate": 3.876240815490665e-05, + "loss": 2.4026, + "step": 70080 + }, + { + "epoch": 140.76, + "learning_rate": 3.871868099829566e-05, + "loss": 2.399, + "step": 70100 + }, + { + "epoch": 140.8, + "learning_rate": 3.867496293083962e-05, + "loss": 2.3907, + "step": 70120 + }, + { + "epoch": 140.84, + "learning_rate": 3.863125398776142e-05, + "loss": 2.3964, + "step": 70140 + }, + { + "epoch": 140.88, + "learning_rate": 3.858755420427651e-05, + "loss": 2.4, + "step": 70160 + }, + { + "epoch": 140.92, + "learning_rate": 3.854386361559306e-05, + "loss": 2.4196, + "step": 70180 + }, + { + "epoch": 140.96, + "learning_rate": 3.8500182256911766e-05, + "loss": 2.4057, + "step": 70200 + }, + { + "epoch": 141.0, + "learning_rate": 3.845651016342592e-05, + "loss": 2.4055, + "step": 70220 + }, + { + "epoch": 141.04, + "learning_rate": 3.8412847370321335e-05, + "loss": 2.3458, + "step": 70240 + }, + { + "epoch": 141.08, + "learning_rate": 3.8369193912776324e-05, + "loss": 2.3606, + "step": 70260 + }, + { + "epoch": 141.12, + "learning_rate": 3.8325549825961713e-05, + "loss": 2.3758, + "step": 70280 + }, + { + "epoch": 141.16, + "learning_rate": 3.828191514504072e-05, + "loss": 2.3652, + "step": 70300 + }, + { + "epoch": 141.2, + "learning_rate": 3.823828990516909e-05, + "loss": 2.3559, + "step": 70320 + }, + { + "epoch": 141.24, + "learning_rate": 3.8194674141494785e-05, + "loss": 2.364, + "step": 70340 + }, + { + "epoch": 141.29, + "learning_rate": 3.815106788915834e-05, + "loss": 2.3633, + "step": 70360 + }, + { + "epoch": 141.33, + "learning_rate": 3.810747118329244e-05, + "loss": 2.3684, + "step": 70380 + }, + { + "epoch": 141.37, + "learning_rate": 3.8063884059022214e-05, + "loss": 2.3849, + "step": 70400 + }, + { + "epoch": 141.41, + "learning_rate": 3.8020306551465024e-05, + "loss": 2.3867, + "step": 70420 + }, + { + "epoch": 141.45, + "learning_rate": 3.7976738695730456e-05, + "loss": 2.3799, + "step": 70440 + }, + { + "epoch": 141.49, + "learning_rate": 3.793318052692036e-05, + "loss": 2.3909, + "step": 70460 + }, + { + "epoch": 141.53, + "learning_rate": 3.788963208012876e-05, + "loss": 2.3672, + "step": 70480 + }, + { + "epoch": 141.57, + "learning_rate": 3.7846093390441854e-05, + "loss": 2.3783, + "step": 70500 + }, + { + "epoch": 141.61, + "learning_rate": 3.780256449293796e-05, + "loss": 2.3941, + "step": 70520 + }, + { + "epoch": 141.65, + "learning_rate": 3.775904542268755e-05, + "loss": 2.384, + "step": 70540 + }, + { + "epoch": 141.69, + "learning_rate": 3.771553621475316e-05, + "loss": 2.3996, + "step": 70560 + }, + { + "epoch": 141.73, + "learning_rate": 3.767203690418935e-05, + "loss": 2.3925, + "step": 70580 + }, + { + "epoch": 141.77, + "learning_rate": 3.762854752604273e-05, + "loss": 2.3755, + "step": 70600 + }, + { + "epoch": 141.81, + "learning_rate": 3.758506811535192e-05, + "loss": 2.3873, + "step": 70620 + }, + { + "epoch": 141.85, + "learning_rate": 3.75415987071475e-05, + "loss": 2.4052, + "step": 70640 + }, + { + "epoch": 141.89, + "learning_rate": 3.7498139336451955e-05, + "loss": 2.3902, + "step": 70660 + }, + { + "epoch": 141.93, + "learning_rate": 3.745469003827973e-05, + "loss": 2.4098, + "step": 70680 + }, + { + "epoch": 141.97, + "learning_rate": 3.741125084763716e-05, + "loss": 2.4101, + "step": 70700 + }, + { + "epoch": 142.01, + "learning_rate": 3.7367821799522404e-05, + "loss": 2.3957, + "step": 70720 + }, + { + "epoch": 142.05, + "learning_rate": 3.73265736301992e-05, + "loss": 2.3342, + "step": 70740 + }, + { + "epoch": 142.09, + "learning_rate": 3.7283164460646194e-05, + "loss": 2.3553, + "step": 70760 + }, + { + "epoch": 142.13, + "learning_rate": 3.723976553681787e-05, + "loss": 2.3565, + "step": 70780 + }, + { + "epoch": 142.17, + "learning_rate": 3.719637689368001e-05, + "loss": 2.359, + "step": 70800 + }, + { + "epoch": 142.21, + "learning_rate": 3.715299856619e-05, + "loss": 2.3593, + "step": 70820 + }, + { + "epoch": 142.25, + "learning_rate": 3.710963058929701e-05, + "loss": 2.37, + "step": 70840 + }, + { + "epoch": 142.29, + "learning_rate": 3.706627299794182e-05, + "loss": 2.3731, + "step": 70860 + }, + { + "epoch": 142.33, + "learning_rate": 3.7022925827056884e-05, + "loss": 2.3687, + "step": 70880 + }, + { + "epoch": 142.37, + "learning_rate": 3.697958911156619e-05, + "loss": 2.3674, + "step": 70900 + }, + { + "epoch": 142.41, + "learning_rate": 3.693626288638538e-05, + "loss": 2.3746, + "step": 70920 + }, + { + "epoch": 142.45, + "learning_rate": 3.6892947186421613e-05, + "loss": 2.3907, + "step": 70940 + }, + { + "epoch": 142.49, + "learning_rate": 3.684964204657354e-05, + "loss": 2.3744, + "step": 70960 + }, + { + "epoch": 142.53, + "learning_rate": 3.680634750173137e-05, + "loss": 2.3859, + "step": 70980 + }, + { + "epoch": 142.57, + "learning_rate": 3.676306358677668e-05, + "loss": 2.3824, + "step": 71000 + }, + { + "epoch": 142.57, + "eval_loss": 2.8772501945495605, + "eval_runtime": 32.5223, + "eval_samples_per_second": 23.553, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004397789483317988, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0002335230619709903, + "step": 71000 + }, + { + "epoch": 142.61, + "learning_rate": 3.67197903365826e-05, + "loss": 2.3808, + "step": 71020 + }, + { + "epoch": 142.65, + "learning_rate": 3.667652778601356e-05, + "loss": 2.3775, + "step": 71040 + }, + { + "epoch": 142.69, + "learning_rate": 3.6633275969925426e-05, + "loss": 2.3795, + "step": 71060 + }, + { + "epoch": 142.73, + "learning_rate": 3.659003492316543e-05, + "loss": 2.3905, + "step": 71080 + }, + { + "epoch": 142.77, + "learning_rate": 3.6546804680572065e-05, + "loss": 2.3841, + "step": 71100 + }, + { + "epoch": 142.81, + "learning_rate": 3.650358527697519e-05, + "loss": 2.3948, + "step": 71120 + }, + { + "epoch": 142.85, + "learning_rate": 3.646037674719588e-05, + "loss": 2.404, + "step": 71140 + }, + { + "epoch": 142.89, + "learning_rate": 3.6417179126046475e-05, + "loss": 2.3971, + "step": 71160 + }, + { + "epoch": 142.93, + "learning_rate": 3.6373992448330486e-05, + "loss": 2.3914, + "step": 71180 + }, + { + "epoch": 142.97, + "learning_rate": 3.633081674884268e-05, + "loss": 2.3752, + "step": 71200 + }, + { + "epoch": 143.01, + "learning_rate": 3.628765206236894e-05, + "loss": 2.3844, + "step": 71220 + }, + { + "epoch": 143.05, + "learning_rate": 3.624449842368624e-05, + "loss": 2.3404, + "step": 71240 + }, + { + "epoch": 143.09, + "learning_rate": 3.620135586756273e-05, + "loss": 2.3327, + "step": 71260 + }, + { + "epoch": 143.13, + "learning_rate": 3.6158224428757535e-05, + "loss": 2.3588, + "step": 71280 + }, + { + "epoch": 143.17, + "learning_rate": 3.611510414202093e-05, + "loss": 2.3554, + "step": 71300 + }, + { + "epoch": 143.21, + "learning_rate": 3.607199504209411e-05, + "loss": 2.3503, + "step": 71320 + }, + { + "epoch": 143.25, + "learning_rate": 3.6028897163709305e-05, + "loss": 2.3444, + "step": 71340 + }, + { + "epoch": 143.29, + "learning_rate": 3.598581054158973e-05, + "loss": 2.37, + "step": 71360 + }, + { + "epoch": 143.33, + "learning_rate": 3.594273521044945e-05, + "loss": 2.3587, + "step": 71380 + }, + { + "epoch": 143.37, + "learning_rate": 3.589967120499353e-05, + "loss": 2.3627, + "step": 71400 + }, + { + "epoch": 143.41, + "learning_rate": 3.585661855991783e-05, + "loss": 2.3547, + "step": 71420 + }, + { + "epoch": 143.45, + "learning_rate": 3.5813577309909104e-05, + "loss": 2.3666, + "step": 71440 + }, + { + "epoch": 143.49, + "learning_rate": 3.577054748964489e-05, + "loss": 2.3725, + "step": 71460 + }, + { + "epoch": 143.53, + "learning_rate": 3.5727529133793535e-05, + "loss": 2.376, + "step": 71480 + }, + { + "epoch": 143.57, + "learning_rate": 3.56845222770142e-05, + "loss": 2.3754, + "step": 71500 + }, + { + "epoch": 143.61, + "learning_rate": 3.564152695395667e-05, + "loss": 2.3841, + "step": 71520 + }, + { + "epoch": 143.65, + "learning_rate": 3.559854319926156e-05, + "loss": 2.3869, + "step": 71540 + }, + { + "epoch": 143.69, + "learning_rate": 3.5555571047560036e-05, + "loss": 2.3887, + "step": 71560 + }, + { + "epoch": 143.73, + "learning_rate": 3.551261053347404e-05, + "loss": 2.3847, + "step": 71580 + }, + { + "epoch": 143.78, + "learning_rate": 3.546966169161605e-05, + "loss": 2.3828, + "step": 71600 + }, + { + "epoch": 143.82, + "learning_rate": 3.5426724556589195e-05, + "loss": 2.3981, + "step": 71620 + }, + { + "epoch": 143.86, + "learning_rate": 3.538379916298715e-05, + "loss": 2.3896, + "step": 71640 + }, + { + "epoch": 143.9, + "learning_rate": 3.534088554539411e-05, + "loss": 2.3818, + "step": 71660 + }, + { + "epoch": 143.94, + "learning_rate": 3.529798373838481e-05, + "loss": 2.378, + "step": 71680 + }, + { + "epoch": 143.98, + "learning_rate": 3.525509377652445e-05, + "loss": 2.397, + "step": 71700 + }, + { + "epoch": 144.02, + "learning_rate": 3.5212215694368746e-05, + "loss": 2.3562, + "step": 71720 + }, + { + "epoch": 144.06, + "learning_rate": 3.5169349526463716e-05, + "loss": 2.3467, + "step": 71740 + }, + { + "epoch": 144.1, + "learning_rate": 3.512649530734591e-05, + "loss": 2.3406, + "step": 71760 + }, + { + "epoch": 144.14, + "learning_rate": 3.50836530715422e-05, + "loss": 2.3483, + "step": 71780 + }, + { + "epoch": 144.18, + "learning_rate": 3.504082285356978e-05, + "loss": 2.3423, + "step": 71800 + }, + { + "epoch": 144.22, + "learning_rate": 3.4998004687936196e-05, + "loss": 2.3615, + "step": 71820 + }, + { + "epoch": 144.26, + "learning_rate": 3.495519860913926e-05, + "loss": 2.3603, + "step": 71840 + }, + { + "epoch": 144.3, + "learning_rate": 3.4912404651667074e-05, + "loss": 2.3647, + "step": 71860 + }, + { + "epoch": 144.34, + "learning_rate": 3.4869622849997924e-05, + "loss": 2.3671, + "step": 71880 + }, + { + "epoch": 144.38, + "learning_rate": 3.482685323860037e-05, + "loss": 2.3703, + "step": 71900 + }, + { + "epoch": 144.42, + "learning_rate": 3.4784095851933105e-05, + "loss": 2.365, + "step": 71920 + }, + { + "epoch": 144.46, + "learning_rate": 3.4741350724444975e-05, + "loss": 2.3687, + "step": 71940 + }, + { + "epoch": 144.5, + "learning_rate": 3.469861789057497e-05, + "loss": 2.3504, + "step": 71960 + }, + { + "epoch": 144.54, + "learning_rate": 3.4655897384752146e-05, + "loss": 2.361, + "step": 71980 + }, + { + "epoch": 144.58, + "learning_rate": 3.461318924139566e-05, + "loss": 2.3682, + "step": 72000 + }, + { + "epoch": 144.58, + "eval_loss": 2.872910499572754, + "eval_runtime": 32.2581, + "eval_samples_per_second": 23.746, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.004201343868974207, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00019491586887148248, + "step": 72000 + }, + { + "epoch": 144.62, + "learning_rate": 3.4570493494914674e-05, + "loss": 2.3858, + "step": 72020 + }, + { + "epoch": 144.66, + "learning_rate": 3.45278101797084e-05, + "loss": 2.3671, + "step": 72040 + }, + { + "epoch": 144.7, + "learning_rate": 3.448513933016599e-05, + "loss": 2.3578, + "step": 72060 + }, + { + "epoch": 144.74, + "learning_rate": 3.4442480980666584e-05, + "loss": 2.3732, + "step": 72080 + }, + { + "epoch": 144.78, + "learning_rate": 3.4399835165579266e-05, + "loss": 2.385, + "step": 72100 + }, + { + "epoch": 144.82, + "learning_rate": 3.435720191926297e-05, + "loss": 2.3682, + "step": 72120 + }, + { + "epoch": 144.86, + "learning_rate": 3.431458127606656e-05, + "loss": 2.3792, + "step": 72140 + }, + { + "epoch": 144.9, + "learning_rate": 3.427197327032868e-05, + "loss": 2.3811, + "step": 72160 + }, + { + "epoch": 144.94, + "learning_rate": 3.422937793637788e-05, + "loss": 2.3823, + "step": 72180 + }, + { + "epoch": 144.98, + "learning_rate": 3.418679530853241e-05, + "loss": 2.3915, + "step": 72200 + }, + { + "epoch": 145.02, + "learning_rate": 3.414422542110033e-05, + "loss": 2.3557, + "step": 72220 + }, + { + "epoch": 145.06, + "learning_rate": 3.4101668308379466e-05, + "loss": 2.3325, + "step": 72240 + }, + { + "epoch": 145.1, + "learning_rate": 3.405912400465726e-05, + "loss": 2.3512, + "step": 72260 + }, + { + "epoch": 145.14, + "learning_rate": 3.401659254421094e-05, + "loss": 2.3372, + "step": 72280 + }, + { + "epoch": 145.18, + "learning_rate": 3.397407396130729e-05, + "loss": 2.3402, + "step": 72300 + }, + { + "epoch": 145.22, + "learning_rate": 3.39315682902028e-05, + "loss": 2.3384, + "step": 72320 + }, + { + "epoch": 145.26, + "learning_rate": 3.3889075565143505e-05, + "loss": 2.3509, + "step": 72340 + }, + { + "epoch": 145.3, + "learning_rate": 3.3846595820365025e-05, + "loss": 2.3418, + "step": 72360 + }, + { + "epoch": 145.34, + "learning_rate": 3.380412909009254e-05, + "loss": 2.3593, + "step": 72380 + }, + { + "epoch": 145.38, + "learning_rate": 3.376167540854069e-05, + "loss": 2.345, + "step": 72400 + }, + { + "epoch": 145.42, + "learning_rate": 3.3719234809913694e-05, + "loss": 2.3507, + "step": 72420 + }, + { + "epoch": 145.46, + "learning_rate": 3.3676807328405124e-05, + "loss": 2.3576, + "step": 72440 + }, + { + "epoch": 145.5, + "learning_rate": 3.3634392998198075e-05, + "loss": 2.3616, + "step": 72460 + }, + { + "epoch": 145.54, + "learning_rate": 3.3591991853464965e-05, + "loss": 2.3591, + "step": 72480 + }, + { + "epoch": 145.58, + "learning_rate": 3.354960392836767e-05, + "loss": 2.3646, + "step": 72500 + }, + { + "epoch": 145.62, + "learning_rate": 3.350722925705736e-05, + "loss": 2.3592, + "step": 72520 + }, + { + "epoch": 145.66, + "learning_rate": 3.346486787367454e-05, + "loss": 2.3774, + "step": 72540 + }, + { + "epoch": 145.7, + "learning_rate": 3.342251981234902e-05, + "loss": 2.3662, + "step": 72560 + }, + { + "epoch": 145.74, + "learning_rate": 3.3380185107199855e-05, + "loss": 2.3787, + "step": 72580 + }, + { + "epoch": 145.78, + "learning_rate": 3.3337863792335365e-05, + "loss": 2.385, + "step": 72600 + }, + { + "epoch": 145.82, + "learning_rate": 3.3295555901853045e-05, + "loss": 2.375, + "step": 72620 + }, + { + "epoch": 145.86, + "learning_rate": 3.3253261469839624e-05, + "loss": 2.3819, + "step": 72640 + }, + { + "epoch": 145.9, + "learning_rate": 3.321098053037097e-05, + "loss": 2.3874, + "step": 72660 + }, + { + "epoch": 145.94, + "learning_rate": 3.316871311751205e-05, + "loss": 2.381, + "step": 72680 + }, + { + "epoch": 145.98, + "learning_rate": 3.312645926531696e-05, + "loss": 2.3723, + "step": 72700 + }, + { + "epoch": 146.02, + "learning_rate": 3.3084219007828864e-05, + "loss": 2.351, + "step": 72720 + }, + { + "epoch": 146.06, + "learning_rate": 3.304199237907999e-05, + "loss": 2.3335, + "step": 72740 + }, + { + "epoch": 146.1, + "learning_rate": 3.299977941309154e-05, + "loss": 2.3392, + "step": 72760 + }, + { + "epoch": 146.14, + "learning_rate": 3.295968978151144e-05, + "loss": 2.3487, + "step": 72780 + }, + { + "epoch": 146.18, + "learning_rate": 3.291750355571772e-05, + "loss": 2.3399, + "step": 72800 + }, + { + "epoch": 146.22, + "learning_rate": 3.287533109298283e-05, + "loss": 2.3411, + "step": 72820 + }, + { + "epoch": 146.27, + "learning_rate": 3.283317242728437e-05, + "loss": 2.3471, + "step": 72840 + }, + { + "epoch": 146.31, + "learning_rate": 3.279102759258882e-05, + "loss": 2.3482, + "step": 72860 + }, + { + "epoch": 146.35, + "learning_rate": 3.27488966228515e-05, + "loss": 2.3386, + "step": 72880 + }, + { + "epoch": 146.39, + "learning_rate": 3.2706779552016595e-05, + "loss": 2.3406, + "step": 72900 + }, + { + "epoch": 146.43, + "learning_rate": 3.266467641401703e-05, + "loss": 2.3579, + "step": 72920 + }, + { + "epoch": 146.47, + "learning_rate": 3.262258724277459e-05, + "loss": 2.3561, + "step": 72940 + }, + { + "epoch": 146.51, + "learning_rate": 3.258051207219971e-05, + "loss": 2.3595, + "step": 72960 + }, + { + "epoch": 146.55, + "learning_rate": 3.253845093619163e-05, + "loss": 2.3523, + "step": 72980 + }, + { + "epoch": 146.59, + "learning_rate": 3.249640386863825e-05, + "loss": 2.3605, + "step": 73000 + }, + { + "epoch": 146.59, + "eval_loss": 2.865365982055664, + "eval_runtime": 34.6395, + "eval_samples_per_second": 22.113, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.005125733699059373, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00011566695521148929, + "step": 73000 + }, + { + "epoch": 146.63, + "learning_rate": 3.24543709034161e-05, + "loss": 2.3598, + "step": 73020 + }, + { + "epoch": 146.67, + "learning_rate": 3.2412352074390415e-05, + "loss": 2.3708, + "step": 73040 + }, + { + "epoch": 146.71, + "learning_rate": 3.237034741541499e-05, + "loss": 2.375, + "step": 73060 + }, + { + "epoch": 146.75, + "learning_rate": 3.232835696033222e-05, + "loss": 2.3596, + "step": 73080 + }, + { + "epoch": 146.79, + "learning_rate": 3.228638074297304e-05, + "loss": 2.3646, + "step": 73100 + }, + { + "epoch": 146.83, + "learning_rate": 3.224441879715699e-05, + "loss": 2.3549, + "step": 73120 + }, + { + "epoch": 146.87, + "learning_rate": 3.220247115669198e-05, + "loss": 2.3619, + "step": 73140 + }, + { + "epoch": 146.91, + "learning_rate": 3.216053785537449e-05, + "loss": 2.3672, + "step": 73160 + }, + { + "epoch": 146.95, + "learning_rate": 3.2118618926989466e-05, + "loss": 2.3838, + "step": 73180 + }, + { + "epoch": 146.99, + "learning_rate": 3.2076714405310174e-05, + "loss": 2.3729, + "step": 73200 + }, + { + "epoch": 147.03, + "learning_rate": 3.203482432409836e-05, + "loss": 2.3313, + "step": 73220 + }, + { + "epoch": 147.07, + "learning_rate": 3.1992948717104085e-05, + "loss": 2.3408, + "step": 73240 + }, + { + "epoch": 147.11, + "learning_rate": 3.19510876180658e-05, + "loss": 2.3274, + "step": 73260 + }, + { + "epoch": 147.15, + "learning_rate": 3.1909241060710194e-05, + "loss": 2.335, + "step": 73280 + }, + { + "epoch": 147.19, + "learning_rate": 3.1867409078752316e-05, + "loss": 2.3488, + "step": 73300 + }, + { + "epoch": 147.23, + "learning_rate": 3.182559170589544e-05, + "loss": 2.328, + "step": 73320 + }, + { + "epoch": 147.27, + "learning_rate": 3.178378897583105e-05, + "loss": 2.3364, + "step": 73340 + }, + { + "epoch": 147.31, + "learning_rate": 3.174200092223887e-05, + "loss": 2.3427, + "step": 73360 + }, + { + "epoch": 147.35, + "learning_rate": 3.1700227578786746e-05, + "loss": 2.3488, + "step": 73380 + }, + { + "epoch": 147.39, + "learning_rate": 3.165846897913075e-05, + "loss": 2.3309, + "step": 73400 + }, + { + "epoch": 147.43, + "learning_rate": 3.1616725156914995e-05, + "loss": 2.3364, + "step": 73420 + }, + { + "epoch": 147.47, + "learning_rate": 3.157499614577173e-05, + "loss": 2.3452, + "step": 73440 + }, + { + "epoch": 147.51, + "learning_rate": 3.1533281979321306e-05, + "loss": 2.3561, + "step": 73460 + }, + { + "epoch": 147.55, + "learning_rate": 3.149158269117202e-05, + "loss": 2.3621, + "step": 73480 + }, + { + "epoch": 147.59, + "learning_rate": 3.144989831492026e-05, + "loss": 2.3614, + "step": 73500 + }, + { + "epoch": 147.63, + "learning_rate": 3.140822888415038e-05, + "loss": 2.3532, + "step": 73520 + }, + { + "epoch": 147.67, + "learning_rate": 3.136657443243468e-05, + "loss": 2.3572, + "step": 73540 + }, + { + "epoch": 147.71, + "learning_rate": 3.132493499333339e-05, + "loss": 2.369, + "step": 73560 + }, + { + "epoch": 147.75, + "learning_rate": 3.1283310600394666e-05, + "loss": 2.3734, + "step": 73580 + }, + { + "epoch": 147.79, + "learning_rate": 3.1241701287154536e-05, + "loss": 2.3577, + "step": 73600 + }, + { + "epoch": 147.83, + "learning_rate": 3.120010708713684e-05, + "loss": 2.352, + "step": 73620 + }, + { + "epoch": 147.87, + "learning_rate": 3.115852803385334e-05, + "loss": 2.3567, + "step": 73640 + }, + { + "epoch": 147.91, + "learning_rate": 3.111696416080344e-05, + "loss": 2.381, + "step": 73660 + }, + { + "epoch": 147.95, + "learning_rate": 3.107541550147448e-05, + "loss": 2.3679, + "step": 73680 + }, + { + "epoch": 147.99, + "learning_rate": 3.1033882089341415e-05, + "loss": 2.3778, + "step": 73700 + }, + { + "epoch": 148.03, + "learning_rate": 3.0992363957866975e-05, + "loss": 2.3118, + "step": 73720 + }, + { + "epoch": 148.07, + "learning_rate": 3.09508611405016e-05, + "loss": 2.3122, + "step": 73740 + }, + { + "epoch": 148.11, + "learning_rate": 3.090937367068331e-05, + "loss": 2.3385, + "step": 73760 + }, + { + "epoch": 148.15, + "learning_rate": 3.0867901581837854e-05, + "loss": 2.317, + "step": 73780 + }, + { + "epoch": 148.19, + "learning_rate": 3.08264449073785e-05, + "loss": 2.3309, + "step": 73800 + }, + { + "epoch": 148.23, + "learning_rate": 3.078500368070621e-05, + "loss": 2.3407, + "step": 73820 + }, + { + "epoch": 148.27, + "learning_rate": 3.0743577935209353e-05, + "loss": 2.3561, + "step": 73840 + }, + { + "epoch": 148.31, + "learning_rate": 3.0702167704263954e-05, + "loss": 2.3425, + "step": 73860 + }, + { + "epoch": 148.35, + "learning_rate": 3.066077302123348e-05, + "loss": 2.3456, + "step": 73880 + }, + { + "epoch": 148.39, + "learning_rate": 3.061939391946886e-05, + "loss": 2.3569, + "step": 73900 + }, + { + "epoch": 148.43, + "learning_rate": 3.0578030432308526e-05, + "loss": 2.3342, + "step": 73920 + }, + { + "epoch": 148.47, + "learning_rate": 3.0536682593078246e-05, + "loss": 2.3421, + "step": 73940 + }, + { + "epoch": 148.51, + "learning_rate": 3.0495350435091273e-05, + "loss": 2.3442, + "step": 73960 + }, + { + "epoch": 148.55, + "learning_rate": 3.0454033991648147e-05, + "loss": 2.3444, + "step": 73980 + }, + { + "epoch": 148.59, + "learning_rate": 3.0412733296036776e-05, + "loss": 2.3446, + "step": 74000 + }, + { + "epoch": 148.59, + "eval_loss": 2.861361265182495, + "eval_runtime": 32.4285, + "eval_samples_per_second": 23.621, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004920170077563462, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00020851465042065565, + "step": 74000 + }, + { + "epoch": 148.63, + "learning_rate": 3.0371448381532418e-05, + "loss": 2.3528, + "step": 74020 + }, + { + "epoch": 148.67, + "learning_rate": 3.0330179281397565e-05, + "loss": 2.3536, + "step": 74040 + }, + { + "epoch": 148.71, + "learning_rate": 3.0288926028881986e-05, + "loss": 2.3618, + "step": 74060 + }, + { + "epoch": 148.76, + "learning_rate": 3.024768865722267e-05, + "loss": 2.3602, + "step": 74080 + }, + { + "epoch": 148.8, + "learning_rate": 3.0206467199643852e-05, + "loss": 2.3514, + "step": 74100 + }, + { + "epoch": 148.84, + "learning_rate": 3.0165261689356884e-05, + "loss": 2.3575, + "step": 74120 + }, + { + "epoch": 148.88, + "learning_rate": 3.01240721595603e-05, + "loss": 2.3668, + "step": 74140 + }, + { + "epoch": 148.92, + "learning_rate": 3.0082898643439804e-05, + "loss": 2.3568, + "step": 74160 + }, + { + "epoch": 148.96, + "learning_rate": 3.0041741174168103e-05, + "loss": 2.3571, + "step": 74180 + }, + { + "epoch": 149.0, + "learning_rate": 3.0000599784905065e-05, + "loss": 2.3523, + "step": 74200 + }, + { + "epoch": 149.04, + "learning_rate": 2.9959474508797535e-05, + "loss": 2.3299, + "step": 74220 + }, + { + "epoch": 149.08, + "learning_rate": 2.991836537897943e-05, + "loss": 2.3093, + "step": 74240 + }, + { + "epoch": 149.12, + "learning_rate": 2.9877272428571613e-05, + "loss": 2.3194, + "step": 74260 + }, + { + "epoch": 149.16, + "learning_rate": 2.9836195690681935e-05, + "loss": 2.3236, + "step": 74280 + }, + { + "epoch": 149.2, + "learning_rate": 2.97951351984052e-05, + "loss": 2.3265, + "step": 74300 + }, + { + "epoch": 149.24, + "learning_rate": 2.9754090984823073e-05, + "loss": 2.322, + "step": 74320 + }, + { + "epoch": 149.28, + "learning_rate": 2.9713063083004188e-05, + "loss": 2.338, + "step": 74340 + }, + { + "epoch": 149.32, + "learning_rate": 2.9672051526003912e-05, + "loss": 2.3292, + "step": 74360 + }, + { + "epoch": 149.36, + "learning_rate": 2.9631056346864594e-05, + "loss": 2.331, + "step": 74380 + }, + { + "epoch": 149.4, + "learning_rate": 2.9590077578615234e-05, + "loss": 2.3382, + "step": 74400 + }, + { + "epoch": 149.44, + "learning_rate": 2.954911525427174e-05, + "loss": 2.3405, + "step": 74420 + }, + { + "epoch": 149.48, + "learning_rate": 2.9508169406836696e-05, + "loss": 2.3456, + "step": 74440 + }, + { + "epoch": 149.52, + "learning_rate": 2.946724006929943e-05, + "loss": 2.3395, + "step": 74460 + }, + { + "epoch": 149.56, + "learning_rate": 2.942632727463598e-05, + "loss": 2.3486, + "step": 74480 + }, + { + "epoch": 149.6, + "learning_rate": 2.9385431055809032e-05, + "loss": 2.3442, + "step": 74500 + }, + { + "epoch": 149.64, + "learning_rate": 2.9344551445767954e-05, + "loss": 2.3505, + "step": 74520 + }, + { + "epoch": 149.68, + "learning_rate": 2.9303688477448667e-05, + "loss": 2.342, + "step": 74540 + }, + { + "epoch": 149.72, + "learning_rate": 2.9262842183773788e-05, + "loss": 2.3538, + "step": 74560 + }, + { + "epoch": 149.76, + "learning_rate": 2.9222012597652372e-05, + "loss": 2.3665, + "step": 74580 + }, + { + "epoch": 149.8, + "learning_rate": 2.9181199751980114e-05, + "loss": 2.3614, + "step": 74600 + }, + { + "epoch": 149.84, + "learning_rate": 2.914040367963921e-05, + "loss": 2.3458, + "step": 74620 + }, + { + "epoch": 149.88, + "learning_rate": 2.909962441349825e-05, + "loss": 2.3492, + "step": 74640 + }, + { + "epoch": 149.92, + "learning_rate": 2.905886198641244e-05, + "loss": 2.3413, + "step": 74660 + }, + { + "epoch": 149.96, + "learning_rate": 2.9018116431223257e-05, + "loss": 2.3673, + "step": 74680 + }, + { + "epoch": 150.0, + "learning_rate": 2.897738778075869e-05, + "loss": 2.3663, + "step": 74700 + }, + { + "epoch": 150.04, + "learning_rate": 2.893667606783308e-05, + "loss": 2.3182, + "step": 74720 + }, + { + "epoch": 150.08, + "learning_rate": 2.88959813252471e-05, + "loss": 2.3119, + "step": 74740 + }, + { + "epoch": 150.12, + "learning_rate": 2.8855303585787813e-05, + "loss": 2.3159, + "step": 74760 + }, + { + "epoch": 150.16, + "learning_rate": 2.881464288222846e-05, + "loss": 2.3223, + "step": 74780 + }, + { + "epoch": 150.2, + "learning_rate": 2.877399924732873e-05, + "loss": 2.323, + "step": 74800 + }, + { + "epoch": 150.24, + "learning_rate": 2.8733372713834396e-05, + "loss": 2.3233, + "step": 74820 + }, + { + "epoch": 150.28, + "learning_rate": 2.8692763314477566e-05, + "loss": 2.3223, + "step": 74840 + }, + { + "epoch": 150.32, + "learning_rate": 2.8652171081976435e-05, + "loss": 2.3147, + "step": 74860 + }, + { + "epoch": 150.36, + "learning_rate": 2.86115960490355e-05, + "loss": 2.3233, + "step": 74880 + }, + { + "epoch": 150.4, + "learning_rate": 2.857103824834533e-05, + "loss": 2.3312, + "step": 74900 + }, + { + "epoch": 150.44, + "learning_rate": 2.853049771258256e-05, + "loss": 2.3383, + "step": 74920 + }, + { + "epoch": 150.48, + "learning_rate": 2.8489974474410003e-05, + "loss": 2.3345, + "step": 74940 + }, + { + "epoch": 150.52, + "learning_rate": 2.844946856647649e-05, + "loss": 2.3286, + "step": 74960 + }, + { + "epoch": 150.56, + "learning_rate": 2.840898002141693e-05, + "loss": 2.3599, + "step": 74980 + }, + { + "epoch": 150.6, + "learning_rate": 2.836850887185214e-05, + "loss": 2.328, + "step": 75000 + }, + { + "epoch": 150.6, + "eval_loss": 2.8561673164367676, + "eval_runtime": 32.3505, + "eval_samples_per_second": 23.678, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.0050671485285182965, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0004443041759831747, + "step": 75000 + }, + { + "epoch": 150.64, + "learning_rate": 2.8328055150389064e-05, + "loss": 2.3322, + "step": 75020 + }, + { + "epoch": 150.68, + "learning_rate": 2.8287618889620537e-05, + "loss": 2.3363, + "step": 75040 + }, + { + "epoch": 150.72, + "learning_rate": 2.824720012212527e-05, + "loss": 2.3536, + "step": 75060 + }, + { + "epoch": 150.76, + "learning_rate": 2.8206798880467967e-05, + "loss": 2.3535, + "step": 75080 + }, + { + "epoch": 150.8, + "learning_rate": 2.816641519719917e-05, + "loss": 2.3523, + "step": 75100 + }, + { + "epoch": 150.84, + "learning_rate": 2.81260491048553e-05, + "loss": 2.3521, + "step": 75120 + }, + { + "epoch": 150.88, + "learning_rate": 2.8087717640344618e-05, + "loss": 2.3606, + "step": 75140 + }, + { + "epoch": 150.92, + "learning_rate": 2.804738594383345e-05, + "loss": 2.3671, + "step": 75160 + }, + { + "epoch": 150.96, + "learning_rate": 2.8007071934146937e-05, + "loss": 2.3512, + "step": 75180 + }, + { + "epoch": 151.0, + "learning_rate": 2.7966775643765307e-05, + "loss": 2.3404, + "step": 75200 + }, + { + "epoch": 151.04, + "learning_rate": 2.792649710515458e-05, + "loss": 2.3077, + "step": 75220 + }, + { + "epoch": 151.08, + "learning_rate": 2.7886236350766433e-05, + "loss": 2.2967, + "step": 75240 + }, + { + "epoch": 151.12, + "learning_rate": 2.7845993413038234e-05, + "loss": 2.3055, + "step": 75260 + }, + { + "epoch": 151.16, + "learning_rate": 2.7805768324393015e-05, + "loss": 2.309, + "step": 75280 + }, + { + "epoch": 151.2, + "learning_rate": 2.7765561117239342e-05, + "loss": 2.3143, + "step": 75300 + }, + { + "epoch": 151.24, + "learning_rate": 2.7725371823971514e-05, + "loss": 2.3325, + "step": 75320 + }, + { + "epoch": 151.29, + "learning_rate": 2.7685200476969257e-05, + "loss": 2.3213, + "step": 75340 + }, + { + "epoch": 151.33, + "learning_rate": 2.7645047108597927e-05, + "loss": 2.316, + "step": 75360 + }, + { + "epoch": 151.37, + "learning_rate": 2.760491175120836e-05, + "loss": 2.3486, + "step": 75380 + }, + { + "epoch": 151.41, + "learning_rate": 2.7564794437136893e-05, + "loss": 2.3276, + "step": 75400 + }, + { + "epoch": 151.45, + "learning_rate": 2.7524695198705334e-05, + "loss": 2.3263, + "step": 75420 + }, + { + "epoch": 151.49, + "learning_rate": 2.748461406822087e-05, + "loss": 2.3389, + "step": 75440 + }, + { + "epoch": 151.53, + "learning_rate": 2.7444551077976204e-05, + "loss": 2.3315, + "step": 75460 + }, + { + "epoch": 151.57, + "learning_rate": 2.740450626024932e-05, + "loss": 2.334, + "step": 75480 + }, + { + "epoch": 151.61, + "learning_rate": 2.7364479647303643e-05, + "loss": 2.3291, + "step": 75500 + }, + { + "epoch": 151.65, + "learning_rate": 2.7324471271387818e-05, + "loss": 2.3449, + "step": 75520 + }, + { + "epoch": 151.69, + "learning_rate": 2.7284481164735942e-05, + "loss": 2.3287, + "step": 75540 + }, + { + "epoch": 151.73, + "learning_rate": 2.7244509359567327e-05, + "loss": 2.3464, + "step": 75560 + }, + { + "epoch": 151.77, + "learning_rate": 2.720455588808648e-05, + "loss": 2.3468, + "step": 75580 + }, + { + "epoch": 151.81, + "learning_rate": 2.7164620782483225e-05, + "loss": 2.3301, + "step": 75600 + }, + { + "epoch": 151.85, + "learning_rate": 2.7124704074932545e-05, + "loss": 2.3451, + "step": 75620 + }, + { + "epoch": 151.89, + "learning_rate": 2.708480579759464e-05, + "loss": 2.3496, + "step": 75640 + }, + { + "epoch": 151.93, + "learning_rate": 2.7044925982614754e-05, + "loss": 2.3412, + "step": 75660 + }, + { + "epoch": 151.97, + "learning_rate": 2.7005064662123413e-05, + "loss": 2.3649, + "step": 75680 + }, + { + "epoch": 152.01, + "learning_rate": 2.6965221868236155e-05, + "loss": 2.3329, + "step": 75700 + }, + { + "epoch": 152.05, + "learning_rate": 2.6925397633053563e-05, + "loss": 2.3069, + "step": 75720 + }, + { + "epoch": 152.09, + "learning_rate": 2.688559198866132e-05, + "loss": 2.311, + "step": 75740 + }, + { + "epoch": 152.13, + "learning_rate": 2.684580496713011e-05, + "loss": 2.3094, + "step": 75760 + }, + { + "epoch": 152.17, + "learning_rate": 2.680603660051565e-05, + "loss": 2.3071, + "step": 75780 + }, + { + "epoch": 152.21, + "learning_rate": 2.676628692085854e-05, + "loss": 2.309, + "step": 75800 + }, + { + "epoch": 152.25, + "learning_rate": 2.6726555960184392e-05, + "loss": 2.3105, + "step": 75820 + }, + { + "epoch": 152.29, + "learning_rate": 2.6686843750503776e-05, + "loss": 2.318, + "step": 75840 + }, + { + "epoch": 152.33, + "learning_rate": 2.6647150323812044e-05, + "loss": 2.3183, + "step": 75860 + }, + { + "epoch": 152.37, + "learning_rate": 2.660747571208949e-05, + "loss": 2.3096, + "step": 75880 + }, + { + "epoch": 152.41, + "learning_rate": 2.6567819947301243e-05, + "loss": 2.3407, + "step": 75900 + }, + { + "epoch": 152.45, + "learning_rate": 2.6528183061397242e-05, + "loss": 2.3339, + "step": 75920 + }, + { + "epoch": 152.49, + "learning_rate": 2.648856508631218e-05, + "loss": 2.3301, + "step": 75940 + }, + { + "epoch": 152.53, + "learning_rate": 2.644896605396554e-05, + "loss": 2.3319, + "step": 75960 + }, + { + "epoch": 152.57, + "learning_rate": 2.6409385996261603e-05, + "loss": 2.3449, + "step": 75980 + }, + { + "epoch": 152.61, + "learning_rate": 2.636982494508926e-05, + "loss": 2.3312, + "step": 76000 + }, + { + "epoch": 152.61, + "eval_loss": 2.8495240211486816, + "eval_runtime": 34.2618, + "eval_samples_per_second": 22.357, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.005261460130357665, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00022891282274441546, + "step": 76000 + }, + { + "epoch": 152.65, + "learning_rate": 2.633028293232216e-05, + "loss": 2.3277, + "step": 76020 + }, + { + "epoch": 152.69, + "learning_rate": 2.6290759989818535e-05, + "loss": 2.3325, + "step": 76040 + }, + { + "epoch": 152.73, + "learning_rate": 2.625125614942139e-05, + "loss": 2.3371, + "step": 76060 + }, + { + "epoch": 152.77, + "learning_rate": 2.621177144295819e-05, + "loss": 2.3324, + "step": 76080 + }, + { + "epoch": 152.81, + "learning_rate": 2.6172305902241068e-05, + "loss": 2.35, + "step": 76100 + }, + { + "epoch": 152.85, + "learning_rate": 2.6132859559066702e-05, + "loss": 2.3158, + "step": 76120 + }, + { + "epoch": 152.89, + "learning_rate": 2.60934324452163e-05, + "loss": 2.3312, + "step": 76140 + }, + { + "epoch": 152.93, + "learning_rate": 2.605402459245559e-05, + "loss": 2.343, + "step": 76160 + }, + { + "epoch": 152.97, + "learning_rate": 2.601463603253471e-05, + "loss": 2.359, + "step": 76180 + }, + { + "epoch": 153.01, + "learning_rate": 2.597526679718839e-05, + "loss": 2.3368, + "step": 76200 + }, + { + "epoch": 153.05, + "learning_rate": 2.593591691813565e-05, + "loss": 2.2886, + "step": 76220 + }, + { + "epoch": 153.09, + "learning_rate": 2.5896586427080006e-05, + "loss": 2.3059, + "step": 76240 + }, + { + "epoch": 153.13, + "learning_rate": 2.5857275355709317e-05, + "loss": 2.291, + "step": 76260 + }, + { + "epoch": 153.17, + "learning_rate": 2.58179837356958e-05, + "loss": 2.3221, + "step": 76280 + }, + { + "epoch": 153.21, + "learning_rate": 2.577871159869603e-05, + "loss": 2.3197, + "step": 76300 + }, + { + "epoch": 153.25, + "learning_rate": 2.5739458976350817e-05, + "loss": 2.3179, + "step": 76320 + }, + { + "epoch": 153.29, + "learning_rate": 2.570022590028531e-05, + "loss": 2.3171, + "step": 76340 + }, + { + "epoch": 153.33, + "learning_rate": 2.5661012402108885e-05, + "loss": 2.3144, + "step": 76360 + }, + { + "epoch": 153.37, + "learning_rate": 2.5621818513415152e-05, + "loss": 2.3125, + "step": 76380 + }, + { + "epoch": 153.41, + "learning_rate": 2.558264426578192e-05, + "loss": 2.3054, + "step": 76400 + }, + { + "epoch": 153.45, + "learning_rate": 2.554348969077117e-05, + "loss": 2.3125, + "step": 76420 + }, + { + "epoch": 153.49, + "learning_rate": 2.5504354819929065e-05, + "loss": 2.3311, + "step": 76440 + }, + { + "epoch": 153.53, + "learning_rate": 2.5465239684785817e-05, + "loss": 2.3359, + "step": 76460 + }, + { + "epoch": 153.57, + "learning_rate": 2.5426144316855805e-05, + "loss": 2.3319, + "step": 76480 + }, + { + "epoch": 153.61, + "learning_rate": 2.5387068747637455e-05, + "loss": 2.3162, + "step": 76500 + }, + { + "epoch": 153.65, + "learning_rate": 2.5348013008613252e-05, + "loss": 2.3325, + "step": 76520 + }, + { + "epoch": 153.69, + "learning_rate": 2.5308977131249722e-05, + "loss": 2.3404, + "step": 76540 + }, + { + "epoch": 153.73, + "learning_rate": 2.52699611469973e-05, + "loss": 2.3168, + "step": 76560 + }, + { + "epoch": 153.78, + "learning_rate": 2.5230965087290547e-05, + "loss": 2.3416, + "step": 76580 + }, + { + "epoch": 153.82, + "learning_rate": 2.5191988983547827e-05, + "loss": 2.3284, + "step": 76600 + }, + { + "epoch": 153.86, + "learning_rate": 2.515303286717149e-05, + "loss": 2.3379, + "step": 76620 + }, + { + "epoch": 153.9, + "learning_rate": 2.511409676954778e-05, + "loss": 2.3335, + "step": 76640 + }, + { + "epoch": 153.94, + "learning_rate": 2.507518072204683e-05, + "loss": 2.3378, + "step": 76660 + }, + { + "epoch": 153.98, + "learning_rate": 2.503628475602256e-05, + "loss": 2.3477, + "step": 76680 + }, + { + "epoch": 154.02, + "learning_rate": 2.499740890281274e-05, + "loss": 2.3212, + "step": 76700 + }, + { + "epoch": 154.06, + "learning_rate": 2.495855319373901e-05, + "loss": 2.3083, + "step": 76720 + }, + { + "epoch": 154.1, + "learning_rate": 2.4919717660106645e-05, + "loss": 2.2846, + "step": 76740 + }, + { + "epoch": 154.14, + "learning_rate": 2.488090233320476e-05, + "loss": 2.3017, + "step": 76760 + }, + { + "epoch": 154.18, + "learning_rate": 2.4842107244306155e-05, + "loss": 2.2984, + "step": 76780 + }, + { + "epoch": 154.22, + "learning_rate": 2.480333242466736e-05, + "loss": 2.3059, + "step": 76800 + }, + { + "epoch": 154.26, + "learning_rate": 2.47645779055285e-05, + "loss": 2.3069, + "step": 76820 + }, + { + "epoch": 154.3, + "learning_rate": 2.4725843718113422e-05, + "loss": 2.3019, + "step": 76840 + }, + { + "epoch": 154.34, + "learning_rate": 2.468712989362955e-05, + "loss": 2.3119, + "step": 76860 + }, + { + "epoch": 154.38, + "learning_rate": 2.464843646326792e-05, + "loss": 2.3225, + "step": 76880 + }, + { + "epoch": 154.42, + "learning_rate": 2.4609763458203145e-05, + "loss": 2.3239, + "step": 76900 + }, + { + "epoch": 154.46, + "learning_rate": 2.457111090959331e-05, + "loss": 2.3037, + "step": 76920 + }, + { + "epoch": 154.5, + "learning_rate": 2.4532478848580154e-05, + "loss": 2.3383, + "step": 76940 + }, + { + "epoch": 154.54, + "learning_rate": 2.449386730628877e-05, + "loss": 2.3236, + "step": 76960 + }, + { + "epoch": 154.58, + "learning_rate": 2.44552763138278e-05, + "loss": 2.3061, + "step": 76980 + }, + { + "epoch": 154.62, + "learning_rate": 2.4416705902289318e-05, + "loss": 2.3207, + "step": 77000 + }, + { + "epoch": 154.62, + "eval_loss": 2.843654155731201, + "eval_runtime": 34.0767, + "eval_samples_per_second": 22.479, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.004635744243968971, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0005488427871293591, + "step": 77000 + }, + { + "epoch": 154.66, + "learning_rate": 2.4378156102748796e-05, + "loss": 2.3291, + "step": 77020 + }, + { + "epoch": 154.7, + "learning_rate": 2.4339626946265138e-05, + "loss": 2.3362, + "step": 77040 + }, + { + "epoch": 154.74, + "learning_rate": 2.430111846388054e-05, + "loss": 2.3299, + "step": 77060 + }, + { + "epoch": 154.78, + "learning_rate": 2.4262630686620624e-05, + "loss": 2.334, + "step": 77080 + }, + { + "epoch": 154.82, + "learning_rate": 2.422416364549429e-05, + "loss": 2.3244, + "step": 77100 + }, + { + "epoch": 154.86, + "learning_rate": 2.4185717371493744e-05, + "loss": 2.3211, + "step": 77120 + }, + { + "epoch": 154.9, + "learning_rate": 2.4147291895594444e-05, + "loss": 2.3366, + "step": 77140 + }, + { + "epoch": 154.94, + "learning_rate": 2.4108887248755115e-05, + "loss": 2.3346, + "step": 77160 + }, + { + "epoch": 154.98, + "learning_rate": 2.4070503461917692e-05, + "loss": 2.3304, + "step": 77180 + }, + { + "epoch": 155.02, + "learning_rate": 2.4032140566007266e-05, + "loss": 2.313, + "step": 77200 + }, + { + "epoch": 155.06, + "learning_rate": 2.3993798591932155e-05, + "loss": 2.2918, + "step": 77220 + }, + { + "epoch": 155.1, + "learning_rate": 2.395547757058379e-05, + "loss": 2.2871, + "step": 77240 + }, + { + "epoch": 155.14, + "learning_rate": 2.3917177532836727e-05, + "loss": 2.291, + "step": 77260 + }, + { + "epoch": 155.18, + "learning_rate": 2.3878898509548608e-05, + "loss": 2.2988, + "step": 77280 + }, + { + "epoch": 155.22, + "learning_rate": 2.384064053156016e-05, + "loss": 2.294, + "step": 77300 + }, + { + "epoch": 155.26, + "learning_rate": 2.3802403629695158e-05, + "loss": 2.3133, + "step": 77320 + }, + { + "epoch": 155.3, + "learning_rate": 2.3764187834760348e-05, + "loss": 2.2792, + "step": 77340 + }, + { + "epoch": 155.34, + "learning_rate": 2.3725993177545526e-05, + "loss": 2.3154, + "step": 77360 + }, + { + "epoch": 155.38, + "learning_rate": 2.368781968882343e-05, + "loss": 2.3188, + "step": 77380 + }, + { + "epoch": 155.42, + "learning_rate": 2.364966739934976e-05, + "loss": 2.3299, + "step": 77400 + }, + { + "epoch": 155.46, + "learning_rate": 2.3611536339863145e-05, + "loss": 2.2995, + "step": 77420 + }, + { + "epoch": 155.5, + "learning_rate": 2.3573426541085024e-05, + "loss": 2.3165, + "step": 77440 + }, + { + "epoch": 155.54, + "learning_rate": 2.3535338033719862e-05, + "loss": 2.3264, + "step": 77460 + }, + { + "epoch": 155.58, + "learning_rate": 2.349727084845481e-05, + "loss": 2.3309, + "step": 77480 + }, + { + "epoch": 155.62, + "learning_rate": 2.345922501595994e-05, + "loss": 2.3094, + "step": 77500 + }, + { + "epoch": 155.66, + "learning_rate": 2.3421200566888095e-05, + "loss": 2.3207, + "step": 77520 + }, + { + "epoch": 155.7, + "learning_rate": 2.338319753187488e-05, + "loss": 2.3193, + "step": 77540 + }, + { + "epoch": 155.74, + "learning_rate": 2.334521594153868e-05, + "loss": 2.3374, + "step": 77560 + }, + { + "epoch": 155.78, + "learning_rate": 2.330725582648053e-05, + "loss": 2.3233, + "step": 77580 + }, + { + "epoch": 155.82, + "learning_rate": 2.3269317217284226e-05, + "loss": 2.3091, + "step": 77600 + }, + { + "epoch": 155.86, + "learning_rate": 2.323140014451623e-05, + "loss": 2.3384, + "step": 77620 + }, + { + "epoch": 155.9, + "learning_rate": 2.319350463872566e-05, + "loss": 2.3326, + "step": 77640 + }, + { + "epoch": 155.94, + "learning_rate": 2.3155630730444183e-05, + "loss": 2.3257, + "step": 77660 + }, + { + "epoch": 155.98, + "learning_rate": 2.3117778450186174e-05, + "loss": 2.328, + "step": 77680 + }, + { + "epoch": 156.02, + "learning_rate": 2.3079947828448534e-05, + "loss": 2.3161, + "step": 77700 + }, + { + "epoch": 156.06, + "learning_rate": 2.304213889571068e-05, + "loss": 2.2837, + "step": 77720 + }, + { + "epoch": 156.1, + "learning_rate": 2.30043516824346e-05, + "loss": 2.2878, + "step": 77740 + }, + { + "epoch": 156.14, + "learning_rate": 2.2966586219064766e-05, + "loss": 2.2998, + "step": 77760 + }, + { + "epoch": 156.18, + "learning_rate": 2.292884253602815e-05, + "loss": 2.3067, + "step": 77780 + }, + { + "epoch": 156.22, + "learning_rate": 2.2891120663734106e-05, + "loss": 2.298, + "step": 77800 + }, + { + "epoch": 156.27, + "learning_rate": 2.2855305114936586e-05, + "loss": 2.2976, + "step": 77820 + }, + { + "epoch": 156.31, + "learning_rate": 2.2817625860989135e-05, + "loss": 2.2924, + "step": 77840 + }, + { + "epoch": 156.35, + "learning_rate": 2.277996850738954e-05, + "loss": 2.2997, + "step": 77860 + }, + { + "epoch": 156.39, + "learning_rate": 2.2742333084477656e-05, + "loss": 2.3096, + "step": 77880 + }, + { + "epoch": 156.43, + "learning_rate": 2.2704719622575682e-05, + "loss": 2.3074, + "step": 77900 + }, + { + "epoch": 156.47, + "learning_rate": 2.2667128151988094e-05, + "loss": 2.3016, + "step": 77920 + }, + { + "epoch": 156.51, + "learning_rate": 2.262955870300168e-05, + "loss": 2.3219, + "step": 77940 + }, + { + "epoch": 156.55, + "learning_rate": 2.2592011305885413e-05, + "loss": 2.3177, + "step": 77960 + }, + { + "epoch": 156.59, + "learning_rate": 2.255448599089063e-05, + "loss": 2.3251, + "step": 77980 + }, + { + "epoch": 156.63, + "learning_rate": 2.2516982788250733e-05, + "loss": 2.3145, + "step": 78000 + }, + { + "epoch": 156.63, + "eval_loss": 2.8391897678375244, + "eval_runtime": 34.232, + "eval_samples_per_second": 22.377, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.004465514195501022, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0002722053108502919, + "step": 78000 + }, + { + "epoch": 156.67, + "learning_rate": 2.2479501728181407e-05, + "loss": 2.3125, + "step": 78020 + }, + { + "epoch": 156.71, + "learning_rate": 2.2442042840880457e-05, + "loss": 2.3142, + "step": 78040 + }, + { + "epoch": 156.75, + "learning_rate": 2.2404606156527835e-05, + "loss": 2.3074, + "step": 78060 + }, + { + "epoch": 156.79, + "learning_rate": 2.2367191705285616e-05, + "loss": 2.3222, + "step": 78080 + }, + { + "epoch": 156.83, + "learning_rate": 2.2329799517297896e-05, + "loss": 2.3233, + "step": 78100 + }, + { + "epoch": 156.87, + "learning_rate": 2.2292429622690958e-05, + "loss": 2.3116, + "step": 78120 + }, + { + "epoch": 156.91, + "learning_rate": 2.2255082051572994e-05, + "loss": 2.3242, + "step": 78140 + }, + { + "epoch": 156.95, + "learning_rate": 2.2217756834034287e-05, + "loss": 2.329, + "step": 78160 + }, + { + "epoch": 156.99, + "learning_rate": 2.2180454000147095e-05, + "loss": 2.3197, + "step": 78180 + }, + { + "epoch": 157.03, + "learning_rate": 2.2143173579965632e-05, + "loss": 2.2989, + "step": 78200 + }, + { + "epoch": 157.07, + "learning_rate": 2.2105915603526078e-05, + "loss": 2.2976, + "step": 78220 + }, + { + "epoch": 157.11, + "learning_rate": 2.206868010084648e-05, + "loss": 2.2872, + "step": 78240 + }, + { + "epoch": 157.15, + "learning_rate": 2.2031467101926816e-05, + "loss": 2.2849, + "step": 78260 + }, + { + "epoch": 157.19, + "learning_rate": 2.1994276636748946e-05, + "loss": 2.3003, + "step": 78280 + }, + { + "epoch": 157.23, + "learning_rate": 2.195710873527656e-05, + "loss": 2.2943, + "step": 78300 + }, + { + "epoch": 157.27, + "learning_rate": 2.1919963427455115e-05, + "loss": 2.2919, + "step": 78320 + }, + { + "epoch": 157.31, + "learning_rate": 2.1882840743211962e-05, + "loss": 2.3043, + "step": 78340 + }, + { + "epoch": 157.35, + "learning_rate": 2.1845740712456193e-05, + "loss": 2.3049, + "step": 78360 + }, + { + "epoch": 157.39, + "learning_rate": 2.180866336507858e-05, + "loss": 2.2987, + "step": 78380 + }, + { + "epoch": 157.43, + "learning_rate": 2.17716087309517e-05, + "loss": 2.3126, + "step": 78400 + }, + { + "epoch": 157.47, + "learning_rate": 2.1734576839929798e-05, + "loss": 2.2964, + "step": 78420 + }, + { + "epoch": 157.51, + "learning_rate": 2.1697567721848823e-05, + "loss": 2.2987, + "step": 78440 + }, + { + "epoch": 157.55, + "learning_rate": 2.166058140652631e-05, + "loss": 2.3111, + "step": 78460 + }, + { + "epoch": 157.59, + "learning_rate": 2.1623617923761468e-05, + "loss": 2.3179, + "step": 78480 + }, + { + "epoch": 157.63, + "learning_rate": 2.158667730333516e-05, + "loss": 2.3033, + "step": 78500 + }, + { + "epoch": 157.67, + "learning_rate": 2.1549759575009732e-05, + "loss": 2.3098, + "step": 78520 + }, + { + "epoch": 157.71, + "learning_rate": 2.1512864768529134e-05, + "loss": 2.2999, + "step": 78540 + }, + { + "epoch": 157.75, + "learning_rate": 2.1475992913618858e-05, + "loss": 2.3063, + "step": 78560 + }, + { + "epoch": 157.79, + "learning_rate": 2.1439144039985898e-05, + "loss": 2.3275, + "step": 78580 + }, + { + "epoch": 157.83, + "learning_rate": 2.1402318177318703e-05, + "loss": 2.3097, + "step": 78600 + }, + { + "epoch": 157.87, + "learning_rate": 2.1365515355287192e-05, + "loss": 2.3232, + "step": 78620 + }, + { + "epoch": 157.91, + "learning_rate": 2.13287356035428e-05, + "loss": 2.3334, + "step": 78640 + }, + { + "epoch": 157.95, + "learning_rate": 2.1291978951718246e-05, + "loss": 2.3172, + "step": 78660 + }, + { + "epoch": 157.99, + "learning_rate": 2.1255245429427723e-05, + "loss": 2.3266, + "step": 78680 + }, + { + "epoch": 158.03, + "learning_rate": 2.1218535066266758e-05, + "loss": 2.2872, + "step": 78700 + }, + { + "epoch": 158.07, + "learning_rate": 2.1181847891812262e-05, + "loss": 2.2895, + "step": 78720 + }, + { + "epoch": 158.11, + "learning_rate": 2.1145183935622377e-05, + "loss": 2.2781, + "step": 78740 + }, + { + "epoch": 158.15, + "learning_rate": 2.1108543227236616e-05, + "loss": 2.2809, + "step": 78760 + }, + { + "epoch": 158.19, + "learning_rate": 2.107192579617574e-05, + "loss": 2.2858, + "step": 78780 + }, + { + "epoch": 158.23, + "learning_rate": 2.1035331671941737e-05, + "loss": 2.2886, + "step": 78800 + }, + { + "epoch": 158.27, + "learning_rate": 2.0998760884017865e-05, + "loss": 2.2886, + "step": 78820 + }, + { + "epoch": 158.31, + "learning_rate": 2.0962213461868485e-05, + "loss": 2.2949, + "step": 78840 + }, + { + "epoch": 158.35, + "learning_rate": 2.092568943493927e-05, + "loss": 2.303, + "step": 78860 + }, + { + "epoch": 158.39, + "learning_rate": 2.0889188832656918e-05, + "loss": 2.2988, + "step": 78880 + }, + { + "epoch": 158.43, + "learning_rate": 2.0852711684429314e-05, + "loss": 2.2984, + "step": 78900 + }, + { + "epoch": 158.47, + "learning_rate": 2.0816258019645446e-05, + "loss": 2.3048, + "step": 78920 + }, + { + "epoch": 158.51, + "learning_rate": 2.077982786767537e-05, + "loss": 2.301, + "step": 78940 + }, + { + "epoch": 158.55, + "learning_rate": 2.0743421257870215e-05, + "loss": 2.3215, + "step": 78960 + }, + { + "epoch": 158.59, + "learning_rate": 2.07070382195621e-05, + "loss": 2.2954, + "step": 78980 + }, + { + "epoch": 158.63, + "learning_rate": 2.06706787820642e-05, + "loss": 2.3084, + "step": 79000 + }, + { + "epoch": 158.63, + "eval_loss": 2.8332622051239014, + "eval_runtime": 34.6722, + "eval_samples_per_second": 22.093, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.00472047805454497, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00022219074401334598, + "step": 79000 + }, + { + "epoch": 158.67, + "learning_rate": 2.063434297467065e-05, + "loss": 2.2921, + "step": 79020 + }, + { + "epoch": 158.71, + "learning_rate": 2.0598030826656577e-05, + "loss": 2.3068, + "step": 79040 + }, + { + "epoch": 158.76, + "learning_rate": 2.056174236727797e-05, + "loss": 2.3107, + "step": 79060 + }, + { + "epoch": 158.8, + "learning_rate": 2.052547762577184e-05, + "loss": 2.3153, + "step": 79080 + }, + { + "epoch": 158.84, + "learning_rate": 2.0489236631356023e-05, + "loss": 2.32, + "step": 79100 + }, + { + "epoch": 158.88, + "learning_rate": 2.045301941322921e-05, + "loss": 2.3119, + "step": 79120 + }, + { + "epoch": 158.92, + "learning_rate": 2.041682600057097e-05, + "loss": 2.3194, + "step": 79140 + }, + { + "epoch": 158.96, + "learning_rate": 2.0380656422541687e-05, + "loss": 2.3171, + "step": 79160 + }, + { + "epoch": 159.0, + "learning_rate": 2.0344510708282556e-05, + "loss": 2.3178, + "step": 79180 + }, + { + "epoch": 159.04, + "learning_rate": 2.0308388886915464e-05, + "loss": 2.2746, + "step": 79200 + }, + { + "epoch": 159.08, + "learning_rate": 2.027229098754317e-05, + "loss": 2.2838, + "step": 79220 + }, + { + "epoch": 159.12, + "learning_rate": 2.02362170392491e-05, + "loss": 2.2788, + "step": 79240 + }, + { + "epoch": 159.16, + "learning_rate": 2.0200167071097343e-05, + "loss": 2.2807, + "step": 79260 + }, + { + "epoch": 159.2, + "learning_rate": 2.016414111213273e-05, + "loss": 2.2926, + "step": 79280 + }, + { + "epoch": 159.24, + "learning_rate": 2.0128139191380723e-05, + "loss": 2.292, + "step": 79300 + }, + { + "epoch": 159.28, + "learning_rate": 2.0092161337847433e-05, + "loss": 2.3013, + "step": 79320 + }, + { + "epoch": 159.32, + "learning_rate": 2.0056207580519538e-05, + "loss": 2.301, + "step": 79340 + }, + { + "epoch": 159.36, + "learning_rate": 2.0020277948364325e-05, + "loss": 2.2993, + "step": 79360 + }, + { + "epoch": 159.4, + "learning_rate": 1.998437247032971e-05, + "loss": 2.2958, + "step": 79380 + }, + { + "epoch": 159.44, + "learning_rate": 1.9948491175344025e-05, + "loss": 2.2877, + "step": 79400 + }, + { + "epoch": 159.48, + "learning_rate": 1.9912634092316206e-05, + "loss": 2.302, + "step": 79420 + }, + { + "epoch": 159.52, + "learning_rate": 1.9876801250135662e-05, + "loss": 2.2864, + "step": 79440 + }, + { + "epoch": 159.56, + "learning_rate": 1.984099267767227e-05, + "loss": 2.2873, + "step": 79460 + }, + { + "epoch": 159.6, + "learning_rate": 1.9805208403776333e-05, + "loss": 2.3094, + "step": 79480 + }, + { + "epoch": 159.64, + "learning_rate": 1.976944845727859e-05, + "loss": 2.296, + "step": 79500 + }, + { + "epoch": 159.68, + "learning_rate": 1.97337128669902e-05, + "loss": 2.3085, + "step": 79520 + }, + { + "epoch": 159.72, + "learning_rate": 1.969800166170268e-05, + "loss": 2.2956, + "step": 79540 + }, + { + "epoch": 159.76, + "learning_rate": 1.966231487018792e-05, + "loss": 2.3012, + "step": 79560 + }, + { + "epoch": 159.8, + "learning_rate": 1.962665252119807e-05, + "loss": 2.313, + "step": 79580 + }, + { + "epoch": 159.84, + "learning_rate": 1.9591014643465706e-05, + "loss": 2.3164, + "step": 79600 + }, + { + "epoch": 159.88, + "learning_rate": 1.955540126570358e-05, + "loss": 2.3035, + "step": 79620 + }, + { + "epoch": 159.92, + "learning_rate": 1.9519812416604748e-05, + "loss": 2.3158, + "step": 79640 + }, + { + "epoch": 159.96, + "learning_rate": 1.9484248124842526e-05, + "loss": 2.3117, + "step": 79660 + }, + { + "epoch": 160.0, + "learning_rate": 1.94487084190704e-05, + "loss": 2.3112, + "step": 79680 + }, + { + "epoch": 160.04, + "learning_rate": 1.9413193327922104e-05, + "loss": 2.2748, + "step": 79700 + }, + { + "epoch": 160.08, + "learning_rate": 1.9377702880011434e-05, + "loss": 2.2529, + "step": 79720 + }, + { + "epoch": 160.12, + "learning_rate": 1.934223710393249e-05, + "loss": 2.2871, + "step": 79740 + }, + { + "epoch": 160.16, + "learning_rate": 1.930679602825935e-05, + "loss": 2.2822, + "step": 79760 + }, + { + "epoch": 160.2, + "learning_rate": 1.927137968154626e-05, + "loss": 2.2839, + "step": 79780 + }, + { + "epoch": 160.24, + "learning_rate": 1.923598809232754e-05, + "loss": 2.2889, + "step": 79800 + }, + { + "epoch": 160.28, + "learning_rate": 1.9200621289117544e-05, + "loss": 2.2873, + "step": 79820 + }, + { + "epoch": 160.32, + "learning_rate": 1.9165279300410694e-05, + "loss": 2.2926, + "step": 79840 + }, + { + "epoch": 160.36, + "learning_rate": 1.9129962154681346e-05, + "loss": 2.2951, + "step": 79860 + }, + { + "epoch": 160.4, + "learning_rate": 1.9094669880383915e-05, + "loss": 2.3022, + "step": 79880 + }, + { + "epoch": 160.44, + "learning_rate": 1.9059402505952744e-05, + "loss": 2.2923, + "step": 79900 + }, + { + "epoch": 160.48, + "learning_rate": 1.9024160059802122e-05, + "loss": 2.2916, + "step": 79920 + }, + { + "epoch": 160.52, + "learning_rate": 1.8988942570326245e-05, + "loss": 2.2724, + "step": 79940 + }, + { + "epoch": 160.56, + "learning_rate": 1.895550909728786e-05, + "loss": 2.2989, + "step": 79960 + }, + { + "epoch": 160.6, + "learning_rate": 1.8920340354920403e-05, + "loss": 2.3106, + "step": 79980 + }, + { + "epoch": 160.64, + "learning_rate": 1.888519665287337e-05, + "loss": 2.308, + "step": 80000 + }, + { + "epoch": 160.64, + "eval_loss": 2.825718641281128, + "eval_runtime": 32.4102, + "eval_samples_per_second": 23.635, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004437183916167702, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0001722512329561938, + "step": 80000 + }, + { + "epoch": 160.68, + "learning_rate": 1.885007801946136e-05, + "loss": 2.3005, + "step": 80020 + }, + { + "epoch": 160.72, + "learning_rate": 1.8814984482978893e-05, + "loss": 2.2994, + "step": 80040 + }, + { + "epoch": 160.76, + "learning_rate": 1.8779916071700136e-05, + "loss": 2.301, + "step": 80060 + }, + { + "epoch": 160.8, + "learning_rate": 1.87448728138791e-05, + "loss": 2.2982, + "step": 80080 + }, + { + "epoch": 160.84, + "learning_rate": 1.8709854737749498e-05, + "loss": 2.3096, + "step": 80100 + }, + { + "epoch": 160.88, + "learning_rate": 1.8674861871524772e-05, + "loss": 2.2997, + "step": 80120 + }, + { + "epoch": 160.92, + "learning_rate": 1.8639894243398053e-05, + "loss": 2.3257, + "step": 80140 + }, + { + "epoch": 160.96, + "learning_rate": 1.8604951881542094e-05, + "loss": 2.2945, + "step": 80160 + }, + { + "epoch": 161.0, + "learning_rate": 1.857003481410934e-05, + "loss": 2.2955, + "step": 80180 + }, + { + "epoch": 161.04, + "learning_rate": 1.8535143069231837e-05, + "loss": 2.2705, + "step": 80200 + }, + { + "epoch": 161.08, + "learning_rate": 1.850027667502126e-05, + "loss": 2.2778, + "step": 80220 + }, + { + "epoch": 161.12, + "learning_rate": 1.8465435659568784e-05, + "loss": 2.2774, + "step": 80240 + }, + { + "epoch": 161.16, + "learning_rate": 1.8430620050945248e-05, + "loss": 2.2842, + "step": 80260 + }, + { + "epoch": 161.2, + "learning_rate": 1.8395829877200903e-05, + "loss": 2.277, + "step": 80280 + }, + { + "epoch": 161.24, + "learning_rate": 1.836106516636559e-05, + "loss": 2.285, + "step": 80300 + }, + { + "epoch": 161.29, + "learning_rate": 1.8326325946448604e-05, + "loss": 2.2727, + "step": 80320 + }, + { + "epoch": 161.33, + "learning_rate": 1.8291612245438705e-05, + "loss": 2.2674, + "step": 80340 + }, + { + "epoch": 161.37, + "learning_rate": 1.8256924091304118e-05, + "loss": 2.283, + "step": 80360 + }, + { + "epoch": 161.41, + "learning_rate": 1.8222261511992406e-05, + "loss": 2.2977, + "step": 80380 + }, + { + "epoch": 161.45, + "learning_rate": 1.8187624535430652e-05, + "loss": 2.29, + "step": 80400 + }, + { + "epoch": 161.49, + "learning_rate": 1.8153013189525193e-05, + "loss": 2.2862, + "step": 80420 + }, + { + "epoch": 161.53, + "learning_rate": 1.811842750216178e-05, + "loss": 2.3057, + "step": 80440 + }, + { + "epoch": 161.57, + "learning_rate": 1.8083867501205483e-05, + "loss": 2.284, + "step": 80460 + }, + { + "epoch": 161.61, + "learning_rate": 1.8049333214500662e-05, + "loss": 2.2936, + "step": 80480 + }, + { + "epoch": 161.65, + "learning_rate": 1.801482466987099e-05, + "loss": 2.2977, + "step": 80500 + }, + { + "epoch": 161.69, + "learning_rate": 1.7980341895119347e-05, + "loss": 2.2943, + "step": 80520 + }, + { + "epoch": 161.73, + "learning_rate": 1.7945884918027888e-05, + "loss": 2.3067, + "step": 80540 + }, + { + "epoch": 161.77, + "learning_rate": 1.7911453766357983e-05, + "loss": 2.2957, + "step": 80560 + }, + { + "epoch": 161.81, + "learning_rate": 1.7877048467850195e-05, + "loss": 2.3029, + "step": 80580 + }, + { + "epoch": 161.85, + "learning_rate": 1.7842669050224242e-05, + "loss": 2.2949, + "step": 80600 + }, + { + "epoch": 161.89, + "learning_rate": 1.780831554117901e-05, + "loss": 2.2976, + "step": 80620 + }, + { + "epoch": 161.93, + "learning_rate": 1.7773987968392508e-05, + "loss": 2.301, + "step": 80640 + }, + { + "epoch": 161.97, + "learning_rate": 1.7739686359521814e-05, + "loss": 2.3058, + "step": 80660 + }, + { + "epoch": 162.01, + "learning_rate": 1.7705410742203122e-05, + "loss": 2.3083, + "step": 80680 + }, + { + "epoch": 162.05, + "learning_rate": 1.767116114405169e-05, + "loss": 2.2719, + "step": 80700 + }, + { + "epoch": 162.09, + "learning_rate": 1.7636937592661796e-05, + "loss": 2.2749, + "step": 80720 + }, + { + "epoch": 162.13, + "learning_rate": 1.7602740115606753e-05, + "loss": 2.2729, + "step": 80740 + }, + { + "epoch": 162.17, + "learning_rate": 1.7568568740438796e-05, + "loss": 2.2762, + "step": 80760 + }, + { + "epoch": 162.21, + "learning_rate": 1.7534423494689263e-05, + "loss": 2.2815, + "step": 80780 + }, + { + "epoch": 162.25, + "learning_rate": 1.7500304405868302e-05, + "loss": 2.2784, + "step": 80800 + }, + { + "epoch": 162.29, + "learning_rate": 1.7466211501465062e-05, + "loss": 2.2636, + "step": 80820 + }, + { + "epoch": 162.33, + "learning_rate": 1.7432144808947587e-05, + "loss": 2.2843, + "step": 80840 + }, + { + "epoch": 162.37, + "learning_rate": 1.7398104355762796e-05, + "loss": 2.2787, + "step": 80860 + }, + { + "epoch": 162.41, + "learning_rate": 1.7364090169336484e-05, + "loss": 2.2655, + "step": 80880 + }, + { + "epoch": 162.45, + "learning_rate": 1.7330102277073206e-05, + "loss": 2.2896, + "step": 80900 + }, + { + "epoch": 162.49, + "learning_rate": 1.7296140706356473e-05, + "loss": 2.2913, + "step": 80920 + }, + { + "epoch": 162.53, + "learning_rate": 1.726220548454845e-05, + "loss": 2.2873, + "step": 80940 + }, + { + "epoch": 162.57, + "learning_rate": 1.722829663899017e-05, + "loss": 2.2857, + "step": 80960 + }, + { + "epoch": 162.61, + "learning_rate": 1.7194414197001328e-05, + "loss": 2.2808, + "step": 80980 + }, + { + "epoch": 162.65, + "learning_rate": 1.716055818588045e-05, + "loss": 2.2838, + "step": 81000 + }, + { + "epoch": 162.65, + "eval_loss": 2.8250410556793213, + "eval_runtime": 32.6221, + "eval_samples_per_second": 23.481, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.004619568906029153, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0001450536698578474, + "step": 81000 + }, + { + "epoch": 162.69, + "learning_rate": 1.71267286329047e-05, + "loss": 2.3072, + "step": 81020 + }, + { + "epoch": 162.73, + "learning_rate": 1.709292556532991e-05, + "loss": 2.2908, + "step": 81040 + }, + { + "epoch": 162.77, + "learning_rate": 1.7059149010390622e-05, + "loss": 2.2944, + "step": 81060 + }, + { + "epoch": 162.81, + "learning_rate": 1.702539899529999e-05, + "loss": 2.3002, + "step": 81080 + }, + { + "epoch": 162.85, + "learning_rate": 1.6991675547249814e-05, + "loss": 2.3044, + "step": 81100 + }, + { + "epoch": 162.89, + "learning_rate": 1.6957978693410416e-05, + "loss": 2.3008, + "step": 81120 + }, + { + "epoch": 162.93, + "learning_rate": 1.6924308460930788e-05, + "loss": 2.3042, + "step": 81140 + }, + { + "epoch": 162.97, + "learning_rate": 1.6890664876938434e-05, + "loss": 2.3128, + "step": 81160 + }, + { + "epoch": 163.01, + "learning_rate": 1.6857047968539342e-05, + "loss": 2.2885, + "step": 81180 + }, + { + "epoch": 163.05, + "learning_rate": 1.6823457762818067e-05, + "loss": 2.2573, + "step": 81200 + }, + { + "epoch": 163.09, + "learning_rate": 1.6789894286837638e-05, + "loss": 2.2841, + "step": 81220 + }, + { + "epoch": 163.13, + "learning_rate": 1.6756357567639546e-05, + "loss": 2.2676, + "step": 81240 + }, + { + "epoch": 163.17, + "learning_rate": 1.67228476322437e-05, + "loss": 2.2764, + "step": 81260 + }, + { + "epoch": 163.21, + "learning_rate": 1.6689364507648436e-05, + "loss": 2.2809, + "step": 81280 + }, + { + "epoch": 163.25, + "learning_rate": 1.6655908220830562e-05, + "loss": 2.2675, + "step": 81300 + }, + { + "epoch": 163.29, + "learning_rate": 1.662247879874515e-05, + "loss": 2.2802, + "step": 81320 + }, + { + "epoch": 163.33, + "learning_rate": 1.6589076268325697e-05, + "loss": 2.2633, + "step": 81340 + }, + { + "epoch": 163.37, + "learning_rate": 1.655570065648403e-05, + "loss": 2.2807, + "step": 81360 + }, + { + "epoch": 163.41, + "learning_rate": 1.6522351990110275e-05, + "loss": 2.2871, + "step": 81380 + }, + { + "epoch": 163.45, + "learning_rate": 1.648903029607283e-05, + "loss": 2.2838, + "step": 81400 + }, + { + "epoch": 163.49, + "learning_rate": 1.6455735601218397e-05, + "loss": 2.2763, + "step": 81420 + }, + { + "epoch": 163.53, + "learning_rate": 1.6422467932371903e-05, + "loss": 2.2802, + "step": 81440 + }, + { + "epoch": 163.57, + "learning_rate": 1.6389227316336515e-05, + "loss": 2.2848, + "step": 81460 + }, + { + "epoch": 163.61, + "learning_rate": 1.635601377989361e-05, + "loss": 2.278, + "step": 81480 + }, + { + "epoch": 163.65, + "learning_rate": 1.632282734980269e-05, + "loss": 2.3008, + "step": 81500 + }, + { + "epoch": 163.69, + "learning_rate": 1.628966805280152e-05, + "loss": 2.2942, + "step": 81520 + }, + { + "epoch": 163.73, + "learning_rate": 1.6256535915605904e-05, + "loss": 2.2874, + "step": 81540 + }, + { + "epoch": 163.78, + "learning_rate": 1.622343096490982e-05, + "loss": 2.3068, + "step": 81560 + }, + { + "epoch": 163.82, + "learning_rate": 1.619035322738533e-05, + "loss": 2.3031, + "step": 81580 + }, + { + "epoch": 163.86, + "learning_rate": 1.6157302729682567e-05, + "loss": 2.2788, + "step": 81600 + }, + { + "epoch": 163.9, + "learning_rate": 1.6124279498429745e-05, + "loss": 2.2861, + "step": 81620 + }, + { + "epoch": 163.94, + "learning_rate": 1.6091283560233022e-05, + "loss": 2.288, + "step": 81640 + }, + { + "epoch": 163.98, + "learning_rate": 1.6058314941676706e-05, + "loss": 2.3088, + "step": 81660 + }, + { + "epoch": 164.02, + "learning_rate": 1.6025373669322962e-05, + "loss": 2.2706, + "step": 81680 + }, + { + "epoch": 164.06, + "learning_rate": 1.5992459769712003e-05, + "loss": 2.265, + "step": 81700 + }, + { + "epoch": 164.1, + "learning_rate": 1.5959573269361954e-05, + "loss": 2.2619, + "step": 81720 + }, + { + "epoch": 164.14, + "learning_rate": 1.5926714194768876e-05, + "loss": 2.2648, + "step": 81740 + }, + { + "epoch": 164.18, + "learning_rate": 1.5893882572406754e-05, + "loss": 2.2764, + "step": 81760 + }, + { + "epoch": 164.22, + "learning_rate": 1.5861078428727393e-05, + "loss": 2.2623, + "step": 81780 + }, + { + "epoch": 164.26, + "learning_rate": 1.582830179016052e-05, + "loss": 2.2686, + "step": 81800 + }, + { + "epoch": 164.3, + "learning_rate": 1.5795552683113678e-05, + "loss": 2.2754, + "step": 81820 + }, + { + "epoch": 164.34, + "learning_rate": 1.5762831133972256e-05, + "loss": 2.2692, + "step": 81840 + }, + { + "epoch": 164.38, + "learning_rate": 1.5730137169099352e-05, + "loss": 2.2791, + "step": 81860 + }, + { + "epoch": 164.42, + "learning_rate": 1.5697470814835963e-05, + "loss": 2.2698, + "step": 81880 + }, + { + "epoch": 164.46, + "learning_rate": 1.5664832097500777e-05, + "loss": 2.2938, + "step": 81900 + }, + { + "epoch": 164.5, + "learning_rate": 1.5632221043390184e-05, + "loss": 2.2791, + "step": 81920 + }, + { + "epoch": 164.54, + "learning_rate": 1.559963767877834e-05, + "loss": 2.2894, + "step": 81940 + }, + { + "epoch": 164.58, + "learning_rate": 1.5567082029917074e-05, + "loss": 2.2805, + "step": 81960 + }, + { + "epoch": 164.62, + "learning_rate": 1.55345541230359e-05, + "loss": 2.2817, + "step": 81980 + }, + { + "epoch": 164.66, + "learning_rate": 1.550205398434193e-05, + "loss": 2.2841, + "step": 82000 + }, + { + "epoch": 164.66, + "eval_loss": 2.8190503120422363, + "eval_runtime": 32.2621, + "eval_samples_per_second": 23.743, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.00591169446020621, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0007389444428819274, + "step": 82000 + }, + { + "epoch": 164.7, + "learning_rate": 1.546958164001994e-05, + "loss": 2.2863, + "step": 82020 + }, + { + "epoch": 164.74, + "learning_rate": 1.543713711623237e-05, + "loss": 2.2954, + "step": 82040 + }, + { + "epoch": 164.78, + "learning_rate": 1.540472043911912e-05, + "loss": 2.2849, + "step": 82060 + }, + { + "epoch": 164.82, + "learning_rate": 1.537233163479775e-05, + "loss": 2.289, + "step": 82080 + }, + { + "epoch": 164.86, + "learning_rate": 1.533997072936333e-05, + "loss": 2.294, + "step": 82100 + }, + { + "epoch": 164.9, + "learning_rate": 1.530763774888847e-05, + "loss": 2.2923, + "step": 82120 + }, + { + "epoch": 164.94, + "learning_rate": 1.5275332719423246e-05, + "loss": 2.2945, + "step": 82140 + }, + { + "epoch": 164.98, + "learning_rate": 1.5243055666995232e-05, + "loss": 2.2875, + "step": 82160 + }, + { + "epoch": 165.02, + "learning_rate": 1.5210806617609508e-05, + "loss": 2.2732, + "step": 82180 + }, + { + "epoch": 165.06, + "learning_rate": 1.5178585597248513e-05, + "loss": 2.2628, + "step": 82200 + }, + { + "epoch": 165.1, + "learning_rate": 1.5146392631872148e-05, + "loss": 2.2485, + "step": 82220 + }, + { + "epoch": 165.14, + "learning_rate": 1.511422774741771e-05, + "loss": 2.2609, + "step": 82240 + }, + { + "epoch": 165.18, + "learning_rate": 1.5082090969799878e-05, + "loss": 2.2668, + "step": 82260 + }, + { + "epoch": 165.22, + "learning_rate": 1.5049982324910645e-05, + "loss": 2.2796, + "step": 82280 + }, + { + "epoch": 165.26, + "learning_rate": 1.5017901838619374e-05, + "loss": 2.2764, + "step": 82300 + }, + { + "epoch": 165.3, + "learning_rate": 1.4985849536772745e-05, + "loss": 2.2651, + "step": 82320 + }, + { + "epoch": 165.34, + "learning_rate": 1.4953825445194714e-05, + "loss": 2.2697, + "step": 82340 + }, + { + "epoch": 165.38, + "learning_rate": 1.492182958968652e-05, + "loss": 2.2805, + "step": 82360 + }, + { + "epoch": 165.42, + "learning_rate": 1.4889861996026616e-05, + "loss": 2.2744, + "step": 82380 + }, + { + "epoch": 165.46, + "learning_rate": 1.4857922689970761e-05, + "loss": 2.2736, + "step": 82400 + }, + { + "epoch": 165.5, + "learning_rate": 1.482601169725184e-05, + "loss": 2.2791, + "step": 82420 + }, + { + "epoch": 165.54, + "learning_rate": 1.4794129043579974e-05, + "loss": 2.2841, + "step": 82440 + }, + { + "epoch": 165.58, + "learning_rate": 1.476227475464244e-05, + "loss": 2.2777, + "step": 82460 + }, + { + "epoch": 165.62, + "learning_rate": 1.4730448856103674e-05, + "loss": 2.2741, + "step": 82480 + }, + { + "epoch": 165.66, + "learning_rate": 1.4698651373605233e-05, + "loss": 2.2862, + "step": 82500 + }, + { + "epoch": 165.7, + "learning_rate": 1.4666882332765747e-05, + "loss": 2.2726, + "step": 82520 + }, + { + "epoch": 165.74, + "learning_rate": 1.463514175918097e-05, + "loss": 2.285, + "step": 82540 + }, + { + "epoch": 165.78, + "learning_rate": 1.4603429678423713e-05, + "loss": 2.2744, + "step": 82560 + }, + { + "epoch": 165.82, + "learning_rate": 1.4571746116043821e-05, + "loss": 2.2768, + "step": 82580 + }, + { + "epoch": 165.86, + "learning_rate": 1.4540091097568175e-05, + "loss": 2.2948, + "step": 82600 + }, + { + "epoch": 165.9, + "learning_rate": 1.4508464648500652e-05, + "loss": 2.2872, + "step": 82620 + }, + { + "epoch": 165.94, + "learning_rate": 1.447686679432212e-05, + "loss": 2.2937, + "step": 82640 + }, + { + "epoch": 165.98, + "learning_rate": 1.4445297560490374e-05, + "loss": 2.2891, + "step": 82660 + }, + { + "epoch": 166.02, + "learning_rate": 1.441375697244019e-05, + "loss": 2.2762, + "step": 82680 + }, + { + "epoch": 166.06, + "learning_rate": 1.4382245055583243e-05, + "loss": 2.2712, + "step": 82700 + }, + { + "epoch": 166.1, + "learning_rate": 1.4350761835308141e-05, + "loss": 2.2625, + "step": 82720 + }, + { + "epoch": 166.14, + "learning_rate": 1.4319307336980292e-05, + "loss": 2.2668, + "step": 82740 + }, + { + "epoch": 166.18, + "learning_rate": 1.4287881585942064e-05, + "loss": 2.2559, + "step": 82760 + }, + { + "epoch": 166.22, + "learning_rate": 1.4256484607512627e-05, + "loss": 2.2649, + "step": 82780 + }, + { + "epoch": 166.27, + "learning_rate": 1.4225116426987917e-05, + "loss": 2.2764, + "step": 82800 + }, + { + "epoch": 166.31, + "learning_rate": 1.419377706964073e-05, + "loss": 2.2608, + "step": 82820 + }, + { + "epoch": 166.35, + "learning_rate": 1.4162466560720616e-05, + "loss": 2.2763, + "step": 82840 + }, + { + "epoch": 166.39, + "learning_rate": 1.4131184925453911e-05, + "loss": 2.2669, + "step": 82860 + }, + { + "epoch": 166.43, + "learning_rate": 1.4099932189043625e-05, + "loss": 2.2614, + "step": 82880 + }, + { + "epoch": 166.47, + "learning_rate": 1.4068708376669526e-05, + "loss": 2.2803, + "step": 82900 + }, + { + "epoch": 166.51, + "learning_rate": 1.4037513513488121e-05, + "loss": 2.2657, + "step": 82920 + }, + { + "epoch": 166.55, + "learning_rate": 1.4006347624632504e-05, + "loss": 2.2847, + "step": 82940 + }, + { + "epoch": 166.59, + "learning_rate": 1.3975210735212484e-05, + "loss": 2.2612, + "step": 82960 + }, + { + "epoch": 166.63, + "learning_rate": 1.3944102870314484e-05, + "loss": 2.2836, + "step": 82980 + }, + { + "epoch": 166.67, + "learning_rate": 1.3913024055001572e-05, + "loss": 2.2897, + "step": 83000 + }, + { + "epoch": 166.67, + "eval_loss": 2.818046808242798, + "eval_runtime": 32.432, + "eval_samples_per_second": 23.619, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.00399408220741636, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00038076588337684946, + "step": 83000 + }, + { + "epoch": 166.71, + "learning_rate": 1.3881974314313356e-05, + "loss": 2.275, + "step": 83020 + }, + { + "epoch": 166.75, + "learning_rate": 1.3850953673266066e-05, + "loss": 2.2805, + "step": 83040 + }, + { + "epoch": 166.79, + "learning_rate": 1.3819962156852468e-05, + "loss": 2.2856, + "step": 83060 + }, + { + "epoch": 166.83, + "learning_rate": 1.3788999790041867e-05, + "loss": 2.2831, + "step": 83080 + }, + { + "epoch": 166.87, + "learning_rate": 1.3758066597780106e-05, + "loss": 2.2762, + "step": 83100 + }, + { + "epoch": 166.91, + "learning_rate": 1.3727162604989451e-05, + "loss": 2.2804, + "step": 83120 + }, + { + "epoch": 166.95, + "learning_rate": 1.3696287836568744e-05, + "loss": 2.2732, + "step": 83140 + }, + { + "epoch": 166.99, + "learning_rate": 1.3665442317393196e-05, + "loss": 2.2792, + "step": 83160 + }, + { + "epoch": 167.03, + "learning_rate": 1.3634626072314482e-05, + "loss": 2.2515, + "step": 83180 + }, + { + "epoch": 167.07, + "learning_rate": 1.3603839126160711e-05, + "loss": 2.2478, + "step": 83200 + }, + { + "epoch": 167.11, + "learning_rate": 1.3573081503736362e-05, + "loss": 2.2573, + "step": 83220 + }, + { + "epoch": 167.15, + "learning_rate": 1.3542353229822308e-05, + "loss": 2.2574, + "step": 83240 + }, + { + "epoch": 167.19, + "learning_rate": 1.3511654329175727e-05, + "loss": 2.2591, + "step": 83260 + }, + { + "epoch": 167.23, + "learning_rate": 1.3480984826530218e-05, + "loss": 2.2443, + "step": 83280 + }, + { + "epoch": 167.27, + "learning_rate": 1.3450344746595612e-05, + "loss": 2.2553, + "step": 83300 + }, + { + "epoch": 167.31, + "learning_rate": 1.3419734114058075e-05, + "loss": 2.262, + "step": 83320 + }, + { + "epoch": 167.35, + "learning_rate": 1.3389152953580047e-05, + "loss": 2.267, + "step": 83340 + }, + { + "epoch": 167.39, + "learning_rate": 1.335860128980021e-05, + "loss": 2.2873, + "step": 83360 + }, + { + "epoch": 167.43, + "learning_rate": 1.3328079147333505e-05, + "loss": 2.2754, + "step": 83380 + }, + { + "epoch": 167.47, + "learning_rate": 1.329758655077104e-05, + "loss": 2.2765, + "step": 83400 + }, + { + "epoch": 167.51, + "learning_rate": 1.326712352468017e-05, + "loss": 2.2693, + "step": 83420 + }, + { + "epoch": 167.55, + "learning_rate": 1.3236690093604399e-05, + "loss": 2.264, + "step": 83440 + }, + { + "epoch": 167.59, + "learning_rate": 1.3206286282063395e-05, + "loss": 2.2648, + "step": 83460 + }, + { + "epoch": 167.63, + "learning_rate": 1.3175912114552963e-05, + "loss": 2.2713, + "step": 83480 + }, + { + "epoch": 167.67, + "learning_rate": 1.3145567615545012e-05, + "loss": 2.2777, + "step": 83500 + }, + { + "epoch": 167.71, + "learning_rate": 1.3115252809487577e-05, + "loss": 2.289, + "step": 83520 + }, + { + "epoch": 167.75, + "learning_rate": 1.3084967720804725e-05, + "loss": 2.2723, + "step": 83540 + }, + { + "epoch": 167.79, + "learning_rate": 1.3054712373896611e-05, + "loss": 2.2811, + "step": 83560 + }, + { + "epoch": 167.83, + "learning_rate": 1.3024486793139423e-05, + "loss": 2.2713, + "step": 83580 + }, + { + "epoch": 167.87, + "learning_rate": 1.2994291002885373e-05, + "loss": 2.2914, + "step": 83600 + }, + { + "epoch": 167.91, + "learning_rate": 1.296412502746267e-05, + "loss": 2.2918, + "step": 83620 + }, + { + "epoch": 167.95, + "learning_rate": 1.2933988891175458e-05, + "loss": 2.2827, + "step": 83640 + }, + { + "epoch": 167.99, + "learning_rate": 1.290388261830393e-05, + "loss": 2.2826, + "step": 83660 + }, + { + "epoch": 168.03, + "learning_rate": 1.287380623310413e-05, + "loss": 2.2567, + "step": 83680 + }, + { + "epoch": 168.07, + "learning_rate": 1.2843759759808061e-05, + "loss": 2.2499, + "step": 83700 + }, + { + "epoch": 168.11, + "learning_rate": 1.2813743222623637e-05, + "loss": 2.2547, + "step": 83720 + }, + { + "epoch": 168.15, + "learning_rate": 1.2783756645734635e-05, + "loss": 2.2607, + "step": 83740 + }, + { + "epoch": 168.19, + "learning_rate": 1.2753800053300707e-05, + "loss": 2.2501, + "step": 83760 + }, + { + "epoch": 168.23, + "learning_rate": 1.2723873469457304e-05, + "loss": 2.2518, + "step": 83780 + }, + { + "epoch": 168.27, + "learning_rate": 1.2693976918315786e-05, + "loss": 2.2674, + "step": 83800 + }, + { + "epoch": 168.31, + "learning_rate": 1.2664110423963221e-05, + "loss": 2.2578, + "step": 83820 + }, + { + "epoch": 168.35, + "learning_rate": 1.2634274010462538e-05, + "loss": 2.2761, + "step": 83840 + }, + { + "epoch": 168.39, + "learning_rate": 1.2604467701852351e-05, + "loss": 2.2632, + "step": 83860 + }, + { + "epoch": 168.43, + "learning_rate": 1.2574691522147103e-05, + "loss": 2.2737, + "step": 83880 + }, + { + "epoch": 168.47, + "learning_rate": 1.254494549533693e-05, + "loss": 2.2708, + "step": 83900 + }, + { + "epoch": 168.51, + "learning_rate": 1.2515229645387638e-05, + "loss": 2.2634, + "step": 83920 + }, + { + "epoch": 168.55, + "learning_rate": 1.2485543996240762e-05, + "loss": 2.2691, + "step": 83940 + }, + { + "epoch": 168.59, + "learning_rate": 1.2457370624828829e-05, + "loss": 2.2728, + "step": 83960 + }, + { + "epoch": 168.63, + "learning_rate": 1.2427743936016312e-05, + "loss": 2.264, + "step": 83980 + }, + { + "epoch": 168.67, + "learning_rate": 1.239814751849187e-05, + "loss": 2.2736, + "step": 84000 + }, + { + "epoch": 168.67, + "eval_loss": 2.8139073848724365, + "eval_runtime": 32.6962, + "eval_samples_per_second": 23.428, + "eval_steps_per_second": 0.122, + "eval_tse_ndup": 0.005459740131815287, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00022672367119640373, + "step": 84000 + }, + { + "epoch": 168.71, + "learning_rate": 1.2368581396100804e-05, + "loss": 2.2798, + "step": 84020 + }, + { + "epoch": 168.76, + "learning_rate": 1.2339045592664011e-05, + "loss": 2.2643, + "step": 84040 + }, + { + "epoch": 168.8, + "learning_rate": 1.2309540131977937e-05, + "loss": 2.2784, + "step": 84060 + }, + { + "epoch": 168.84, + "learning_rate": 1.2280065037814604e-05, + "loss": 2.2662, + "step": 84080 + }, + { + "epoch": 168.88, + "learning_rate": 1.2250620333921564e-05, + "loss": 2.2775, + "step": 84100 + }, + { + "epoch": 168.92, + "learning_rate": 1.2221206044021888e-05, + "loss": 2.2729, + "step": 84120 + }, + { + "epoch": 168.96, + "learning_rate": 1.2191822191814134e-05, + "loss": 2.2845, + "step": 84140 + }, + { + "epoch": 169.0, + "learning_rate": 1.2162468800972342e-05, + "loss": 2.2824, + "step": 84160 + }, + { + "epoch": 169.04, + "learning_rate": 1.213314589514603e-05, + "loss": 2.2596, + "step": 84180 + }, + { + "epoch": 169.08, + "learning_rate": 1.2103853497960095e-05, + "loss": 2.2431, + "step": 84200 + }, + { + "epoch": 169.12, + "learning_rate": 1.2074591633014903e-05, + "loss": 2.2273, + "step": 84220 + }, + { + "epoch": 169.16, + "learning_rate": 1.2045360323886223e-05, + "loss": 2.2499, + "step": 84240 + }, + { + "epoch": 169.2, + "learning_rate": 1.2016159594125187e-05, + "loss": 2.2723, + "step": 84260 + }, + { + "epoch": 169.24, + "learning_rate": 1.1986989467258314e-05, + "loss": 2.2672, + "step": 84280 + }, + { + "epoch": 169.28, + "learning_rate": 1.1957849966787394e-05, + "loss": 2.2671, + "step": 84300 + }, + { + "epoch": 169.32, + "learning_rate": 1.1928741116189663e-05, + "loss": 2.2723, + "step": 84320 + }, + { + "epoch": 169.36, + "learning_rate": 1.1899662938917555e-05, + "loss": 2.2669, + "step": 84340 + }, + { + "epoch": 169.4, + "learning_rate": 1.1870615458398838e-05, + "loss": 2.263, + "step": 84360 + }, + { + "epoch": 169.44, + "learning_rate": 1.1841598698036548e-05, + "loss": 2.2667, + "step": 84380 + }, + { + "epoch": 169.48, + "learning_rate": 1.1812612681208962e-05, + "loss": 2.2571, + "step": 84400 + }, + { + "epoch": 169.52, + "learning_rate": 1.1783657431269606e-05, + "loss": 2.2705, + "step": 84420 + }, + { + "epoch": 169.56, + "learning_rate": 1.1754732971547177e-05, + "loss": 2.2728, + "step": 84440 + }, + { + "epoch": 169.6, + "learning_rate": 1.17258393253456e-05, + "loss": 2.259, + "step": 84460 + }, + { + "epoch": 169.64, + "learning_rate": 1.1696976515943958e-05, + "loss": 2.2672, + "step": 84480 + }, + { + "epoch": 169.68, + "learning_rate": 1.1668144566596523e-05, + "loss": 2.2611, + "step": 84500 + }, + { + "epoch": 169.72, + "learning_rate": 1.1639343500532624e-05, + "loss": 2.2716, + "step": 84520 + }, + { + "epoch": 169.76, + "learning_rate": 1.1610573340956799e-05, + "loss": 2.2719, + "step": 84540 + }, + { + "epoch": 169.8, + "learning_rate": 1.1581834111048657e-05, + "loss": 2.2682, + "step": 84560 + }, + { + "epoch": 169.84, + "learning_rate": 1.155312583396284e-05, + "loss": 2.2702, + "step": 84580 + }, + { + "epoch": 169.88, + "learning_rate": 1.1524448532829107e-05, + "loss": 2.2591, + "step": 84600 + }, + { + "epoch": 169.92, + "learning_rate": 1.1495802230752234e-05, + "loss": 2.2802, + "step": 84620 + }, + { + "epoch": 169.96, + "learning_rate": 1.1467186950812053e-05, + "loss": 2.2686, + "step": 84640 + }, + { + "epoch": 170.0, + "learning_rate": 1.1438602716063329e-05, + "loss": 2.2733, + "step": 84660 + }, + { + "epoch": 170.04, + "learning_rate": 1.1410049549535911e-05, + "loss": 2.2503, + "step": 84680 + }, + { + "epoch": 170.08, + "learning_rate": 1.1381527474234565e-05, + "loss": 2.2647, + "step": 84700 + }, + { + "epoch": 170.12, + "learning_rate": 1.1353036513138986e-05, + "loss": 2.246, + "step": 84720 + }, + { + "epoch": 170.16, + "learning_rate": 1.1324576689203842e-05, + "loss": 2.2571, + "step": 84740 + }, + { + "epoch": 170.2, + "learning_rate": 1.1296148025358694e-05, + "loss": 2.2449, + "step": 84760 + }, + { + "epoch": 170.24, + "learning_rate": 1.126775054450802e-05, + "loss": 2.2551, + "step": 84780 + }, + { + "epoch": 170.28, + "learning_rate": 1.1239384269531133e-05, + "loss": 2.2524, + "step": 84800 + }, + { + "epoch": 170.32, + "learning_rate": 1.1211049223282227e-05, + "loss": 2.2551, + "step": 84820 + }, + { + "epoch": 170.36, + "learning_rate": 1.1182745428590374e-05, + "loss": 2.2528, + "step": 84840 + }, + { + "epoch": 170.4, + "learning_rate": 1.1154472908259395e-05, + "loss": 2.2538, + "step": 84860 + }, + { + "epoch": 170.44, + "learning_rate": 1.1126231685067956e-05, + "loss": 2.2631, + "step": 84880 + }, + { + "epoch": 170.48, + "learning_rate": 1.1098021781769503e-05, + "loss": 2.2516, + "step": 84900 + }, + { + "epoch": 170.52, + "learning_rate": 1.1069843221092263e-05, + "loss": 2.2637, + "step": 84920 + }, + { + "epoch": 170.56, + "learning_rate": 1.1041696025739162e-05, + "loss": 2.2477, + "step": 84940 + }, + { + "epoch": 170.6, + "learning_rate": 1.1013580218387897e-05, + "loss": 2.2765, + "step": 84960 + }, + { + "epoch": 170.64, + "learning_rate": 1.0985495821690878e-05, + "loss": 2.2606, + "step": 84980 + }, + { + "epoch": 170.68, + "learning_rate": 1.0957442858275185e-05, + "loss": 2.2623, + "step": 85000 + }, + { + "epoch": 170.68, + "eval_loss": 2.8101019859313965, + "eval_runtime": 32.3376, + "eval_samples_per_second": 23.688, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.005201261590475699, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0002198469683783, + "step": 85000 + }, + { + "epoch": 170.72, + "learning_rate": 1.0929421350742603e-05, + "loss": 2.268, + "step": 85020 + }, + { + "epoch": 170.76, + "learning_rate": 1.0901431321669508e-05, + "loss": 2.2613, + "step": 85040 + }, + { + "epoch": 170.8, + "learning_rate": 1.0873472793607025e-05, + "loss": 2.2647, + "step": 85060 + }, + { + "epoch": 170.84, + "learning_rate": 1.0845545789080785e-05, + "loss": 2.2756, + "step": 85080 + }, + { + "epoch": 170.88, + "learning_rate": 1.0817650330591094e-05, + "loss": 2.2657, + "step": 85100 + }, + { + "epoch": 170.92, + "learning_rate": 1.0789786440612808e-05, + "loss": 2.2673, + "step": 85120 + }, + { + "epoch": 170.96, + "learning_rate": 1.0761954141595364e-05, + "loss": 2.2848, + "step": 85140 + }, + { + "epoch": 171.0, + "learning_rate": 1.0734153455962765e-05, + "loss": 2.2779, + "step": 85160 + }, + { + "epoch": 171.04, + "learning_rate": 1.0706384406113468e-05, + "loss": 2.2496, + "step": 85180 + }, + { + "epoch": 171.08, + "learning_rate": 1.0678647014420556e-05, + "loss": 2.2399, + "step": 85200 + }, + { + "epoch": 171.12, + "learning_rate": 1.0650941303231499e-05, + "loss": 2.2489, + "step": 85220 + }, + { + "epoch": 171.16, + "learning_rate": 1.0623267294868322e-05, + "loss": 2.2537, + "step": 85240 + }, + { + "epoch": 171.2, + "learning_rate": 1.0595625011627425e-05, + "loss": 2.2408, + "step": 85260 + }, + { + "epoch": 171.24, + "learning_rate": 1.0568014475779742e-05, + "loss": 2.2472, + "step": 85280 + }, + { + "epoch": 171.29, + "learning_rate": 1.0540435709570585e-05, + "loss": 2.266, + "step": 85300 + }, + { + "epoch": 171.33, + "learning_rate": 1.0512888735219639e-05, + "loss": 2.2474, + "step": 85320 + }, + { + "epoch": 171.37, + "learning_rate": 1.048537357492101e-05, + "loss": 2.2616, + "step": 85340 + }, + { + "epoch": 171.41, + "learning_rate": 1.0457890250843178e-05, + "loss": 2.2563, + "step": 85360 + }, + { + "epoch": 171.45, + "learning_rate": 1.0430438785128966e-05, + "loss": 2.267, + "step": 85380 + }, + { + "epoch": 171.49, + "learning_rate": 1.0403019199895487e-05, + "loss": 2.265, + "step": 85400 + }, + { + "epoch": 171.53, + "learning_rate": 1.0375631517234241e-05, + "loss": 2.2688, + "step": 85420 + }, + { + "epoch": 171.57, + "learning_rate": 1.0348275759210996e-05, + "loss": 2.2572, + "step": 85440 + }, + { + "epoch": 171.61, + "learning_rate": 1.0320951947865764e-05, + "loss": 2.2621, + "step": 85460 + }, + { + "epoch": 171.65, + "learning_rate": 1.0293660105212854e-05, + "loss": 2.2631, + "step": 85480 + }, + { + "epoch": 171.69, + "learning_rate": 1.0266400253240815e-05, + "loss": 2.2495, + "step": 85500 + }, + { + "epoch": 171.73, + "learning_rate": 1.0239172413912435e-05, + "loss": 2.2642, + "step": 85520 + }, + { + "epoch": 171.77, + "learning_rate": 1.0211976609164658e-05, + "loss": 2.2624, + "step": 85540 + }, + { + "epoch": 171.81, + "learning_rate": 1.0184812860908655e-05, + "loss": 2.2508, + "step": 85560 + }, + { + "epoch": 171.85, + "learning_rate": 1.0157681191029805e-05, + "loss": 2.2732, + "step": 85580 + }, + { + "epoch": 171.89, + "learning_rate": 1.0130581621387575e-05, + "loss": 2.2727, + "step": 85600 + }, + { + "epoch": 171.93, + "learning_rate": 1.0103514173815598e-05, + "loss": 2.261, + "step": 85620 + }, + { + "epoch": 171.97, + "learning_rate": 1.0076478870121636e-05, + "loss": 2.2632, + "step": 85640 + }, + { + "epoch": 172.01, + "learning_rate": 1.004947573208756e-05, + "loss": 2.2604, + "step": 85660 + }, + { + "epoch": 172.05, + "learning_rate": 1.0022504781469277e-05, + "loss": 2.2442, + "step": 85680 + }, + { + "epoch": 172.09, + "learning_rate": 9.995566039996796e-06, + "loss": 2.2461, + "step": 85700 + }, + { + "epoch": 172.13, + "learning_rate": 9.968659529374219e-06, + "loss": 2.2364, + "step": 85720 + }, + { + "epoch": 172.17, + "learning_rate": 9.941785271279592e-06, + "loss": 2.2482, + "step": 85740 + }, + { + "epoch": 172.21, + "learning_rate": 9.914943287365046e-06, + "loss": 2.2615, + "step": 85760 + }, + { + "epoch": 172.25, + "learning_rate": 9.88813359925665e-06, + "loss": 2.2413, + "step": 85780 + }, + { + "epoch": 172.29, + "learning_rate": 9.861356228554524e-06, + "loss": 2.2602, + "step": 85800 + }, + { + "epoch": 172.33, + "learning_rate": 9.834611196832689e-06, + "loss": 2.241, + "step": 85820 + }, + { + "epoch": 172.37, + "learning_rate": 9.80789852563913e-06, + "loss": 2.2506, + "step": 85840 + }, + { + "epoch": 172.41, + "learning_rate": 9.781218236495776e-06, + "loss": 2.2583, + "step": 85860 + }, + { + "epoch": 172.45, + "learning_rate": 9.754570350898451e-06, + "loss": 2.2628, + "step": 85880 + }, + { + "epoch": 172.49, + "learning_rate": 9.727954890316886e-06, + "loss": 2.2649, + "step": 85900 + }, + { + "epoch": 172.53, + "learning_rate": 9.701371876194642e-06, + "loss": 2.2499, + "step": 85920 + }, + { + "epoch": 172.57, + "learning_rate": 9.674821329949225e-06, + "loss": 2.2528, + "step": 85940 + }, + { + "epoch": 172.61, + "learning_rate": 9.649628403870708e-06, + "loss": 2.2797, + "step": 85960 + }, + { + "epoch": 172.65, + "learning_rate": 9.623141231487904e-06, + "loss": 2.2686, + "step": 85980 + }, + { + "epoch": 172.69, + "learning_rate": 9.596686590010917e-06, + "loss": 2.2443, + "step": 86000 + }, + { + "epoch": 172.69, + "eval_loss": 2.8058507442474365, + "eval_runtime": 32.501, + "eval_samples_per_second": 23.568, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.00505093629186253, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0006958043225993617, + "step": 86000 + }, + { + "epoch": 172.73, + "learning_rate": 9.570264500753773e-06, + "loss": 2.2681, + "step": 86020 + }, + { + "epoch": 172.77, + "learning_rate": 9.543874985004275e-06, + "loss": 2.2504, + "step": 86040 + }, + { + "epoch": 172.81, + "learning_rate": 9.517518064023972e-06, + "loss": 2.2585, + "step": 86060 + }, + { + "epoch": 172.85, + "learning_rate": 9.491193759048166e-06, + "loss": 2.2642, + "step": 86080 + }, + { + "epoch": 172.89, + "learning_rate": 9.464902091285883e-06, + "loss": 2.2573, + "step": 86100 + }, + { + "epoch": 172.93, + "learning_rate": 9.438643081919818e-06, + "loss": 2.2444, + "step": 86120 + }, + { + "epoch": 172.97, + "learning_rate": 9.4124167521064e-06, + "loss": 2.2633, + "step": 86140 + }, + { + "epoch": 173.01, + "learning_rate": 9.386223122975702e-06, + "loss": 2.2685, + "step": 86160 + }, + { + "epoch": 173.05, + "learning_rate": 9.360062215631476e-06, + "loss": 2.2358, + "step": 86180 + }, + { + "epoch": 173.09, + "learning_rate": 9.333934051151066e-06, + "loss": 2.2469, + "step": 86200 + }, + { + "epoch": 173.13, + "learning_rate": 9.30783865058546e-06, + "loss": 2.2481, + "step": 86220 + }, + { + "epoch": 173.17, + "learning_rate": 9.281776034959288e-06, + "loss": 2.24, + "step": 86240 + }, + { + "epoch": 173.21, + "learning_rate": 9.25574622527069e-06, + "loss": 2.2562, + "step": 86260 + }, + { + "epoch": 173.25, + "learning_rate": 9.229749242491436e-06, + "loss": 2.2437, + "step": 86280 + }, + { + "epoch": 173.29, + "learning_rate": 9.203785107566815e-06, + "loss": 2.2503, + "step": 86300 + }, + { + "epoch": 173.33, + "learning_rate": 9.177853841415674e-06, + "loss": 2.25, + "step": 86320 + }, + { + "epoch": 173.37, + "learning_rate": 9.151955464930345e-06, + "loss": 2.2467, + "step": 86340 + }, + { + "epoch": 173.41, + "learning_rate": 9.126089998976684e-06, + "loss": 2.2536, + "step": 86360 + }, + { + "epoch": 173.45, + "learning_rate": 9.100257464394062e-06, + "loss": 2.27, + "step": 86380 + }, + { + "epoch": 173.49, + "learning_rate": 9.074457881995252e-06, + "loss": 2.2459, + "step": 86400 + }, + { + "epoch": 173.53, + "learning_rate": 9.04869127256654e-06, + "loss": 2.2519, + "step": 86420 + }, + { + "epoch": 173.57, + "learning_rate": 9.022957656867565e-06, + "loss": 2.2508, + "step": 86440 + }, + { + "epoch": 173.61, + "learning_rate": 8.997257055631503e-06, + "loss": 2.27, + "step": 86460 + }, + { + "epoch": 173.65, + "learning_rate": 8.971589489564819e-06, + "loss": 2.2432, + "step": 86480 + }, + { + "epoch": 173.69, + "learning_rate": 8.945954979347421e-06, + "loss": 2.2539, + "step": 86500 + }, + { + "epoch": 173.73, + "learning_rate": 8.92035354563257e-06, + "loss": 2.2704, + "step": 86520 + }, + { + "epoch": 173.78, + "learning_rate": 8.894785209046885e-06, + "loss": 2.2624, + "step": 86540 + }, + { + "epoch": 173.82, + "learning_rate": 8.87052596426936e-06, + "loss": 2.2587, + "step": 86560 + }, + { + "epoch": 173.86, + "learning_rate": 8.845022226311828e-06, + "loss": 2.2595, + "step": 86580 + }, + { + "epoch": 173.9, + "learning_rate": 8.819551646176543e-06, + "loss": 2.2476, + "step": 86600 + }, + { + "epoch": 173.94, + "learning_rate": 8.794114244384698e-06, + "loss": 2.27, + "step": 86620 + }, + { + "epoch": 173.98, + "learning_rate": 8.768710041430728e-06, + "loss": 2.2599, + "step": 86640 + }, + { + "epoch": 174.02, + "learning_rate": 8.743339057782351e-06, + "loss": 2.2398, + "step": 86660 + }, + { + "epoch": 174.06, + "learning_rate": 8.718001313880515e-06, + "loss": 2.2524, + "step": 86680 + }, + { + "epoch": 174.1, + "learning_rate": 8.692696830139368e-06, + "loss": 2.2294, + "step": 86700 + }, + { + "epoch": 174.14, + "learning_rate": 8.667425626946296e-06, + "loss": 2.2332, + "step": 86720 + }, + { + "epoch": 174.18, + "learning_rate": 8.642187724661833e-06, + "loss": 2.2629, + "step": 86740 + }, + { + "epoch": 174.22, + "learning_rate": 8.616983143619723e-06, + "loss": 2.2403, + "step": 86760 + }, + { + "epoch": 174.26, + "learning_rate": 8.591811904126818e-06, + "loss": 2.2509, + "step": 86780 + }, + { + "epoch": 174.3, + "learning_rate": 8.566674026463135e-06, + "loss": 2.2529, + "step": 86800 + }, + { + "epoch": 174.34, + "learning_rate": 8.541569530881816e-06, + "loss": 2.2456, + "step": 86820 + }, + { + "epoch": 174.38, + "learning_rate": 8.516498437609094e-06, + "loss": 2.2591, + "step": 86840 + }, + { + "epoch": 174.42, + "learning_rate": 8.491460766844317e-06, + "loss": 2.2574, + "step": 86860 + }, + { + "epoch": 174.46, + "learning_rate": 8.466456538759837e-06, + "loss": 2.2597, + "step": 86880 + }, + { + "epoch": 174.5, + "learning_rate": 8.441485773501163e-06, + "loss": 2.2531, + "step": 86900 + }, + { + "epoch": 174.54, + "learning_rate": 8.416548491186753e-06, + "loss": 2.2468, + "step": 86920 + }, + { + "epoch": 174.58, + "learning_rate": 8.391644711908136e-06, + "loss": 2.2596, + "step": 86940 + }, + { + "epoch": 174.62, + "learning_rate": 8.366774455729836e-06, + "loss": 2.2466, + "step": 86960 + }, + { + "epoch": 174.66, + "learning_rate": 8.341937742689371e-06, + "loss": 2.2572, + "step": 86980 + }, + { + "epoch": 174.7, + "learning_rate": 8.317134592797248e-06, + "loss": 2.2529, + "step": 87000 + }, + { + "epoch": 174.7, + "eval_loss": 2.804830312728882, + "eval_runtime": 34.7025, + "eval_samples_per_second": 22.073, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.005019331459687481, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0007729413967548602, + "step": 87000 + }, + { + "epoch": 174.74, + "learning_rate": 8.292365026036886e-06, + "loss": 2.2579, + "step": 87020 + }, + { + "epoch": 174.78, + "learning_rate": 8.267629062364691e-06, + "loss": 2.2381, + "step": 87040 + }, + { + "epoch": 174.82, + "learning_rate": 8.242926721709982e-06, + "loss": 2.2591, + "step": 87060 + }, + { + "epoch": 174.86, + "learning_rate": 8.218258023975006e-06, + "loss": 2.2494, + "step": 87080 + }, + { + "epoch": 174.9, + "learning_rate": 8.193622989034844e-06, + "loss": 2.2619, + "step": 87100 + }, + { + "epoch": 174.94, + "learning_rate": 8.16902163673754e-06, + "loss": 2.2534, + "step": 87120 + }, + { + "epoch": 174.98, + "learning_rate": 8.14445398690396e-06, + "loss": 2.2412, + "step": 87140 + }, + { + "epoch": 175.02, + "learning_rate": 8.119920059327785e-06, + "loss": 2.2472, + "step": 87160 + }, + { + "epoch": 175.06, + "learning_rate": 8.095419873775572e-06, + "loss": 2.2189, + "step": 87180 + }, + { + "epoch": 175.1, + "learning_rate": 8.07095344998668e-06, + "loss": 2.2434, + "step": 87200 + }, + { + "epoch": 175.14, + "learning_rate": 8.046520807673274e-06, + "loss": 2.2341, + "step": 87220 + }, + { + "epoch": 175.18, + "learning_rate": 8.022121966520263e-06, + "loss": 2.2471, + "step": 87240 + }, + { + "epoch": 175.22, + "learning_rate": 7.99775694618538e-06, + "loss": 2.2506, + "step": 87260 + }, + { + "epoch": 175.26, + "learning_rate": 7.973425766299092e-06, + "loss": 2.2475, + "step": 87280 + }, + { + "epoch": 175.3, + "learning_rate": 7.949128446464554e-06, + "loss": 2.2443, + "step": 87300 + }, + { + "epoch": 175.34, + "learning_rate": 7.924865006257703e-06, + "loss": 2.2496, + "step": 87320 + }, + { + "epoch": 175.38, + "learning_rate": 7.900635465227147e-06, + "loss": 2.2512, + "step": 87340 + }, + { + "epoch": 175.42, + "learning_rate": 7.876439842894195e-06, + "loss": 2.2457, + "step": 87360 + }, + { + "epoch": 175.46, + "learning_rate": 7.852278158752818e-06, + "loss": 2.2622, + "step": 87380 + }, + { + "epoch": 175.5, + "learning_rate": 7.828150432269643e-06, + "loss": 2.2424, + "step": 87400 + }, + { + "epoch": 175.54, + "learning_rate": 7.804056682883947e-06, + "loss": 2.2569, + "step": 87420 + }, + { + "epoch": 175.58, + "learning_rate": 7.779996930007638e-06, + "loss": 2.2523, + "step": 87440 + }, + { + "epoch": 175.62, + "learning_rate": 7.755971193025213e-06, + "loss": 2.2373, + "step": 87460 + }, + { + "epoch": 175.66, + "learning_rate": 7.731979491293784e-06, + "loss": 2.2513, + "step": 87480 + }, + { + "epoch": 175.7, + "learning_rate": 7.708021844143032e-06, + "loss": 2.2506, + "step": 87500 + }, + { + "epoch": 175.74, + "learning_rate": 7.68409827087519e-06, + "loss": 2.2418, + "step": 87520 + }, + { + "epoch": 175.78, + "learning_rate": 7.66020879076505e-06, + "loss": 2.2467, + "step": 87540 + }, + { + "epoch": 175.82, + "learning_rate": 7.636353423059938e-06, + "loss": 2.2477, + "step": 87560 + }, + { + "epoch": 175.86, + "learning_rate": 7.612532186979699e-06, + "loss": 2.2722, + "step": 87580 + }, + { + "epoch": 175.9, + "learning_rate": 7.588745101716677e-06, + "loss": 2.2495, + "step": 87600 + }, + { + "epoch": 175.94, + "learning_rate": 7.56499218643566e-06, + "loss": 2.2607, + "step": 87620 + }, + { + "epoch": 175.98, + "learning_rate": 7.541273460273996e-06, + "loss": 2.2557, + "step": 87640 + }, + { + "epoch": 176.02, + "learning_rate": 7.517588942341397e-06, + "loss": 2.2556, + "step": 87660 + }, + { + "epoch": 176.06, + "learning_rate": 7.493938651720062e-06, + "loss": 2.2342, + "step": 87680 + }, + { + "epoch": 176.1, + "learning_rate": 7.470322607464608e-06, + "loss": 2.2395, + "step": 87700 + }, + { + "epoch": 176.14, + "learning_rate": 7.446740828602044e-06, + "loss": 2.2355, + "step": 87720 + }, + { + "epoch": 176.18, + "learning_rate": 7.423193334131806e-06, + "loss": 2.2207, + "step": 87740 + }, + { + "epoch": 176.22, + "learning_rate": 7.399680143025645e-06, + "loss": 2.2422, + "step": 87760 + }, + { + "epoch": 176.27, + "learning_rate": 7.376201274227762e-06, + "loss": 2.2449, + "step": 87780 + }, + { + "epoch": 176.31, + "learning_rate": 7.352756746654627e-06, + "loss": 2.2506, + "step": 87800 + }, + { + "epoch": 176.35, + "learning_rate": 7.329346579195084e-06, + "loss": 2.2432, + "step": 87820 + }, + { + "epoch": 176.39, + "learning_rate": 7.305970790710259e-06, + "loss": 2.2587, + "step": 87840 + }, + { + "epoch": 176.43, + "learning_rate": 7.28262940003363e-06, + "loss": 2.2492, + "step": 87860 + }, + { + "epoch": 176.47, + "learning_rate": 7.259322425970938e-06, + "loss": 2.2379, + "step": 87880 + }, + { + "epoch": 176.51, + "learning_rate": 7.236049887300167e-06, + "loss": 2.2483, + "step": 87900 + }, + { + "epoch": 176.55, + "learning_rate": 7.212811802771591e-06, + "loss": 2.2469, + "step": 87920 + }, + { + "epoch": 176.59, + "learning_rate": 7.189608191107716e-06, + "loss": 2.2408, + "step": 87940 + }, + { + "epoch": 176.63, + "learning_rate": 7.166439071003284e-06, + "loss": 2.2388, + "step": 87960 + }, + { + "epoch": 176.67, + "learning_rate": 7.143304461125206e-06, + "loss": 2.2515, + "step": 87980 + }, + { + "epoch": 176.71, + "learning_rate": 7.12020438011265e-06, + "loss": 2.2556, + "step": 88000 + }, + { + "epoch": 176.71, + "eval_loss": 2.803500175476074, + "eval_runtime": 32.4862, + "eval_samples_per_second": 23.579, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.005054233176822753, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00032644806922367385, + "step": 88000 + }, + { + "epoch": 176.75, + "learning_rate": 7.097138846576928e-06, + "loss": 2.2566, + "step": 88020 + }, + { + "epoch": 176.79, + "learning_rate": 7.074107879101516e-06, + "loss": 2.2468, + "step": 88040 + }, + { + "epoch": 176.83, + "learning_rate": 7.051111496242047e-06, + "loss": 2.2601, + "step": 88060 + }, + { + "epoch": 176.87, + "learning_rate": 7.028149716526295e-06, + "loss": 2.2436, + "step": 88080 + }, + { + "epoch": 176.91, + "learning_rate": 7.005222558454161e-06, + "loss": 2.253, + "step": 88100 + }, + { + "epoch": 176.95, + "learning_rate": 6.982330040497626e-06, + "loss": 2.2481, + "step": 88120 + }, + { + "epoch": 176.99, + "learning_rate": 6.959472181100768e-06, + "loss": 2.2507, + "step": 88140 + }, + { + "epoch": 177.03, + "learning_rate": 6.9366489986797955e-06, + "loss": 2.239, + "step": 88160 + }, + { + "epoch": 177.07, + "learning_rate": 6.913860511622894e-06, + "loss": 2.2294, + "step": 88180 + }, + { + "epoch": 177.11, + "learning_rate": 6.891106738290348e-06, + "loss": 2.247, + "step": 88200 + }, + { + "epoch": 177.15, + "learning_rate": 6.868387697014466e-06, + "loss": 2.238, + "step": 88220 + }, + { + "epoch": 177.19, + "learning_rate": 6.845703406099574e-06, + "loss": 2.2518, + "step": 88240 + }, + { + "epoch": 177.23, + "learning_rate": 6.823053883821984e-06, + "loss": 2.2396, + "step": 88260 + }, + { + "epoch": 177.27, + "learning_rate": 6.800439148429993e-06, + "loss": 2.2317, + "step": 88280 + }, + { + "epoch": 177.31, + "learning_rate": 6.777859218143923e-06, + "loss": 2.2443, + "step": 88300 + }, + { + "epoch": 177.35, + "learning_rate": 6.755314111155986e-06, + "loss": 2.2442, + "step": 88320 + }, + { + "epoch": 177.39, + "learning_rate": 6.732803845630392e-06, + "loss": 2.2432, + "step": 88340 + }, + { + "epoch": 177.43, + "learning_rate": 6.710328439703212e-06, + "loss": 2.2539, + "step": 88360 + }, + { + "epoch": 177.47, + "learning_rate": 6.6878879114825265e-06, + "loss": 2.2395, + "step": 88380 + }, + { + "epoch": 177.51, + "learning_rate": 6.66548227904823e-06, + "loss": 2.2333, + "step": 88400 + }, + { + "epoch": 177.55, + "learning_rate": 6.6431115604521474e-06, + "loss": 2.2436, + "step": 88420 + }, + { + "epoch": 177.59, + "learning_rate": 6.620775773717963e-06, + "loss": 2.2368, + "step": 88440 + }, + { + "epoch": 177.63, + "learning_rate": 6.598474936841209e-06, + "loss": 2.2546, + "step": 88460 + }, + { + "epoch": 177.67, + "learning_rate": 6.576209067789285e-06, + "loss": 2.2543, + "step": 88480 + }, + { + "epoch": 177.71, + "learning_rate": 6.553978184501369e-06, + "loss": 2.2357, + "step": 88500 + }, + { + "epoch": 177.75, + "learning_rate": 6.531782304888523e-06, + "loss": 2.2387, + "step": 88520 + }, + { + "epoch": 177.79, + "learning_rate": 6.509621446833525e-06, + "loss": 2.2552, + "step": 88540 + }, + { + "epoch": 177.83, + "learning_rate": 6.487495628191003e-06, + "loss": 2.2478, + "step": 88560 + }, + { + "epoch": 177.87, + "learning_rate": 6.465404866787323e-06, + "loss": 2.2434, + "step": 88580 + }, + { + "epoch": 177.91, + "learning_rate": 6.44334918042061e-06, + "loss": 2.2497, + "step": 88600 + }, + { + "epoch": 177.95, + "learning_rate": 6.421328586860747e-06, + "loss": 2.2528, + "step": 88620 + }, + { + "epoch": 177.99, + "learning_rate": 6.399343103849309e-06, + "loss": 2.2564, + "step": 88640 + }, + { + "epoch": 178.03, + "learning_rate": 6.3773927490996034e-06, + "loss": 2.229, + "step": 88660 + }, + { + "epoch": 178.07, + "learning_rate": 6.355477540296645e-06, + "loss": 2.2342, + "step": 88680 + }, + { + "epoch": 178.11, + "learning_rate": 6.3335974950971264e-06, + "loss": 2.239, + "step": 88700 + }, + { + "epoch": 178.15, + "learning_rate": 6.311752631129375e-06, + "loss": 2.2377, + "step": 88720 + }, + { + "epoch": 178.19, + "learning_rate": 6.28994296599344e-06, + "loss": 2.2363, + "step": 88740 + }, + { + "epoch": 178.23, + "learning_rate": 6.268168517260964e-06, + "loss": 2.2455, + "step": 88760 + }, + { + "epoch": 178.27, + "learning_rate": 6.24642930247521e-06, + "loss": 2.2312, + "step": 88780 + }, + { + "epoch": 178.31, + "learning_rate": 6.224725339151088e-06, + "loss": 2.2386, + "step": 88800 + }, + { + "epoch": 178.35, + "learning_rate": 6.203056644775079e-06, + "loss": 2.2604, + "step": 88820 + }, + { + "epoch": 178.39, + "learning_rate": 6.181423236805278e-06, + "loss": 2.2351, + "step": 88840 + }, + { + "epoch": 178.43, + "learning_rate": 6.159825132671304e-06, + "loss": 2.247, + "step": 88860 + }, + { + "epoch": 178.47, + "learning_rate": 6.13826234977436e-06, + "loss": 2.2409, + "step": 88880 + }, + { + "epoch": 178.51, + "learning_rate": 6.116734905487226e-06, + "loss": 2.2363, + "step": 88900 + }, + { + "epoch": 178.55, + "learning_rate": 6.095242817154145e-06, + "loss": 2.2363, + "step": 88920 + }, + { + "epoch": 178.59, + "learning_rate": 6.073786102090906e-06, + "loss": 2.225, + "step": 88940 + }, + { + "epoch": 178.63, + "learning_rate": 6.052364777584801e-06, + "loss": 2.2426, + "step": 88960 + }, + { + "epoch": 178.67, + "learning_rate": 6.0309788608946204e-06, + "loss": 2.2518, + "step": 88980 + }, + { + "epoch": 178.71, + "learning_rate": 6.00962836925058e-06, + "loss": 2.2562, + "step": 89000 + }, + { + "epoch": 178.71, + "eval_loss": 2.801001787185669, + "eval_runtime": 32.9214, + "eval_samples_per_second": 23.268, + "eval_steps_per_second": 0.122, + "eval_tse_ndup": 0.005682577265537312, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0001450536698578474, + "step": 89000 + }, + { + "epoch": 178.76, + "learning_rate": 5.988313319854394e-06, + "loss": 2.2459, + "step": 89020 + }, + { + "epoch": 178.8, + "learning_rate": 5.96703372987924e-06, + "loss": 2.2422, + "step": 89040 + }, + { + "epoch": 178.84, + "learning_rate": 5.945789616469666e-06, + "loss": 2.2378, + "step": 89060 + }, + { + "epoch": 178.88, + "learning_rate": 5.924580996741685e-06, + "loss": 2.2502, + "step": 89080 + }, + { + "epoch": 178.92, + "learning_rate": 5.903407887782697e-06, + "loss": 2.2502, + "step": 89100 + }, + { + "epoch": 178.96, + "learning_rate": 5.882270306651505e-06, + "loss": 2.2523, + "step": 89120 + }, + { + "epoch": 179.0, + "learning_rate": 5.86222252773897e-06, + "loss": 2.2415, + "step": 89140 + }, + { + "epoch": 179.04, + "learning_rate": 5.841154274828869e-06, + "loss": 2.2444, + "step": 89160 + }, + { + "epoch": 179.08, + "learning_rate": 5.820121599903189e-06, + "loss": 2.2408, + "step": 89180 + }, + { + "epoch": 179.12, + "learning_rate": 5.799124519907545e-06, + "loss": 2.2282, + "step": 89200 + }, + { + "epoch": 179.16, + "learning_rate": 5.778163051758906e-06, + "loss": 2.2409, + "step": 89220 + }, + { + "epoch": 179.2, + "learning_rate": 5.757237212345557e-06, + "loss": 2.2336, + "step": 89240 + }, + { + "epoch": 179.24, + "learning_rate": 5.736347018527061e-06, + "loss": 2.2396, + "step": 89260 + }, + { + "epoch": 179.28, + "learning_rate": 5.715492487134289e-06, + "loss": 2.2373, + "step": 89280 + }, + { + "epoch": 179.32, + "learning_rate": 5.694673634969333e-06, + "loss": 2.2487, + "step": 89300 + }, + { + "epoch": 179.36, + "learning_rate": 5.673890478805582e-06, + "loss": 2.2337, + "step": 89320 + }, + { + "epoch": 179.4, + "learning_rate": 5.653143035387648e-06, + "loss": 2.2302, + "step": 89340 + }, + { + "epoch": 179.44, + "learning_rate": 5.632431321431364e-06, + "loss": 2.2344, + "step": 89360 + }, + { + "epoch": 179.48, + "learning_rate": 5.6117553536237935e-06, + "loss": 2.2416, + "step": 89380 + }, + { + "epoch": 179.52, + "learning_rate": 5.5911151486231884e-06, + "loss": 2.2351, + "step": 89400 + }, + { + "epoch": 179.56, + "learning_rate": 5.570510723058997e-06, + "loss": 2.2527, + "step": 89420 + }, + { + "epoch": 179.6, + "learning_rate": 5.549942093531812e-06, + "loss": 2.2435, + "step": 89440 + }, + { + "epoch": 179.64, + "learning_rate": 5.529409276613417e-06, + "loss": 2.2393, + "step": 89460 + }, + { + "epoch": 179.68, + "learning_rate": 5.508912288846735e-06, + "loss": 2.2348, + "step": 89480 + }, + { + "epoch": 179.72, + "learning_rate": 5.48845114674581e-06, + "loss": 2.2268, + "step": 89500 + }, + { + "epoch": 179.76, + "learning_rate": 5.4680258667958396e-06, + "loss": 2.2312, + "step": 89520 + }, + { + "epoch": 179.8, + "learning_rate": 5.44763646545306e-06, + "loss": 2.2398, + "step": 89540 + }, + { + "epoch": 179.84, + "learning_rate": 5.427282959144886e-06, + "loss": 2.2422, + "step": 89560 + }, + { + "epoch": 179.88, + "learning_rate": 5.406965364269745e-06, + "loss": 2.2544, + "step": 89580 + }, + { + "epoch": 179.92, + "learning_rate": 5.386683697197159e-06, + "loss": 2.2546, + "step": 89600 + }, + { + "epoch": 179.96, + "learning_rate": 5.366437974267713e-06, + "loss": 2.2443, + "step": 89620 + }, + { + "epoch": 180.0, + "learning_rate": 5.346228211793003e-06, + "loss": 2.2494, + "step": 89640 + }, + { + "epoch": 180.04, + "learning_rate": 5.3260544260556935e-06, + "loss": 2.2153, + "step": 89660 + }, + { + "epoch": 180.08, + "learning_rate": 5.305916633309404e-06, + "loss": 2.23, + "step": 89680 + }, + { + "epoch": 180.12, + "learning_rate": 5.285814849778831e-06, + "loss": 2.2274, + "step": 89700 + }, + { + "epoch": 180.16, + "learning_rate": 5.2657490916595885e-06, + "loss": 2.2398, + "step": 89720 + }, + { + "epoch": 180.2, + "learning_rate": 5.245719375118308e-06, + "loss": 2.2451, + "step": 89740 + }, + { + "epoch": 180.24, + "learning_rate": 5.225725716292562e-06, + "loss": 2.2345, + "step": 89760 + }, + { + "epoch": 180.28, + "learning_rate": 5.2057681312909e-06, + "loss": 2.251, + "step": 89780 + }, + { + "epoch": 180.32, + "learning_rate": 5.185846636192765e-06, + "loss": 2.2419, + "step": 89800 + }, + { + "epoch": 180.36, + "learning_rate": 5.165961247048551e-06, + "loss": 2.2322, + "step": 89820 + }, + { + "epoch": 180.4, + "learning_rate": 5.146111979879564e-06, + "loss": 2.2557, + "step": 89840 + }, + { + "epoch": 180.44, + "learning_rate": 5.126298850677991e-06, + "loss": 2.2288, + "step": 89860 + }, + { + "epoch": 180.48, + "learning_rate": 5.106521875406933e-06, + "loss": 2.2448, + "step": 89880 + }, + { + "epoch": 180.52, + "learning_rate": 5.086781070000302e-06, + "loss": 2.2448, + "step": 89900 + }, + { + "epoch": 180.56, + "learning_rate": 5.067076450362956e-06, + "loss": 2.2252, + "step": 89920 + }, + { + "epoch": 180.6, + "learning_rate": 5.0474080323705156e-06, + "loss": 2.2457, + "step": 89940 + }, + { + "epoch": 180.64, + "learning_rate": 5.027775831869486e-06, + "loss": 2.2447, + "step": 89960 + }, + { + "epoch": 180.68, + "learning_rate": 5.008179864677165e-06, + "loss": 2.2259, + "step": 89980 + }, + { + "epoch": 180.72, + "learning_rate": 4.9886201465816855e-06, + "loss": 2.2287, + "step": 90000 + }, + { + "epoch": 180.72, + "eval_loss": 2.7974812984466553, + "eval_runtime": 32.3185, + "eval_samples_per_second": 23.702, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.005042780224166673, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00018366086295735527, + "step": 90000 + }, + { + "epoch": 180.76, + "learning_rate": 4.969096693341957e-06, + "loss": 2.2493, + "step": 90020 + }, + { + "epoch": 180.8, + "learning_rate": 4.949609520687659e-06, + "loss": 2.2367, + "step": 90040 + }, + { + "epoch": 180.84, + "learning_rate": 4.9301586443192604e-06, + "loss": 2.2459, + "step": 90060 + }, + { + "epoch": 180.88, + "learning_rate": 4.910744079907981e-06, + "loss": 2.2356, + "step": 90080 + }, + { + "epoch": 180.92, + "learning_rate": 4.891365843095791e-06, + "loss": 2.2409, + "step": 90100 + }, + { + "epoch": 180.96, + "learning_rate": 4.87202394949538e-06, + "loss": 2.2375, + "step": 90120 + }, + { + "epoch": 181.0, + "learning_rate": 4.852718414690166e-06, + "loss": 2.2487, + "step": 90140 + }, + { + "epoch": 181.04, + "learning_rate": 4.8334492542342736e-06, + "loss": 2.2347, + "step": 90160 + }, + { + "epoch": 181.08, + "learning_rate": 4.8142164836525e-06, + "loss": 2.2238, + "step": 90180 + }, + { + "epoch": 181.12, + "learning_rate": 4.795020118440357e-06, + "loss": 2.2411, + "step": 90200 + }, + { + "epoch": 181.16, + "learning_rate": 4.775860174064001e-06, + "loss": 2.2268, + "step": 90220 + }, + { + "epoch": 181.2, + "learning_rate": 4.756736665960259e-06, + "loss": 2.2304, + "step": 90240 + }, + { + "epoch": 181.24, + "learning_rate": 4.737649609536593e-06, + "loss": 2.2379, + "step": 90260 + }, + { + "epoch": 181.29, + "learning_rate": 4.718599020171105e-06, + "loss": 2.2309, + "step": 90280 + }, + { + "epoch": 181.33, + "learning_rate": 4.699584913212507e-06, + "loss": 2.2351, + "step": 90300 + }, + { + "epoch": 181.37, + "learning_rate": 4.680607303980117e-06, + "loss": 2.2248, + "step": 90320 + }, + { + "epoch": 181.41, + "learning_rate": 4.661666207763854e-06, + "loss": 2.2342, + "step": 90340 + }, + { + "epoch": 181.45, + "learning_rate": 4.6427616398242225e-06, + "loss": 2.2272, + "step": 90360 + }, + { + "epoch": 181.49, + "learning_rate": 4.623893615392283e-06, + "loss": 2.221, + "step": 90380 + }, + { + "epoch": 181.53, + "learning_rate": 4.605062149669675e-06, + "loss": 2.2176, + "step": 90400 + }, + { + "epoch": 181.57, + "learning_rate": 4.586267257828547e-06, + "loss": 2.2464, + "step": 90420 + }, + { + "epoch": 181.61, + "learning_rate": 4.567508955011634e-06, + "loss": 2.245, + "step": 90440 + }, + { + "epoch": 181.65, + "learning_rate": 4.5487872563321435e-06, + "loss": 2.2459, + "step": 90460 + }, + { + "epoch": 181.69, + "learning_rate": 4.530102176873807e-06, + "loss": 2.2259, + "step": 90480 + }, + { + "epoch": 181.73, + "learning_rate": 4.511453731690868e-06, + "loss": 2.2488, + "step": 90500 + }, + { + "epoch": 181.77, + "learning_rate": 4.4928419358080285e-06, + "loss": 2.2429, + "step": 90520 + }, + { + "epoch": 181.81, + "learning_rate": 4.474266804220495e-06, + "loss": 2.2407, + "step": 90540 + }, + { + "epoch": 181.85, + "learning_rate": 4.4557283518938955e-06, + "loss": 2.2504, + "step": 90560 + }, + { + "epoch": 181.89, + "learning_rate": 4.437226593764338e-06, + "loss": 2.2373, + "step": 90580 + }, + { + "epoch": 181.93, + "learning_rate": 4.418761544738348e-06, + "loss": 2.2487, + "step": 90600 + }, + { + "epoch": 181.97, + "learning_rate": 4.400333219692893e-06, + "loss": 2.2488, + "step": 90620 + }, + { + "epoch": 182.01, + "learning_rate": 4.381941633475317e-06, + "loss": 2.2296, + "step": 90640 + }, + { + "epoch": 182.05, + "learning_rate": 4.363586800903419e-06, + "loss": 2.2346, + "step": 90660 + }, + { + "epoch": 182.09, + "learning_rate": 4.345268736765346e-06, + "loss": 2.2183, + "step": 90680 + }, + { + "epoch": 182.13, + "learning_rate": 4.32698745581962e-06, + "loss": 2.2323, + "step": 90700 + }, + { + "epoch": 182.17, + "learning_rate": 4.308742972795149e-06, + "loss": 2.2199, + "step": 90720 + }, + { + "epoch": 182.21, + "learning_rate": 4.2905353023911854e-06, + "loss": 2.2346, + "step": 90740 + }, + { + "epoch": 182.25, + "learning_rate": 4.272364459277317e-06, + "loss": 2.2228, + "step": 90760 + }, + { + "epoch": 182.29, + "learning_rate": 4.25423045809345e-06, + "loss": 2.2461, + "step": 90780 + }, + { + "epoch": 182.33, + "learning_rate": 4.236133313449847e-06, + "loss": 2.2288, + "step": 90800 + }, + { + "epoch": 182.37, + "learning_rate": 4.218073039927045e-06, + "loss": 2.2272, + "step": 90820 + }, + { + "epoch": 182.41, + "learning_rate": 4.200049652075866e-06, + "loss": 2.2304, + "step": 90840 + }, + { + "epoch": 182.45, + "learning_rate": 4.182063164417432e-06, + "loss": 2.2391, + "step": 90860 + }, + { + "epoch": 182.49, + "learning_rate": 4.164113591443142e-06, + "loss": 2.2345, + "step": 90880 + }, + { + "epoch": 182.53, + "learning_rate": 4.146200947614642e-06, + "loss": 2.2282, + "step": 90900 + }, + { + "epoch": 182.57, + "learning_rate": 4.128325247363807e-06, + "loss": 2.241, + "step": 90920 + }, + { + "epoch": 182.61, + "learning_rate": 4.110486505092781e-06, + "loss": 2.2317, + "step": 90940 + }, + { + "epoch": 182.65, + "learning_rate": 4.092684735173913e-06, + "loss": 2.2429, + "step": 90960 + }, + { + "epoch": 182.69, + "learning_rate": 4.0749199519497685e-06, + "loss": 2.2348, + "step": 90980 + }, + { + "epoch": 182.73, + "learning_rate": 4.057192169733115e-06, + "loss": 2.2395, + "step": 91000 + }, + { + "epoch": 182.73, + "eval_loss": 2.7962591648101807, + "eval_runtime": 32.3706, + "eval_samples_per_second": 23.663, + "eval_steps_per_second": 0.124, + "eval_tse_ndup": 0.005289853434004025, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0008227262837249782, + "step": 91000 + }, + { + "epoch": 182.77, + "learning_rate": 4.0395014028069e-06, + "loss": 2.2523, + "step": 91020 + }, + { + "epoch": 182.81, + "learning_rate": 4.021847665424267e-06, + "loss": 2.2452, + "step": 91040 + }, + { + "epoch": 182.85, + "learning_rate": 4.004230971808498e-06, + "loss": 2.2331, + "step": 91060 + }, + { + "epoch": 182.89, + "learning_rate": 3.986651336153052e-06, + "loss": 2.2553, + "step": 91080 + }, + { + "epoch": 182.93, + "learning_rate": 3.969108772621521e-06, + "loss": 2.2306, + "step": 91100 + }, + { + "epoch": 182.97, + "learning_rate": 3.951603295347639e-06, + "loss": 2.224, + "step": 91120 + }, + { + "epoch": 183.01, + "learning_rate": 3.934134918435256e-06, + "loss": 2.2329, + "step": 91140 + }, + { + "epoch": 183.05, + "learning_rate": 3.916703655958298e-06, + "loss": 2.2249, + "step": 91160 + }, + { + "epoch": 183.09, + "learning_rate": 3.899309521960859e-06, + "loss": 2.2277, + "step": 91180 + }, + { + "epoch": 183.13, + "learning_rate": 3.88195253045705e-06, + "loss": 2.2391, + "step": 91200 + }, + { + "epoch": 183.17, + "learning_rate": 3.864632695431097e-06, + "loss": 2.2429, + "step": 91220 + }, + { + "epoch": 183.21, + "learning_rate": 3.847350030837282e-06, + "loss": 2.2379, + "step": 91240 + }, + { + "epoch": 183.25, + "learning_rate": 3.830104550599922e-06, + "loss": 2.2303, + "step": 91260 + }, + { + "epoch": 183.29, + "learning_rate": 3.8128962686134085e-06, + "loss": 2.2369, + "step": 91280 + }, + { + "epoch": 183.33, + "learning_rate": 3.795725198742117e-06, + "loss": 2.2277, + "step": 91300 + }, + { + "epoch": 183.37, + "learning_rate": 3.7785913548204987e-06, + "loss": 2.2313, + "step": 91320 + }, + { + "epoch": 183.41, + "learning_rate": 3.761494750652961e-06, + "loss": 2.2261, + "step": 91340 + }, + { + "epoch": 183.45, + "learning_rate": 3.7444354000139402e-06, + "loss": 2.2337, + "step": 91360 + }, + { + "epoch": 183.49, + "learning_rate": 3.7274133166478266e-06, + "loss": 2.2363, + "step": 91380 + }, + { + "epoch": 183.53, + "learning_rate": 3.7104285142690265e-06, + "loss": 2.2344, + "step": 91400 + }, + { + "epoch": 183.57, + "learning_rate": 3.693481006561894e-06, + "loss": 2.229, + "step": 91420 + }, + { + "epoch": 183.61, + "learning_rate": 3.676570807180707e-06, + "loss": 2.2241, + "step": 91440 + }, + { + "epoch": 183.65, + "learning_rate": 3.6596979297497223e-06, + "loss": 2.222, + "step": 91460 + }, + { + "epoch": 183.69, + "learning_rate": 3.6428623878631053e-06, + "loss": 2.2271, + "step": 91480 + }, + { + "epoch": 183.73, + "learning_rate": 3.6260641950849605e-06, + "loss": 2.2197, + "step": 91500 + }, + { + "epoch": 183.78, + "learning_rate": 3.609303364949268e-06, + "loss": 2.2466, + "step": 91520 + }, + { + "epoch": 183.82, + "learning_rate": 3.5925799109599423e-06, + "loss": 2.2402, + "step": 91540 + }, + { + "epoch": 183.86, + "learning_rate": 3.5758938465907723e-06, + "loss": 2.24, + "step": 91560 + }, + { + "epoch": 183.9, + "learning_rate": 3.5592451852854036e-06, + "loss": 2.2361, + "step": 91580 + }, + { + "epoch": 183.94, + "learning_rate": 3.542633940457368e-06, + "loss": 2.239, + "step": 91600 + }, + { + "epoch": 183.98, + "learning_rate": 3.526060125490044e-06, + "loss": 2.2471, + "step": 91620 + }, + { + "epoch": 184.02, + "learning_rate": 3.509523753736671e-06, + "loss": 2.236, + "step": 91640 + }, + { + "epoch": 184.06, + "learning_rate": 3.4930248385202755e-06, + "loss": 2.2221, + "step": 91660 + }, + { + "epoch": 184.1, + "learning_rate": 3.4765633931337473e-06, + "loss": 2.2257, + "step": 91680 + }, + { + "epoch": 184.14, + "learning_rate": 3.4601394308397927e-06, + "loss": 2.2315, + "step": 91700 + }, + { + "epoch": 184.18, + "learning_rate": 3.4437529648708778e-06, + "loss": 2.2246, + "step": 91720 + }, + { + "epoch": 184.22, + "learning_rate": 3.4274040084292968e-06, + "loss": 2.2391, + "step": 91740 + }, + { + "epoch": 184.26, + "learning_rate": 3.4110925746871023e-06, + "loss": 2.2289, + "step": 91760 + }, + { + "epoch": 184.3, + "learning_rate": 3.394818676786132e-06, + "loss": 2.2289, + "step": 91780 + }, + { + "epoch": 184.34, + "learning_rate": 3.3785823278379537e-06, + "loss": 2.2303, + "step": 91800 + }, + { + "epoch": 184.38, + "learning_rate": 3.3623835409239022e-06, + "loss": 2.2291, + "step": 91820 + }, + { + "epoch": 184.42, + "learning_rate": 3.346222329095072e-06, + "loss": 2.2304, + "step": 91840 + }, + { + "epoch": 184.46, + "learning_rate": 3.3300987053722366e-06, + "loss": 2.2418, + "step": 91860 + }, + { + "epoch": 184.5, + "learning_rate": 3.3140126827459205e-06, + "loss": 2.2128, + "step": 91880 + }, + { + "epoch": 184.54, + "learning_rate": 3.297964274176324e-06, + "loss": 2.2281, + "step": 91900 + }, + { + "epoch": 184.58, + "learning_rate": 3.2819534925933817e-06, + "loss": 2.2358, + "step": 91920 + }, + { + "epoch": 184.62, + "learning_rate": 3.265980350896675e-06, + "loss": 2.2251, + "step": 91940 + }, + { + "epoch": 184.66, + "learning_rate": 3.250044861955487e-06, + "loss": 2.2286, + "step": 91960 + }, + { + "epoch": 184.7, + "learning_rate": 3.2341470386087425e-06, + "loss": 2.238, + "step": 91980 + }, + { + "epoch": 184.74, + "learning_rate": 3.2182868936650347e-06, + "loss": 2.2391, + "step": 92000 + }, + { + "epoch": 184.74, + "eval_loss": 2.7947802543640137, + "eval_runtime": 32.5532, + "eval_samples_per_second": 23.531, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.00485469302637311, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0009536928854430179, + "step": 92000 + }, + { + "epoch": 184.78, + "learning_rate": 3.202464439902608e-06, + "loss": 2.2339, + "step": 92020 + }, + { + "epoch": 184.82, + "learning_rate": 3.1866796900693052e-06, + "loss": 2.2385, + "step": 92040 + }, + { + "epoch": 184.86, + "learning_rate": 3.1709326568826357e-06, + "loss": 2.2335, + "step": 92060 + }, + { + "epoch": 184.9, + "learning_rate": 3.155223353029685e-06, + "loss": 2.233, + "step": 92080 + }, + { + "epoch": 184.94, + "learning_rate": 3.1395517911671612e-06, + "loss": 2.2267, + "step": 92100 + }, + { + "epoch": 184.98, + "learning_rate": 3.1239179839213596e-06, + "loss": 2.2375, + "step": 92120 + }, + { + "epoch": 185.02, + "learning_rate": 3.108321943888154e-06, + "loss": 2.2479, + "step": 92140 + }, + { + "epoch": 185.06, + "learning_rate": 3.0927636836329987e-06, + "loss": 2.2437, + "step": 92160 + }, + { + "epoch": 185.1, + "learning_rate": 3.0772432156908893e-06, + "loss": 2.224, + "step": 92180 + }, + { + "epoch": 185.14, + "learning_rate": 3.061760552566395e-06, + "loss": 2.2281, + "step": 92200 + }, + { + "epoch": 185.18, + "learning_rate": 3.0463157067336156e-06, + "loss": 2.2291, + "step": 92220 + }, + { + "epoch": 185.22, + "learning_rate": 3.0309086906361917e-06, + "loss": 2.2217, + "step": 92240 + }, + { + "epoch": 185.26, + "learning_rate": 3.01553951668726e-06, + "loss": 2.2171, + "step": 92260 + }, + { + "epoch": 185.3, + "learning_rate": 3.000208197269505e-06, + "loss": 2.2283, + "step": 92280 + }, + { + "epoch": 185.34, + "learning_rate": 2.984914744735101e-06, + "loss": 2.2206, + "step": 92300 + }, + { + "epoch": 185.38, + "learning_rate": 2.9696591714056754e-06, + "loss": 2.2253, + "step": 92320 + }, + { + "epoch": 185.42, + "learning_rate": 2.9544414895723902e-06, + "loss": 2.238, + "step": 92340 + }, + { + "epoch": 185.46, + "learning_rate": 2.9392617114958553e-06, + "loss": 2.2287, + "step": 92360 + }, + { + "epoch": 185.5, + "learning_rate": 2.924119849406143e-06, + "loss": 2.2299, + "step": 92380 + }, + { + "epoch": 185.54, + "learning_rate": 2.909015915502772e-06, + "loss": 2.2187, + "step": 92400 + }, + { + "epoch": 185.58, + "learning_rate": 2.8939499219547028e-06, + "loss": 2.2345, + "step": 92420 + }, + { + "epoch": 185.62, + "learning_rate": 2.8789218809003594e-06, + "loss": 2.2261, + "step": 92440 + }, + { + "epoch": 185.66, + "learning_rate": 2.8639318044475396e-06, + "loss": 2.2464, + "step": 92460 + }, + { + "epoch": 185.7, + "learning_rate": 2.848979704673488e-06, + "loss": 2.2232, + "step": 92480 + }, + { + "epoch": 185.74, + "learning_rate": 2.834065593624835e-06, + "loss": 2.2445, + "step": 92500 + }, + { + "epoch": 185.78, + "learning_rate": 2.8191894833176248e-06, + "loss": 2.2443, + "step": 92520 + }, + { + "epoch": 185.82, + "learning_rate": 2.8043513857372472e-06, + "loss": 2.2317, + "step": 92540 + }, + { + "epoch": 185.86, + "learning_rate": 2.7895513128385007e-06, + "loss": 2.2262, + "step": 92560 + }, + { + "epoch": 185.9, + "learning_rate": 2.774789276545553e-06, + "loss": 2.2288, + "step": 92580 + }, + { + "epoch": 185.94, + "learning_rate": 2.760065288751884e-06, + "loss": 2.2383, + "step": 92600 + }, + { + "epoch": 185.98, + "learning_rate": 2.7453793613203606e-06, + "loss": 2.2211, + "step": 92620 + }, + { + "epoch": 186.02, + "learning_rate": 2.730731506083162e-06, + "loss": 2.2251, + "step": 92640 + }, + { + "epoch": 186.06, + "learning_rate": 2.7161217348418144e-06, + "loss": 2.2255, + "step": 92660 + }, + { + "epoch": 186.1, + "learning_rate": 2.701550059367125e-06, + "loss": 2.2226, + "step": 92680 + }, + { + "epoch": 186.14, + "learning_rate": 2.6870164913992414e-06, + "loss": 2.2346, + "step": 92700 + }, + { + "epoch": 186.18, + "learning_rate": 2.672521042647591e-06, + "loss": 2.2156, + "step": 92720 + }, + { + "epoch": 186.22, + "learning_rate": 2.6580637247909044e-06, + "loss": 2.2271, + "step": 92740 + }, + { + "epoch": 186.27, + "learning_rate": 2.6436445494771867e-06, + "loss": 2.2341, + "step": 92760 + }, + { + "epoch": 186.31, + "learning_rate": 2.6292635283236778e-06, + "loss": 2.2329, + "step": 92780 + }, + { + "epoch": 186.35, + "learning_rate": 2.614920672916943e-06, + "loss": 2.2281, + "step": 92800 + }, + { + "epoch": 186.39, + "learning_rate": 2.6006159948127386e-06, + "loss": 2.2206, + "step": 92820 + }, + { + "epoch": 186.43, + "learning_rate": 2.5863495055360897e-06, + "loss": 2.2282, + "step": 92840 + }, + { + "epoch": 186.47, + "learning_rate": 2.572121216581247e-06, + "loss": 2.2106, + "step": 92860 + }, + { + "epoch": 186.51, + "learning_rate": 2.557931139411696e-06, + "loss": 2.2344, + "step": 92880 + }, + { + "epoch": 186.55, + "learning_rate": 2.5437792854601238e-06, + "loss": 2.2319, + "step": 92900 + }, + { + "epoch": 186.59, + "learning_rate": 2.5296656661284113e-06, + "loss": 2.2371, + "step": 92920 + }, + { + "epoch": 186.63, + "learning_rate": 2.5155902927876564e-06, + "loss": 2.2254, + "step": 92940 + }, + { + "epoch": 186.67, + "learning_rate": 2.5015531767781263e-06, + "loss": 2.224, + "step": 92960 + }, + { + "epoch": 186.71, + "learning_rate": 2.4875543294092808e-06, + "loss": 2.2309, + "step": 92980 + }, + { + "epoch": 186.75, + "learning_rate": 2.4735937619597362e-06, + "loss": 2.2385, + "step": 93000 + }, + { + "epoch": 186.75, + "eval_loss": 2.7932307720184326, + "eval_runtime": 34.6624, + "eval_samples_per_second": 22.099, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.004893996783465221, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00019499318091499953, + "step": 93000 + }, + { + "epoch": 186.79, + "learning_rate": 2.4596714856772674e-06, + "loss": 2.2285, + "step": 93020 + }, + { + "epoch": 186.83, + "learning_rate": 2.445787511778802e-06, + "loss": 2.2448, + "step": 93040 + }, + { + "epoch": 186.87, + "learning_rate": 2.4319418514504033e-06, + "loss": 2.238, + "step": 93060 + }, + { + "epoch": 186.91, + "learning_rate": 2.4181345158472758e-06, + "loss": 2.2385, + "step": 93080 + }, + { + "epoch": 186.95, + "learning_rate": 2.404365516093732e-06, + "loss": 2.229, + "step": 93100 + }, + { + "epoch": 186.99, + "learning_rate": 2.3906348632832143e-06, + "loss": 2.2222, + "step": 93120 + }, + { + "epoch": 187.03, + "learning_rate": 2.377626272045452e-06, + "loss": 2.2158, + "step": 93140 + }, + { + "epoch": 187.07, + "learning_rate": 2.3639704275643104e-06, + "loss": 2.2265, + "step": 93160 + }, + { + "epoch": 187.11, + "learning_rate": 2.3503529625717835e-06, + "loss": 2.2362, + "step": 93180 + }, + { + "epoch": 187.15, + "learning_rate": 2.336773888039223e-06, + "loss": 2.2334, + "step": 93200 + }, + { + "epoch": 187.19, + "learning_rate": 2.323233214907028e-06, + "loss": 2.2173, + "step": 93220 + }, + { + "epoch": 187.23, + "learning_rate": 2.3097309540847045e-06, + "loss": 2.241, + "step": 93240 + }, + { + "epoch": 187.27, + "learning_rate": 2.2962671164507508e-06, + "loss": 2.2318, + "step": 93260 + }, + { + "epoch": 187.31, + "learning_rate": 2.282841712852757e-06, + "loss": 2.2215, + "step": 93280 + }, + { + "epoch": 187.35, + "learning_rate": 2.269454754107303e-06, + "loss": 2.2251, + "step": 93300 + }, + { + "epoch": 187.39, + "learning_rate": 2.2561062510000506e-06, + "loss": 2.209, + "step": 93320 + }, + { + "epoch": 187.43, + "learning_rate": 2.2427962142856305e-06, + "loss": 2.2284, + "step": 93340 + }, + { + "epoch": 187.47, + "learning_rate": 2.229524654687698e-06, + "loss": 2.231, + "step": 93360 + }, + { + "epoch": 187.51, + "learning_rate": 2.2162915828989163e-06, + "loss": 2.2219, + "step": 93380 + }, + { + "epoch": 187.55, + "learning_rate": 2.2030970095809233e-06, + "loss": 2.21, + "step": 93400 + }, + { + "epoch": 187.59, + "learning_rate": 2.189940945364366e-06, + "loss": 2.2307, + "step": 93420 + }, + { + "epoch": 187.63, + "learning_rate": 2.176823400848821e-06, + "loss": 2.2276, + "step": 93440 + }, + { + "epoch": 187.67, + "learning_rate": 2.1637443866028837e-06, + "loss": 2.2275, + "step": 93460 + }, + { + "epoch": 187.71, + "learning_rate": 2.150703913164065e-06, + "loss": 2.2299, + "step": 93480 + }, + { + "epoch": 187.75, + "learning_rate": 2.1377019910388373e-06, + "loss": 2.2274, + "step": 93500 + }, + { + "epoch": 187.79, + "learning_rate": 2.1247386307026163e-06, + "loss": 2.2255, + "step": 93520 + }, + { + "epoch": 187.83, + "learning_rate": 2.1118138425997526e-06, + "loss": 2.2303, + "step": 93540 + }, + { + "epoch": 187.87, + "learning_rate": 2.098927637143522e-06, + "loss": 2.2331, + "step": 93560 + }, + { + "epoch": 187.91, + "learning_rate": 2.086080024716086e-06, + "loss": 2.2402, + "step": 93580 + }, + { + "epoch": 187.95, + "learning_rate": 2.073271015668543e-06, + "loss": 2.2302, + "step": 93600 + }, + { + "epoch": 187.99, + "learning_rate": 2.0605006203208876e-06, + "loss": 2.2294, + "step": 93620 + }, + { + "epoch": 188.03, + "learning_rate": 2.0477688489619895e-06, + "loss": 2.2267, + "step": 93640 + }, + { + "epoch": 188.07, + "learning_rate": 2.0350757118496044e-06, + "loss": 2.2224, + "step": 93660 + }, + { + "epoch": 188.11, + "learning_rate": 2.0224212192103686e-06, + "loss": 2.2179, + "step": 93680 + }, + { + "epoch": 188.15, + "learning_rate": 2.0098053812397876e-06, + "loss": 2.2084, + "step": 93700 + }, + { + "epoch": 188.19, + "learning_rate": 1.9972282081021974e-06, + "loss": 2.2257, + "step": 93720 + }, + { + "epoch": 188.23, + "learning_rate": 1.9846897099308037e-06, + "loss": 2.2267, + "step": 93740 + }, + { + "epoch": 188.27, + "learning_rate": 1.9721898968276475e-06, + "loss": 2.2126, + "step": 93760 + }, + { + "epoch": 188.31, + "learning_rate": 1.959728778863612e-06, + "loss": 2.2228, + "step": 93780 + }, + { + "epoch": 188.35, + "learning_rate": 1.9473063660783886e-06, + "loss": 2.2328, + "step": 93800 + }, + { + "epoch": 188.39, + "learning_rate": 1.934922668480477e-06, + "loss": 2.2316, + "step": 93820 + }, + { + "epoch": 188.43, + "learning_rate": 1.9225776960472284e-06, + "loss": 2.2265, + "step": 93840 + }, + { + "epoch": 188.47, + "learning_rate": 1.9102714587247385e-06, + "loss": 2.2354, + "step": 93860 + }, + { + "epoch": 188.51, + "learning_rate": 1.8980039664279302e-06, + "loss": 2.2333, + "step": 93880 + }, + { + "epoch": 188.55, + "learning_rate": 1.8857752290405039e-06, + "loss": 2.2236, + "step": 93900 + }, + { + "epoch": 188.59, + "learning_rate": 1.8735852564149225e-06, + "loss": 2.2179, + "step": 93920 + }, + { + "epoch": 188.63, + "learning_rate": 1.861434058372441e-06, + "loss": 2.2307, + "step": 93940 + }, + { + "epoch": 188.67, + "learning_rate": 1.8493216447030392e-06, + "loss": 2.2206, + "step": 93960 + }, + { + "epoch": 188.71, + "learning_rate": 1.8372480251655e-06, + "loss": 2.2289, + "step": 93980 + }, + { + "epoch": 188.76, + "learning_rate": 1.825213209487292e-06, + "loss": 2.2353, + "step": 94000 + }, + { + "epoch": 188.76, + "eval_loss": 2.792614698410034, + "eval_runtime": 32.5432, + "eval_samples_per_second": 23.538, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.005256866790202048, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0010757916481313652, + "step": 94000 + }, + { + "epoch": 188.8, + "learning_rate": 1.8132172073646592e-06, + "loss": 2.2468, + "step": 94020 + }, + { + "epoch": 188.84, + "learning_rate": 1.8012600284625648e-06, + "loss": 2.2132, + "step": 94040 + }, + { + "epoch": 188.88, + "learning_rate": 1.7893416824146914e-06, + "loss": 2.2433, + "step": 94060 + }, + { + "epoch": 188.92, + "learning_rate": 1.7774621788234468e-06, + "loss": 2.2237, + "step": 94080 + }, + { + "epoch": 188.96, + "learning_rate": 1.7656215272599086e-06, + "loss": 2.2295, + "step": 94100 + }, + { + "epoch": 189.0, + "learning_rate": 1.75381973726389e-06, + "loss": 2.2331, + "step": 94120 + }, + { + "epoch": 189.04, + "learning_rate": 1.742056818343879e-06, + "loss": 2.2052, + "step": 94140 + }, + { + "epoch": 189.08, + "learning_rate": 1.7303327799770563e-06, + "loss": 2.2359, + "step": 94160 + }, + { + "epoch": 189.12, + "learning_rate": 1.7186476316092437e-06, + "loss": 2.2086, + "step": 94180 + }, + { + "epoch": 189.16, + "learning_rate": 1.7070013826549713e-06, + "loss": 2.2237, + "step": 94200 + }, + { + "epoch": 189.2, + "learning_rate": 1.6953940424974168e-06, + "loss": 2.2239, + "step": 94220 + }, + { + "epoch": 189.24, + "learning_rate": 1.6838256204883885e-06, + "loss": 2.2264, + "step": 94240 + }, + { + "epoch": 189.28, + "learning_rate": 1.6722961259483638e-06, + "loss": 2.225, + "step": 94260 + }, + { + "epoch": 189.32, + "learning_rate": 1.6608055681664514e-06, + "loss": 2.232, + "step": 94280 + }, + { + "epoch": 189.36, + "learning_rate": 1.6493539564003845e-06, + "loss": 2.2309, + "step": 94300 + }, + { + "epoch": 189.4, + "learning_rate": 1.637941299876511e-06, + "loss": 2.2195, + "step": 94320 + }, + { + "epoch": 189.44, + "learning_rate": 1.6265676077898029e-06, + "loss": 2.2306, + "step": 94340 + }, + { + "epoch": 189.48, + "learning_rate": 1.6152328893038582e-06, + "loss": 2.2175, + "step": 94360 + }, + { + "epoch": 189.52, + "learning_rate": 1.603937153550833e-06, + "loss": 2.2305, + "step": 94380 + }, + { + "epoch": 189.56, + "learning_rate": 1.5926804096315029e-06, + "loss": 2.2243, + "step": 94400 + }, + { + "epoch": 189.6, + "learning_rate": 1.58146266661523e-06, + "loss": 2.2278, + "step": 94420 + }, + { + "epoch": 189.64, + "learning_rate": 1.5702839335399455e-06, + "loss": 2.2379, + "step": 94440 + }, + { + "epoch": 189.68, + "learning_rate": 1.5591442194121453e-06, + "loss": 2.2383, + "step": 94460 + }, + { + "epoch": 189.72, + "learning_rate": 1.5480435332069e-06, + "loss": 2.2197, + "step": 94480 + }, + { + "epoch": 189.76, + "learning_rate": 1.5369818838678385e-06, + "loss": 2.2246, + "step": 94500 + }, + { + "epoch": 189.8, + "learning_rate": 1.525959280307121e-06, + "loss": 2.2264, + "step": 94520 + }, + { + "epoch": 189.84, + "learning_rate": 1.5149757314054658e-06, + "loss": 2.2185, + "step": 94540 + }, + { + "epoch": 189.88, + "learning_rate": 1.5040312460121164e-06, + "loss": 2.2154, + "step": 94560 + }, + { + "epoch": 189.92, + "learning_rate": 1.4931258329448583e-06, + "loss": 2.2221, + "step": 94580 + }, + { + "epoch": 189.96, + "learning_rate": 1.4822595009899687e-06, + "loss": 2.2308, + "step": 94600 + }, + { + "epoch": 190.0, + "learning_rate": 1.4714322589022666e-06, + "loss": 2.2308, + "step": 94620 + }, + { + "epoch": 190.04, + "learning_rate": 1.460644115405063e-06, + "loss": 2.2098, + "step": 94640 + }, + { + "epoch": 190.08, + "learning_rate": 1.4498950791901666e-06, + "loss": 2.2203, + "step": 94660 + }, + { + "epoch": 190.12, + "learning_rate": 1.4391851589178939e-06, + "loss": 2.2231, + "step": 94680 + }, + { + "epoch": 190.16, + "learning_rate": 1.428514363217015e-06, + "loss": 2.2294, + "step": 94700 + }, + { + "epoch": 190.2, + "learning_rate": 1.4178827006848138e-06, + "loss": 2.2153, + "step": 94720 + }, + { + "epoch": 190.24, + "learning_rate": 1.4072901798870164e-06, + "loss": 2.2373, + "step": 94740 + }, + { + "epoch": 190.28, + "learning_rate": 1.3967368093578347e-06, + "loss": 2.2379, + "step": 94760 + }, + { + "epoch": 190.32, + "learning_rate": 1.3862225975999232e-06, + "loss": 2.2227, + "step": 94780 + }, + { + "epoch": 190.36, + "learning_rate": 1.3757475530843944e-06, + "loss": 2.2204, + "step": 94800 + }, + { + "epoch": 190.4, + "learning_rate": 1.365311684250814e-06, + "loss": 2.2203, + "step": 94820 + }, + { + "epoch": 190.44, + "learning_rate": 1.3549149995071508e-06, + "loss": 2.2279, + "step": 94840 + }, + { + "epoch": 190.48, + "learning_rate": 1.3445575072298377e-06, + "loss": 2.2293, + "step": 94860 + }, + { + "epoch": 190.52, + "learning_rate": 1.3342392157637162e-06, + "loss": 2.235, + "step": 94880 + }, + { + "epoch": 190.56, + "learning_rate": 1.3239601334220586e-06, + "loss": 2.2289, + "step": 94900 + }, + { + "epoch": 190.6, + "learning_rate": 1.3137202684865125e-06, + "loss": 2.2258, + "step": 94920 + }, + { + "epoch": 190.64, + "learning_rate": 1.3035196292071727e-06, + "loss": 2.2234, + "step": 94940 + }, + { + "epoch": 190.68, + "learning_rate": 1.2933582238025099e-06, + "loss": 2.2144, + "step": 94960 + }, + { + "epoch": 190.72, + "learning_rate": 1.2832360604593697e-06, + "loss": 2.2205, + "step": 94980 + }, + { + "epoch": 190.76, + "learning_rate": 1.2731531473330116e-06, + "loss": 2.2273, + "step": 95000 + }, + { + "epoch": 190.76, + "eval_loss": 2.7919530868530273, + "eval_runtime": 32.4852, + "eval_samples_per_second": 23.58, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.005757550742713724, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00033324745999826035, + "step": 95000 + }, + { + "epoch": 190.8, + "learning_rate": 1.2631094925470488e-06, + "loss": 2.2272, + "step": 95020 + }, + { + "epoch": 190.84, + "learning_rate": 1.253105104193475e-06, + "loss": 2.2243, + "step": 95040 + }, + { + "epoch": 190.88, + "learning_rate": 1.2431399903326368e-06, + "loss": 2.2207, + "step": 95060 + }, + { + "epoch": 190.92, + "learning_rate": 1.2332141589932567e-06, + "loss": 2.228, + "step": 95080 + }, + { + "epoch": 190.96, + "learning_rate": 1.223327618172404e-06, + "loss": 2.2174, + "step": 95100 + }, + { + "epoch": 191.0, + "learning_rate": 1.2134803758354684e-06, + "loss": 2.2236, + "step": 95120 + }, + { + "epoch": 191.04, + "learning_rate": 1.2036724399162037e-06, + "loss": 2.2249, + "step": 95140 + }, + { + "epoch": 191.08, + "learning_rate": 1.1939038183166885e-06, + "loss": 2.213, + "step": 95160 + }, + { + "epoch": 191.12, + "learning_rate": 1.1841745189073272e-06, + "loss": 2.2276, + "step": 95180 + }, + { + "epoch": 191.16, + "learning_rate": 1.174484549526833e-06, + "loss": 2.2279, + "step": 95200 + }, + { + "epoch": 191.2, + "learning_rate": 1.1648339179822331e-06, + "loss": 2.228, + "step": 95220 + }, + { + "epoch": 191.24, + "learning_rate": 1.1552226320488913e-06, + "loss": 2.2301, + "step": 95240 + }, + { + "epoch": 191.29, + "learning_rate": 1.1456506994704242e-06, + "loss": 2.2262, + "step": 95260 + }, + { + "epoch": 191.33, + "learning_rate": 1.136118127958774e-06, + "loss": 2.2206, + "step": 95280 + }, + { + "epoch": 191.37, + "learning_rate": 1.1266249251941586e-06, + "loss": 2.2265, + "step": 95300 + }, + { + "epoch": 191.41, + "learning_rate": 1.1171710988250927e-06, + "loss": 2.2357, + "step": 95320 + }, + { + "epoch": 191.45, + "learning_rate": 1.1082264430986532e-06, + "loss": 2.216, + "step": 95340 + }, + { + "epoch": 191.49, + "learning_rate": 1.0988494225797285e-06, + "loss": 2.233, + "step": 95360 + }, + { + "epoch": 191.53, + "learning_rate": 1.0895118008345751e-06, + "loss": 2.227, + "step": 95380 + }, + { + "epoch": 191.57, + "learning_rate": 1.0802135853863248e-06, + "loss": 2.2243, + "step": 95400 + }, + { + "epoch": 191.61, + "learning_rate": 1.0709547837263966e-06, + "loss": 2.2136, + "step": 95420 + }, + { + "epoch": 191.65, + "learning_rate": 1.061735403314429e-06, + "loss": 2.2201, + "step": 95440 + }, + { + "epoch": 191.69, + "learning_rate": 1.0525554515783187e-06, + "loss": 2.2291, + "step": 95460 + }, + { + "epoch": 191.73, + "learning_rate": 1.0434149359141888e-06, + "loss": 2.2214, + "step": 95480 + }, + { + "epoch": 191.77, + "learning_rate": 1.0343138636863815e-06, + "loss": 2.2165, + "step": 95500 + }, + { + "epoch": 191.81, + "learning_rate": 1.0252522422274647e-06, + "loss": 2.2024, + "step": 95520 + }, + { + "epoch": 191.85, + "learning_rate": 1.016230078838226e-06, + "loss": 2.2112, + "step": 95540 + }, + { + "epoch": 191.89, + "learning_rate": 1.007247380787657e-06, + "loss": 2.2375, + "step": 95560 + }, + { + "epoch": 191.93, + "learning_rate": 9.983041553129623e-07, + "loss": 2.2368, + "step": 95580 + }, + { + "epoch": 191.97, + "learning_rate": 9.894004096195287e-07, + "loss": 2.2269, + "step": 95600 + }, + { + "epoch": 192.01, + "learning_rate": 9.805361508809562e-07, + "loss": 2.224, + "step": 95620 + }, + { + "epoch": 192.05, + "learning_rate": 9.717113862389992e-07, + "loss": 2.2207, + "step": 95640 + }, + { + "epoch": 192.09, + "learning_rate": 9.629261228036257e-07, + "loss": 2.2336, + "step": 95660 + }, + { + "epoch": 192.13, + "learning_rate": 9.54180367652957e-07, + "loss": 2.2333, + "step": 95680 + }, + { + "epoch": 192.17, + "learning_rate": 9.454741278333012e-07, + "loss": 2.225, + "step": 95700 + }, + { + "epoch": 192.21, + "learning_rate": 9.368074103590974e-07, + "loss": 2.2256, + "step": 95720 + }, + { + "epoch": 192.25, + "learning_rate": 9.281802222129765e-07, + "loss": 2.218, + "step": 95740 + }, + { + "epoch": 192.29, + "learning_rate": 9.195925703457176e-07, + "loss": 2.2243, + "step": 95760 + }, + { + "epoch": 192.33, + "learning_rate": 9.110444616762193e-07, + "loss": 2.2131, + "step": 95780 + }, + { + "epoch": 192.37, + "learning_rate": 9.025359030915448e-07, + "loss": 2.2215, + "step": 95800 + }, + { + "epoch": 192.41, + "learning_rate": 8.940669014468883e-07, + "loss": 2.2335, + "step": 95820 + }, + { + "epoch": 192.45, + "learning_rate": 8.856374635655695e-07, + "loss": 2.2058, + "step": 95840 + }, + { + "epoch": 192.49, + "learning_rate": 8.772475962390337e-07, + "loss": 2.213, + "step": 95860 + }, + { + "epoch": 192.53, + "learning_rate": 8.688973062268402e-07, + "loss": 2.2265, + "step": 95880 + }, + { + "epoch": 192.57, + "learning_rate": 8.605866002566853e-07, + "loss": 2.2336, + "step": 95900 + }, + { + "epoch": 192.61, + "learning_rate": 8.523154850243353e-07, + "loss": 2.2157, + "step": 95920 + }, + { + "epoch": 192.65, + "learning_rate": 8.440839671936818e-07, + "loss": 2.232, + "step": 95940 + }, + { + "epoch": 192.69, + "learning_rate": 8.358920533967196e-07, + "loss": 2.226, + "step": 95960 + }, + { + "epoch": 192.73, + "learning_rate": 8.277397502335194e-07, + "loss": 2.2255, + "step": 95980 + }, + { + "epoch": 192.77, + "learning_rate": 8.196270642722437e-07, + "loss": 2.2244, + "step": 96000 + }, + { + "epoch": 192.77, + "eval_loss": 2.7916061878204346, + "eval_runtime": 34.1236, + "eval_samples_per_second": 22.448, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.005598931489431555, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.00036951087746272225, + "step": 96000 + }, + { + "epoch": 192.81, + "learning_rate": 8.115540020491364e-07, + "loss": 2.218, + "step": 96020 + }, + { + "epoch": 192.85, + "learning_rate": 8.035205700685167e-07, + "loss": 2.2233, + "step": 96040 + }, + { + "epoch": 192.89, + "learning_rate": 7.955267748027851e-07, + "loss": 2.2202, + "step": 96060 + }, + { + "epoch": 192.93, + "learning_rate": 7.875726226923896e-07, + "loss": 2.2312, + "step": 96080 + }, + { + "epoch": 192.97, + "learning_rate": 7.79658120145843e-07, + "loss": 2.2154, + "step": 96100 + }, + { + "epoch": 193.01, + "learning_rate": 7.717832735397335e-07, + "loss": 2.2221, + "step": 96120 + }, + { + "epoch": 193.05, + "learning_rate": 7.639480892186634e-07, + "loss": 2.2261, + "step": 96140 + }, + { + "epoch": 193.09, + "learning_rate": 7.561525734953112e-07, + "loss": 2.22, + "step": 96160 + }, + { + "epoch": 193.13, + "learning_rate": 7.483967326503805e-07, + "loss": 2.2292, + "step": 96180 + }, + { + "epoch": 193.17, + "learning_rate": 7.406805729326116e-07, + "loss": 2.2083, + "step": 96200 + }, + { + "epoch": 193.21, + "learning_rate": 7.330041005587763e-07, + "loss": 2.219, + "step": 96220 + }, + { + "epoch": 193.25, + "learning_rate": 7.253673217136658e-07, + "loss": 2.2103, + "step": 96240 + }, + { + "epoch": 193.29, + "learning_rate": 7.177702425500976e-07, + "loss": 2.2173, + "step": 96260 + }, + { + "epoch": 193.33, + "learning_rate": 7.102128691889032e-07, + "loss": 2.2153, + "step": 96280 + }, + { + "epoch": 193.37, + "learning_rate": 7.026952077189175e-07, + "loss": 2.2255, + "step": 96300 + }, + { + "epoch": 193.41, + "learning_rate": 6.952172641969901e-07, + "loss": 2.2178, + "step": 96320 + }, + { + "epoch": 193.45, + "learning_rate": 6.87779044647957e-07, + "loss": 2.2262, + "step": 96340 + }, + { + "epoch": 193.49, + "learning_rate": 6.803805550646691e-07, + "loss": 2.2259, + "step": 96360 + }, + { + "epoch": 193.53, + "learning_rate": 6.730218014079415e-07, + "loss": 2.2253, + "step": 96380 + }, + { + "epoch": 193.57, + "learning_rate": 6.657027896065982e-07, + "loss": 2.2338, + "step": 96400 + }, + { + "epoch": 193.61, + "learning_rate": 6.584235255574334e-07, + "loss": 2.2295, + "step": 96420 + }, + { + "epoch": 193.65, + "learning_rate": 6.511840151252169e-07, + "loss": 2.2277, + "step": 96440 + }, + { + "epoch": 193.69, + "learning_rate": 6.439842641426941e-07, + "loss": 2.233, + "step": 96460 + }, + { + "epoch": 193.73, + "learning_rate": 6.368242784105693e-07, + "loss": 2.2257, + "step": 96480 + }, + { + "epoch": 193.78, + "learning_rate": 6.297040636975282e-07, + "loss": 2.2342, + "step": 96500 + }, + { + "epoch": 193.82, + "learning_rate": 6.226236257401819e-07, + "loss": 2.2203, + "step": 96520 + }, + { + "epoch": 193.86, + "learning_rate": 6.15582970243117e-07, + "loss": 2.2278, + "step": 96540 + }, + { + "epoch": 193.9, + "learning_rate": 6.085821028788629e-07, + "loss": 2.208, + "step": 96560 + }, + { + "epoch": 193.94, + "learning_rate": 6.01621029287891e-07, + "loss": 2.222, + "step": 96580 + }, + { + "epoch": 193.98, + "learning_rate": 5.946997550786204e-07, + "loss": 2.2128, + "step": 96600 + }, + { + "epoch": 194.02, + "learning_rate": 5.878182858273795e-07, + "loss": 2.227, + "step": 96620 + }, + { + "epoch": 194.06, + "learning_rate": 5.809766270784666e-07, + "loss": 2.2052, + "step": 96640 + }, + { + "epoch": 194.1, + "learning_rate": 5.741747843440614e-07, + "loss": 2.2324, + "step": 96660 + }, + { + "epoch": 194.14, + "learning_rate": 5.674127631043025e-07, + "loss": 2.2295, + "step": 96680 + }, + { + "epoch": 194.18, + "learning_rate": 5.606905688072151e-07, + "loss": 2.2056, + "step": 96700 + }, + { + "epoch": 194.22, + "learning_rate": 5.540082068687558e-07, + "loss": 2.2274, + "step": 96720 + }, + { + "epoch": 194.26, + "learning_rate": 5.473656826727847e-07, + "loss": 2.234, + "step": 96740 + }, + { + "epoch": 194.3, + "learning_rate": 5.407630015710597e-07, + "loss": 2.2151, + "step": 96760 + }, + { + "epoch": 194.34, + "learning_rate": 5.342001688832421e-07, + "loss": 2.2244, + "step": 96780 + }, + { + "epoch": 194.38, + "learning_rate": 5.276771898968913e-07, + "loss": 2.2237, + "step": 96800 + }, + { + "epoch": 194.42, + "learning_rate": 5.211940698674534e-07, + "loss": 2.206, + "step": 96820 + }, + { + "epoch": 194.46, + "learning_rate": 5.147508140182556e-07, + "loss": 2.2127, + "step": 96840 + }, + { + "epoch": 194.5, + "learning_rate": 5.08347427540512e-07, + "loss": 2.208, + "step": 96860 + }, + { + "epoch": 194.54, + "learning_rate": 5.019839155933292e-07, + "loss": 2.2221, + "step": 96880 + }, + { + "epoch": 194.58, + "learning_rate": 4.956602833036561e-07, + "loss": 2.2229, + "step": 96900 + }, + { + "epoch": 194.62, + "learning_rate": 4.893765357663393e-07, + "loss": 2.2334, + "step": 96920 + }, + { + "epoch": 194.66, + "learning_rate": 4.831326780440792e-07, + "loss": 2.2312, + "step": 96940 + }, + { + "epoch": 194.7, + "learning_rate": 4.769287151674406e-07, + "loss": 2.2277, + "step": 96960 + }, + { + "epoch": 194.74, + "learning_rate": 4.707646521348363e-07, + "loss": 2.2115, + "step": 96980 + }, + { + "epoch": 194.78, + "learning_rate": 4.646404939125548e-07, + "loss": 2.2263, + "step": 97000 + }, + { + "epoch": 194.78, + "eval_loss": 2.7910659313201904, + "eval_runtime": 34.6779, + "eval_samples_per_second": 22.089, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.004969120972029905, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0004829113690826827, + "step": 97000 + }, + { + "epoch": 194.82, + "learning_rate": 4.5855624543470476e-07, + "loss": 2.2351, + "step": 97020 + }, + { + "epoch": 194.86, + "learning_rate": 4.52511911603265e-07, + "loss": 2.227, + "step": 97040 + }, + { + "epoch": 194.9, + "learning_rate": 4.465074972880401e-07, + "loss": 2.2251, + "step": 97060 + }, + { + "epoch": 194.94, + "learning_rate": 4.405430073266825e-07, + "loss": 2.2253, + "step": 97080 + }, + { + "epoch": 194.98, + "learning_rate": 4.3461844652467607e-07, + "loss": 2.225, + "step": 97100 + }, + { + "epoch": 195.02, + "learning_rate": 4.2873381965532476e-07, + "loss": 2.2203, + "step": 97120 + }, + { + "epoch": 195.06, + "learning_rate": 4.228891314597694e-07, + "loss": 2.2191, + "step": 97140 + }, + { + "epoch": 195.1, + "learning_rate": 4.170843866469709e-07, + "loss": 2.2043, + "step": 97160 + }, + { + "epoch": 195.14, + "learning_rate": 4.1131958989371055e-07, + "loss": 2.2181, + "step": 97180 + }, + { + "epoch": 195.18, + "learning_rate": 4.0559474584457303e-07, + "loss": 2.2269, + "step": 97200 + }, + { + "epoch": 195.22, + "learning_rate": 3.9990985911196876e-07, + "loss": 2.2246, + "step": 97220 + }, + { + "epoch": 195.26, + "learning_rate": 3.9426493427611177e-07, + "loss": 2.2209, + "step": 97240 + }, + { + "epoch": 195.3, + "learning_rate": 3.886599758850029e-07, + "loss": 2.2205, + "step": 97260 + }, + { + "epoch": 195.34, + "learning_rate": 3.8309498845446326e-07, + "loss": 2.2202, + "step": 97280 + }, + { + "epoch": 195.38, + "learning_rate": 3.775699764681062e-07, + "loss": 2.2134, + "step": 97300 + }, + { + "epoch": 195.42, + "learning_rate": 3.720849443773267e-07, + "loss": 2.2237, + "step": 97320 + }, + { + "epoch": 195.46, + "learning_rate": 3.6663989660132293e-07, + "loss": 2.2209, + "step": 97340 + }, + { + "epoch": 195.5, + "learning_rate": 3.6150414068175653e-07, + "loss": 2.2338, + "step": 97360 + }, + { + "epoch": 195.54, + "learning_rate": 3.5613707490825266e-07, + "loss": 2.2208, + "step": 97380 + }, + { + "epoch": 195.58, + "learning_rate": 3.508100062984354e-07, + "loss": 2.2151, + "step": 97400 + }, + { + "epoch": 195.62, + "learning_rate": 3.455229391442216e-07, + "loss": 2.2331, + "step": 97420 + }, + { + "epoch": 195.66, + "learning_rate": 3.4027587770530946e-07, + "loss": 2.2292, + "step": 97440 + }, + { + "epoch": 195.7, + "learning_rate": 3.350688262091617e-07, + "loss": 2.1997, + "step": 97460 + }, + { + "epoch": 195.74, + "learning_rate": 3.2990178885100033e-07, + "loss": 2.2254, + "step": 97480 + }, + { + "epoch": 195.78, + "learning_rate": 3.2477476979381747e-07, + "loss": 2.2258, + "step": 97500 + }, + { + "epoch": 195.82, + "learning_rate": 3.196877731683645e-07, + "loss": 2.2272, + "step": 97520 + }, + { + "epoch": 195.86, + "learning_rate": 3.146408030731407e-07, + "loss": 2.2154, + "step": 97540 + }, + { + "epoch": 195.9, + "learning_rate": 3.096338635743934e-07, + "loss": 2.2199, + "step": 97560 + }, + { + "epoch": 195.94, + "learning_rate": 3.046669587061235e-07, + "loss": 2.2416, + "step": 97580 + }, + { + "epoch": 195.98, + "learning_rate": 2.9974009247007996e-07, + "loss": 2.2142, + "step": 97600 + }, + { + "epoch": 196.02, + "learning_rate": 2.9485326883575414e-07, + "loss": 2.2172, + "step": 97620 + }, + { + "epoch": 196.06, + "learning_rate": 2.900064917403633e-07, + "loss": 2.2223, + "step": 97640 + }, + { + "epoch": 196.1, + "learning_rate": 2.851997650888727e-07, + "loss": 2.2183, + "step": 97660 + }, + { + "epoch": 196.14, + "learning_rate": 2.804330927539789e-07, + "loss": 2.2281, + "step": 97680 + }, + { + "epoch": 196.18, + "learning_rate": 2.757064785760932e-07, + "loss": 2.2094, + "step": 97700 + }, + { + "epoch": 196.22, + "learning_rate": 2.7101992636336946e-07, + "loss": 2.2199, + "step": 97720 + }, + { + "epoch": 196.27, + "learning_rate": 2.663734398916762e-07, + "loss": 2.2064, + "step": 97740 + }, + { + "epoch": 196.31, + "learning_rate": 2.617670229046132e-07, + "loss": 2.2303, + "step": 97760 + }, + { + "epoch": 196.35, + "learning_rate": 2.572006791134729e-07, + "loss": 2.2094, + "step": 97780 + }, + { + "epoch": 196.39, + "learning_rate": 2.5267441219728463e-07, + "loss": 2.2267, + "step": 97800 + }, + { + "epoch": 196.43, + "learning_rate": 2.4818822580277566e-07, + "loss": 2.2177, + "step": 97820 + }, + { + "epoch": 196.47, + "learning_rate": 2.437421235443882e-07, + "loss": 2.2096, + "step": 97840 + }, + { + "epoch": 196.51, + "learning_rate": 2.393361090042678e-07, + "loss": 2.232, + "step": 97860 + }, + { + "epoch": 196.55, + "learning_rate": 2.3497018573225837e-07, + "loss": 2.2266, + "step": 97880 + }, + { + "epoch": 196.59, + "learning_rate": 2.3064435724590717e-07, + "loss": 2.229, + "step": 97900 + }, + { + "epoch": 196.63, + "learning_rate": 2.2635862703045408e-07, + "loss": 2.2223, + "step": 97920 + }, + { + "epoch": 196.67, + "learning_rate": 2.2211299853883704e-07, + "loss": 2.2332, + "step": 97940 + }, + { + "epoch": 196.71, + "learning_rate": 2.1790747519167543e-07, + "loss": 2.213, + "step": 97960 + }, + { + "epoch": 196.75, + "learning_rate": 2.1374206037729215e-07, + "loss": 2.2247, + "step": 97980 + }, + { + "epoch": 196.79, + "learning_rate": 2.0961675745168607e-07, + "loss": 2.2142, + "step": 98000 + }, + { + "epoch": 196.79, + "eval_loss": 2.7909278869628906, + "eval_runtime": 32.5067, + "eval_samples_per_second": 23.564, + "eval_steps_per_second": 0.123, + "eval_tse_ndup": 0.00532480664652228, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.000541762110418916, + "step": 98000 + }, + { + "epoch": 196.83, + "learning_rate": 2.0553156973852627e-07, + "loss": 2.2325, + "step": 98020 + }, + { + "epoch": 196.87, + "learning_rate": 2.014865005291966e-07, + "loss": 2.2224, + "step": 98040 + }, + { + "epoch": 196.91, + "learning_rate": 1.9748155308271786e-07, + "loss": 2.2195, + "step": 98060 + }, + { + "epoch": 196.95, + "learning_rate": 1.9351673062580898e-07, + "loss": 2.2253, + "step": 98080 + }, + { + "epoch": 196.99, + "learning_rate": 1.8959203635285915e-07, + "loss": 2.2261, + "step": 98100 + }, + { + "epoch": 197.03, + "learning_rate": 1.8570747342592232e-07, + "loss": 2.2054, + "step": 98120 + }, + { + "epoch": 197.07, + "learning_rate": 1.8186304497472273e-07, + "loss": 2.2216, + "step": 98140 + }, + { + "epoch": 197.11, + "learning_rate": 1.780587540966383e-07, + "loss": 2.2241, + "step": 98160 + }, + { + "epoch": 197.15, + "learning_rate": 1.7429460385672835e-07, + "loss": 2.2185, + "step": 98180 + }, + { + "epoch": 197.19, + "learning_rate": 1.705705972876892e-07, + "loss": 2.2119, + "step": 98200 + }, + { + "epoch": 197.23, + "learning_rate": 1.6688673738989859e-07, + "loss": 2.2173, + "step": 98220 + }, + { + "epoch": 197.27, + "learning_rate": 1.6324302713136564e-07, + "loss": 2.2196, + "step": 98240 + }, + { + "epoch": 197.31, + "learning_rate": 1.5963946944776986e-07, + "loss": 2.2166, + "step": 98260 + }, + { + "epoch": 197.35, + "learning_rate": 1.5607606724242774e-07, + "loss": 2.2241, + "step": 98280 + }, + { + "epoch": 197.39, + "learning_rate": 1.5255282338630938e-07, + "loss": 2.2242, + "step": 98300 + }, + { + "epoch": 197.43, + "learning_rate": 1.4906974071802748e-07, + "loss": 2.2285, + "step": 98320 + }, + { + "epoch": 197.47, + "learning_rate": 1.4562682204383726e-07, + "loss": 2.2149, + "step": 98340 + }, + { + "epoch": 197.51, + "learning_rate": 1.4222407013764204e-07, + "loss": 2.227, + "step": 98360 + }, + { + "epoch": 197.55, + "learning_rate": 1.3886148774097106e-07, + "loss": 2.2266, + "step": 98380 + }, + { + "epoch": 197.59, + "learning_rate": 1.3553907756299611e-07, + "loss": 2.2335, + "step": 98400 + }, + { + "epoch": 197.63, + "learning_rate": 1.322568422805315e-07, + "loss": 2.2229, + "step": 98420 + }, + { + "epoch": 197.67, + "learning_rate": 1.2901478453801185e-07, + "loss": 2.2226, + "step": 98440 + }, + { + "epoch": 197.71, + "learning_rate": 1.258129069474978e-07, + "loss": 2.2217, + "step": 98460 + }, + { + "epoch": 197.75, + "learning_rate": 1.226512120886869e-07, + "loss": 2.2135, + "step": 98480 + }, + { + "epoch": 197.79, + "learning_rate": 1.1952970250890816e-07, + "loss": 2.2211, + "step": 98500 + }, + { + "epoch": 197.83, + "learning_rate": 1.1644838072309428e-07, + "loss": 2.2276, + "step": 98520 + }, + { + "epoch": 197.87, + "learning_rate": 1.1340724921380941e-07, + "loss": 2.2272, + "step": 98540 + }, + { + "epoch": 197.91, + "learning_rate": 1.1040631043125472e-07, + "loss": 2.2138, + "step": 98560 + }, + { + "epoch": 197.95, + "learning_rate": 1.0744556679321283e-07, + "loss": 2.2306, + "step": 98580 + }, + { + "epoch": 197.99, + "learning_rate": 1.045250206851145e-07, + "loss": 2.205, + "step": 98600 + }, + { + "epoch": 198.03, + "learning_rate": 1.0164467445998305e-07, + "loss": 2.2362, + "step": 98620 + }, + { + "epoch": 198.07, + "learning_rate": 9.880453043846771e-08, + "loss": 2.2202, + "step": 98640 + }, + { + "epoch": 198.11, + "learning_rate": 9.600459090882141e-08, + "loss": 2.2305, + "step": 98660 + }, + { + "epoch": 198.15, + "learning_rate": 9.324485812689521e-08, + "loss": 2.2226, + "step": 98680 + }, + { + "epoch": 198.19, + "learning_rate": 9.052533431616605e-08, + "loss": 2.2242, + "step": 98700 + }, + { + "epoch": 198.23, + "learning_rate": 8.78460216677035e-08, + "loss": 2.2248, + "step": 98720 + }, + { + "epoch": 198.27, + "learning_rate": 8.520692234018079e-08, + "loss": 2.2218, + "step": 98740 + }, + { + "epoch": 198.31, + "learning_rate": 8.260803845986376e-08, + "loss": 2.227, + "step": 98760 + }, + { + "epoch": 198.35, + "learning_rate": 8.004937212063857e-08, + "loss": 2.2127, + "step": 98780 + }, + { + "epoch": 198.39, + "learning_rate": 7.753092538397289e-08, + "loss": 2.2144, + "step": 98800 + }, + { + "epoch": 198.43, + "learning_rate": 7.505270027893252e-08, + "loss": 2.2042, + "step": 98820 + }, + { + "epoch": 198.47, + "learning_rate": 7.261469880217586e-08, + "loss": 2.2226, + "step": 98840 + }, + { + "epoch": 198.51, + "learning_rate": 7.021692291795944e-08, + "loss": 2.2164, + "step": 98860 + }, + { + "epoch": 198.55, + "learning_rate": 6.785937455813241e-08, + "loss": 2.2016, + "step": 98880 + }, + { + "epoch": 198.59, + "learning_rate": 6.554205562211979e-08, + "loss": 2.2232, + "step": 98900 + }, + { + "epoch": 198.63, + "learning_rate": 6.326496797695591e-08, + "loss": 2.2336, + "step": 98920 + }, + { + "epoch": 198.67, + "learning_rate": 6.102811345723436e-08, + "loss": 2.2152, + "step": 98940 + }, + { + "epoch": 198.71, + "learning_rate": 5.883149386515796e-08, + "loss": 2.2262, + "step": 98960 + }, + { + "epoch": 198.76, + "learning_rate": 5.6675110970499935e-08, + "loss": 2.2207, + "step": 98980 + }, + { + "epoch": 198.8, + "learning_rate": 5.455896651062053e-08, + "loss": 2.2103, + "step": 99000 + }, + { + "epoch": 198.8, + "eval_loss": 2.7908263206481934, + "eval_runtime": 34.6548, + "eval_samples_per_second": 22.104, + "eval_steps_per_second": 0.115, + "eval_tse_ndup": 0.005303010261352325, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.000559893819151147, + "step": 99000 + }, + { + "epoch": 198.84, + "learning_rate": 5.2483062190461507e-08, + "loss": 2.2325, + "step": 99020 + }, + { + "epoch": 198.88, + "learning_rate": 5.044739968253498e-08, + "loss": 2.2234, + "step": 99040 + }, + { + "epoch": 198.92, + "learning_rate": 4.845198062694567e-08, + "loss": 2.2207, + "step": 99060 + }, + { + "epoch": 198.96, + "learning_rate": 4.659360948651159e-08, + "loss": 2.2135, + "step": 99080 + }, + { + "epoch": 199.0, + "learning_rate": 4.467666975750961e-08, + "loss": 2.2245, + "step": 99100 + }, + { + "epoch": 199.04, + "learning_rate": 4.2799978130209087e-08, + "loss": 2.2213, + "step": 99120 + }, + { + "epoch": 199.08, + "learning_rate": 4.0963536116628284e-08, + "loss": 2.2218, + "step": 99140 + }, + { + "epoch": 199.12, + "learning_rate": 3.916734519636145e-08, + "loss": 2.221, + "step": 99160 + }, + { + "epoch": 199.16, + "learning_rate": 3.7411406816562076e-08, + "loss": 2.209, + "step": 99180 + }, + { + "epoch": 199.2, + "learning_rate": 3.569572239195962e-08, + "loss": 2.2179, + "step": 99200 + }, + { + "epoch": 199.24, + "learning_rate": 3.40202933048539e-08, + "loss": 2.225, + "step": 99220 + }, + { + "epoch": 199.28, + "learning_rate": 3.238512090510404e-08, + "loss": 2.213, + "step": 99240 + }, + { + "epoch": 199.32, + "learning_rate": 3.0790206510133976e-08, + "loss": 2.2239, + "step": 99260 + }, + { + "epoch": 199.36, + "learning_rate": 2.9235551404943606e-08, + "loss": 2.2338, + "step": 99280 + }, + { + "epoch": 199.4, + "learning_rate": 2.772115684209209e-08, + "loss": 2.2207, + "step": 99300 + }, + { + "epoch": 199.44, + "learning_rate": 2.6247024041692326e-08, + "loss": 2.2152, + "step": 99320 + }, + { + "epoch": 199.48, + "learning_rate": 2.4813154191433153e-08, + "loss": 2.2306, + "step": 99340 + }, + { + "epoch": 199.52, + "learning_rate": 2.341954844655714e-08, + "loss": 2.216, + "step": 99360 + }, + { + "epoch": 199.56, + "learning_rate": 2.2066207929860584e-08, + "loss": 2.2116, + "step": 99380 + }, + { + "epoch": 199.6, + "learning_rate": 2.0753133731715725e-08, + "loss": 2.2314, + "step": 99400 + }, + { + "epoch": 199.64, + "learning_rate": 1.9480326910031877e-08, + "loss": 2.22, + "step": 99420 + }, + { + "epoch": 199.68, + "learning_rate": 1.8247788490299844e-08, + "loss": 2.222, + "step": 99440 + }, + { + "epoch": 199.72, + "learning_rate": 1.7055519465541957e-08, + "loss": 2.2225, + "step": 99460 + }, + { + "epoch": 199.76, + "learning_rate": 1.5903520796350933e-08, + "loss": 2.215, + "step": 99480 + }, + { + "epoch": 199.8, + "learning_rate": 1.4791793410873223e-08, + "loss": 2.2275, + "step": 99500 + }, + { + "epoch": 199.84, + "learning_rate": 1.3720338204803451e-08, + "loss": 2.2209, + "step": 99520 + }, + { + "epoch": 199.88, + "learning_rate": 1.2689156041401085e-08, + "loss": 2.2209, + "step": 99540 + }, + { + "epoch": 199.92, + "learning_rate": 1.1698247751462666e-08, + "loss": 2.2183, + "step": 99560 + }, + { + "epoch": 199.96, + "learning_rate": 1.074761413334957e-08, + "loss": 2.2203, + "step": 99580 + }, + { + "epoch": 200.0, + "learning_rate": 9.837255952971359e-09, + "loss": 2.2169, + "step": 99600 + }, + { + "epoch": 200.04, + "learning_rate": 8.967173943780216e-09, + "loss": 2.2349, + "step": 99620 + }, + { + "epoch": 200.08, + "learning_rate": 8.137368806798717e-09, + "loss": 2.2219, + "step": 99640 + }, + { + "epoch": 200.12, + "learning_rate": 7.347841210569862e-09, + "loss": 2.2388, + "step": 99660 + }, + { + "epoch": 200.16, + "learning_rate": 6.598591791218134e-09, + "loss": 2.2275, + "step": 99680 + }, + { + "epoch": 200.2, + "learning_rate": 5.889621152382896e-09, + "loss": 2.2246, + "step": 99700 + }, + { + "epoch": 200.24, + "learning_rate": 5.220929865284996e-09, + "loss": 2.2103, + "step": 99720 + }, + { + "epoch": 200.28, + "learning_rate": 4.592518468665707e-09, + "loss": 2.2204, + "step": 99740 + }, + { + "epoch": 200.32, + "learning_rate": 4.004387468831139e-09, + "loss": 2.2311, + "step": 99760 + }, + { + "epoch": 200.36, + "learning_rate": 3.4565373396244772e-09, + "loss": 2.2103, + "step": 99780 + }, + { + "epoch": 200.4, + "learning_rate": 2.94896852243709e-09, + "loss": 2.2182, + "step": 99800 + }, + { + "epoch": 200.44, + "learning_rate": 2.4816814262140775e-09, + "loss": 2.2244, + "step": 99820 + }, + { + "epoch": 200.48, + "learning_rate": 2.0546764274376184e-09, + "loss": 2.2101, + "step": 99840 + }, + { + "epoch": 200.52, + "learning_rate": 1.6679538701325215e-09, + "loss": 2.2177, + "step": 99860 + }, + { + "epoch": 200.56, + "learning_rate": 1.3215140658828783e-09, + "loss": 2.2225, + "step": 99880 + }, + { + "epoch": 200.6, + "learning_rate": 1.0153572938043087e-09, + "loss": 2.2122, + "step": 99900 + }, + { + "epoch": 200.64, + "learning_rate": 7.49483800560613e-10, + "loss": 2.2194, + "step": 99920 + }, + { + "epoch": 200.68, + "learning_rate": 5.238938003637727e-10, + "loss": 2.2239, + "step": 99940 + }, + { + "epoch": 200.72, + "learning_rate": 3.3858747496839926e-10, + "loss": 2.2266, + "step": 99960 + }, + { + "epoch": 200.76, + "learning_rate": 1.9356497366618265e-10, + "loss": 2.2195, + "step": 99980 + }, + { + "epoch": 200.8, + "learning_rate": 8.882641330809626e-11, + "loss": 2.2254, + "step": 100000 + }, + { + "epoch": 200.8, + "eval_loss": 2.7908167839050293, + "eval_runtime": 34.1422, + "eval_samples_per_second": 22.436, + "eval_steps_per_second": 0.117, + "eval_tse_ndup": 0.005517566164021638, + "eval_tse_nnof": 0.0, + "eval_tse_nnon": 0.0, + "eval_tse_time": 0.0, + "eval_tse_type": 0.0006845493166852346, + "step": 100000 + }, + { + "epoch": 200.8, + "step": 100000, + "total_flos": 1.1151182873493504e+18, + "train_loss": 3.3319753212356567, + "train_runtime": 62203.6746, + "train_samples_per_second": 205.776, + "train_steps_per_second": 1.608 + } + ], + "max_steps": 100000, + "num_train_epochs": 201, + "total_flos": 1.1151182873493504e+18, + "trial_name": null, + "trial_params": null +}