diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -2,7 +2,7 @@ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, - "global_step": 1784, + "global_step": 892, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, @@ -10,10745 +10,5375 @@ { "epoch": 0.0, "learning_rate": 0.0, - "loss": 4.9039, + "loss": 4.8105, "step": 1 }, { "epoch": 0.0, "learning_rate": 0.0, - "loss": 5.0473, + "loss": 5.0644, "step": 2 }, { "epoch": 0.0, "learning_rate": 6.000000000000001e-08, - "loss": 5.1588, + "loss": 4.9238, "step": 3 }, { "epoch": 0.0, "learning_rate": 1.2000000000000002e-07, - "loss": 4.7, + "loss": 4.9152, "step": 4 }, { - "epoch": 0.0, + "epoch": 0.01, "learning_rate": 1.8e-07, - "loss": 4.7898, + "loss": 4.7529, "step": 5 }, { - "epoch": 0.0, + "epoch": 0.01, "learning_rate": 2.4000000000000003e-07, - "loss": 4.8486, + "loss": 4.8704, "step": 6 }, { - "epoch": 0.0, + "epoch": 0.01, "learning_rate": 3.0000000000000004e-07, - "loss": 4.7039, + "loss": 4.8358, "step": 7 }, { - "epoch": 0.0, + "epoch": 0.01, "learning_rate": 3.6e-07, - "loss": 4.8578, + "loss": 4.7615, "step": 8 }, { "epoch": 0.01, "learning_rate": 4.2e-07, - "loss": 5.0793, + "loss": 4.9183, "step": 9 }, { "epoch": 0.01, "learning_rate": 4.800000000000001e-07, - "loss": 4.7626, + "loss": 4.8171, "step": 10 }, { "epoch": 0.01, "learning_rate": 5.4e-07, - "loss": 4.8163, + "loss": 4.882, "step": 11 }, { "epoch": 0.01, "learning_rate": 6.000000000000001e-07, - "loss": 4.7764, + "loss": 4.8536, "step": 12 }, { "epoch": 0.01, "learning_rate": 6.599999999999999e-07, - "loss": 4.9868, + "loss": 4.8384, "step": 13 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 7.2e-07, - "loss": 4.601, + "loss": 4.7957, "step": 14 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 7.799999999999999e-07, - "loss": 4.7462, + "loss": 4.7335, "step": 15 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 8.4e-07, - "loss": 4.8539, + "loss": 4.8636, "step": 16 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 9e-07, - "loss": 4.8762, + "loss": 4.8799, "step": 17 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 9.600000000000001e-07, - "loss": 4.7755, + "loss": 4.9855, "step": 18 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 1.0200000000000002e-06, - "loss": 4.7324, + "loss": 4.6993, "step": 19 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 1.08e-06, - "loss": 5.0197, + "loss": 4.8669, "step": 20 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 1.14e-06, - "loss": 4.7546, + "loss": 4.796, "step": 21 }, { - "epoch": 0.01, + "epoch": 0.02, "learning_rate": 1.2000000000000002e-06, - "loss": 5.0524, + "loss": 4.7392, "step": 22 }, { - "epoch": 0.01, + "epoch": 0.03, "learning_rate": 1.26e-06, - "loss": 4.9026, + "loss": 4.7951, "step": 23 }, { - "epoch": 0.01, + "epoch": 0.03, "learning_rate": 1.3199999999999999e-06, - "loss": 4.6494, + "loss": 4.912, "step": 24 }, { - "epoch": 0.01, + "epoch": 0.03, "learning_rate": 1.38e-06, - "loss": 4.8204, + "loss": 4.7894, "step": 25 }, { - "epoch": 0.01, + "epoch": 0.03, "learning_rate": 1.44e-06, - "loss": 4.9052, + "loss": 4.9259, "step": 26 }, { - "epoch": 0.02, + "epoch": 0.03, "learning_rate": 1.5e-06, - "loss": 4.7384, + "loss": 4.6791, "step": 27 }, { - "epoch": 0.02, + "epoch": 0.03, "learning_rate": 1.5599999999999999e-06, - "loss": 4.4794, + "loss": 4.6056, "step": 28 }, { - "epoch": 0.02, + "epoch": 0.03, "learning_rate": 1.62e-06, - "loss": 4.8224, + "loss": 4.6111, "step": 29 }, { - "epoch": 0.02, + "epoch": 0.03, "learning_rate": 1.68e-06, - "loss": 4.6417, + "loss": 4.7901, "step": 30 }, { - "epoch": 0.02, + "epoch": 0.03, "learning_rate": 1.74e-06, - "loss": 4.5346, + "loss": 4.7425, "step": 31 }, { - "epoch": 0.02, - "learning_rate": 1.74e-06, - "loss": 4.8925, + "epoch": 0.04, + "learning_rate": 1.8e-06, + "loss": 4.533, "step": 32 }, { - "epoch": 0.02, - "learning_rate": 1.8e-06, - "loss": 4.6022, + "epoch": 0.04, + "learning_rate": 1.86e-06, + "loss": 4.637, "step": 33 }, { - "epoch": 0.02, - "learning_rate": 1.86e-06, - "loss": 4.5056, + "epoch": 0.04, + "learning_rate": 1.9200000000000003e-06, + "loss": 4.4931, "step": 34 }, { - "epoch": 0.02, - "learning_rate": 1.9200000000000003e-06, - "loss": 4.7537, + "epoch": 0.04, + "learning_rate": 1.98e-06, + "loss": 4.7015, "step": 35 }, { - "epoch": 0.02, - "learning_rate": 1.98e-06, - "loss": 4.9873, + "epoch": 0.04, + "learning_rate": 2.0400000000000004e-06, + "loss": 4.6452, "step": 36 }, { - "epoch": 0.02, - "learning_rate": 2.0400000000000004e-06, - "loss": 4.6271, + "epoch": 0.04, + "learning_rate": 2.1000000000000002e-06, + "loss": 4.6921, "step": 37 }, { - "epoch": 0.02, - "learning_rate": 2.1000000000000002e-06, - "loss": 4.7265, + "epoch": 0.04, + "learning_rate": 2.16e-06, + "loss": 4.6732, "step": 38 }, { - "epoch": 0.02, - "learning_rate": 2.16e-06, - "loss": 4.8251, + "epoch": 0.04, + "learning_rate": 2.22e-06, + "loss": 4.8648, "step": 39 }, { - "epoch": 0.02, - "learning_rate": 2.22e-06, - "loss": 4.816, + "epoch": 0.04, + "learning_rate": 2.28e-06, + "loss": 4.6368, "step": 40 }, { - "epoch": 0.02, - "learning_rate": 2.28e-06, - "loss": 4.6273, + "epoch": 0.05, + "learning_rate": 2.34e-06, + "loss": 4.4882, "step": 41 }, { - "epoch": 0.02, - "learning_rate": 2.34e-06, - "loss": 4.419, + "epoch": 0.05, + "learning_rate": 2.4000000000000003e-06, + "loss": 4.6836, "step": 42 }, { - "epoch": 0.02, - "learning_rate": 2.4000000000000003e-06, - "loss": 4.726, + "epoch": 0.05, + "learning_rate": 2.46e-06, + "loss": 4.4747, "step": 43 }, { - "epoch": 0.02, - "learning_rate": 2.46e-06, - "loss": 4.5527, + "epoch": 0.05, + "learning_rate": 2.52e-06, + "loss": 4.6423, "step": 44 }, { - "epoch": 0.03, - "learning_rate": 2.52e-06, - "loss": 4.8068, + "epoch": 0.05, + "learning_rate": 2.58e-06, + "loss": 4.7406, "step": 45 }, { - "epoch": 0.03, - "learning_rate": 2.58e-06, - "loss": 4.7964, + "epoch": 0.05, + "learning_rate": 2.6399999999999997e-06, + "loss": 4.871, "step": 46 }, { - "epoch": 0.03, - "learning_rate": 2.6399999999999997e-06, - "loss": 4.9131, + "epoch": 0.05, + "learning_rate": 2.7e-06, + "loss": 4.756, "step": 47 }, { - "epoch": 0.03, - "learning_rate": 2.7e-06, - "loss": 5.2795, + "epoch": 0.05, + "learning_rate": 2.76e-06, + "loss": 5.13, "step": 48 }, { - "epoch": 0.03, + "epoch": 0.05, "learning_rate": 2.76e-06, - "loss": 5.3103, + "loss": 5.2669, "step": 49 }, { - "epoch": 0.03, + "epoch": 0.06, "learning_rate": 2.82e-06, - "loss": 5.3139, + "loss": 4.9543, "step": 50 }, { - "epoch": 0.03, + "epoch": 0.06, "learning_rate": 2.88e-06, - "loss": 4.4328, + "loss": 4.3671, "step": 51 }, { - "epoch": 0.03, + "epoch": 0.06, "learning_rate": 2.9400000000000002e-06, - "loss": 4.5604, + "loss": 4.4949, "step": 52 }, { - "epoch": 0.03, + "epoch": 0.06, "learning_rate": 3e-06, - "loss": 4.562, + "loss": 4.3571, "step": 53 }, { - "epoch": 0.03, + "epoch": 0.06, "learning_rate": 3.06e-06, - "loss": 4.4274, + "loss": 4.4288, "step": 54 }, { - "epoch": 0.03, + "epoch": 0.06, "learning_rate": 3.1199999999999998e-06, - "loss": 4.4539, + "loss": 4.4257, "step": 55 }, { - "epoch": 0.03, + "epoch": 0.06, "learning_rate": 3.18e-06, - "loss": 4.576, + "loss": 4.4883, "step": 56 }, { - "epoch": 0.03, + "epoch": 0.06, "learning_rate": 3.24e-06, - "loss": 4.4693, + "loss": 4.45, "step": 57 }, { - "epoch": 0.03, + "epoch": 0.07, "learning_rate": 3.3e-06, - "loss": 4.4279, + "loss": 4.6948, "step": 58 }, { - "epoch": 0.03, + "epoch": 0.07, "learning_rate": 3.36e-06, - "loss": 4.4521, + "loss": 4.429, "step": 59 }, { - "epoch": 0.03, + "epoch": 0.07, "learning_rate": 3.4200000000000003e-06, - "loss": 4.4023, + "loss": 4.286, "step": 60 }, { - "epoch": 0.03, + "epoch": 0.07, "learning_rate": 3.48e-06, - "loss": 4.4207, + "loss": 4.5093, "step": 61 }, { - "epoch": 0.03, + "epoch": 0.07, "learning_rate": 3.54e-06, - "loss": 4.352, + "loss": 4.4479, "step": 62 }, { - "epoch": 0.04, + "epoch": 0.07, "learning_rate": 3.6e-06, - "loss": 4.7222, + "loss": 4.3196, "step": 63 }, { - "epoch": 0.04, + "epoch": 0.07, "learning_rate": 3.66e-06, - "loss": 4.2245, + "loss": 4.4664, "step": 64 }, { - "epoch": 0.04, + "epoch": 0.07, "learning_rate": 3.72e-06, - "loss": 4.4388, + "loss": 4.2118, "step": 65 }, { - "epoch": 0.04, + "epoch": 0.07, "learning_rate": 3.7800000000000002e-06, - "loss": 4.5199, + "loss": 4.5129, "step": 66 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 3.8400000000000005e-06, - "loss": 4.5636, + "loss": 4.2765, "step": 67 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 3.9e-06, - "loss": 4.5562, + "loss": 4.3555, "step": 68 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 3.96e-06, - "loss": 4.4263, + "loss": 4.2651, "step": 69 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 4.0200000000000005e-06, - "loss": 4.3701, + "loss": 4.3872, "step": 70 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 4.080000000000001e-06, - "loss": 4.4671, + "loss": 4.5608, "step": 71 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 4.14e-06, - "loss": 4.24, + "loss": 4.4479, "step": 72 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 4.2000000000000004e-06, - "loss": 4.4483, + "loss": 4.3873, "step": 73 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 4.26e-06, - "loss": 4.2266, + "loss": 4.4395, "step": 74 }, { - "epoch": 0.04, + "epoch": 0.08, "learning_rate": 4.32e-06, - "loss": 4.7742, + "loss": 4.2003, "step": 75 }, { - "epoch": 0.04, + "epoch": 0.09, "learning_rate": 4.3799999999999996e-06, - "loss": 4.5626, + "loss": 4.5215, "step": 76 }, { - "epoch": 0.04, + "epoch": 0.09, "learning_rate": 4.44e-06, - "loss": 4.3647, + "loss": 4.3927, "step": 77 }, { - "epoch": 0.04, + "epoch": 0.09, "learning_rate": 4.5e-06, - "loss": 4.2969, + "loss": 4.3106, "step": 78 }, { - "epoch": 0.04, + "epoch": 0.09, "learning_rate": 4.56e-06, - "loss": 4.2667, + "loss": 4.1722, "step": 79 }, { - "epoch": 0.04, + "epoch": 0.09, "learning_rate": 4.62e-06, - "loss": 4.6044, + "loss": 4.3752, "step": 80 }, { - "epoch": 0.05, + "epoch": 0.09, "learning_rate": 4.68e-06, - "loss": 4.2306, + "loss": 4.4628, "step": 81 }, { - "epoch": 0.05, + "epoch": 0.09, "learning_rate": 4.74e-06, - "loss": 4.3939, + "loss": 4.2682, "step": 82 }, { - "epoch": 0.05, + "epoch": 0.09, "learning_rate": 4.800000000000001e-06, - "loss": 4.3016, + "loss": 4.4709, "step": 83 }, { - "epoch": 0.05, + "epoch": 0.09, "learning_rate": 4.86e-06, - "loss": 4.3591, + "loss": 4.4093, "step": 84 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 4.92e-06, - "loss": 4.2502, + "loss": 4.3576, "step": 85 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 4.980000000000001e-06, - "loss": 4.2694, + "loss": 4.3302, "step": 86 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 5.04e-06, - "loss": 4.6084, + "loss": 4.3482, "step": 87 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 5.1e-06, - "loss": 4.2422, + "loss": 4.3374, "step": 88 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 5.16e-06, - "loss": 4.599, + "loss": 4.3922, "step": 89 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 5.22e-06, - "loss": 4.1419, + "loss": 4.2921, "step": 90 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 5.279999999999999e-06, - "loss": 4.5418, + "loss": 4.5432, "step": 91 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 5.34e-06, - "loss": 4.2179, + "loss": 4.4631, "step": 92 }, { - "epoch": 0.05, + "epoch": 0.1, "learning_rate": 5.4e-06, - "loss": 4.0608, + "loss": 4.177, "step": 93 }, { - "epoch": 0.05, + "epoch": 0.11, "learning_rate": 5.46e-06, - "loss": 4.5138, + "loss": 4.3825, "step": 94 }, { - "epoch": 0.05, + "epoch": 0.11, "learning_rate": 5.52e-06, - "loss": 4.7229, + "loss": 4.6525, "step": 95 }, { - "epoch": 0.05, + "epoch": 0.11, "learning_rate": 5.58e-06, - "loss": 4.287, + "loss": 4.5859, "step": 96 }, { - "epoch": 0.05, + "epoch": 0.11, "learning_rate": 5.64e-06, - "loss": 4.4828, + "loss": 4.3767, "step": 97 }, { - "epoch": 0.05, + "epoch": 0.11, "learning_rate": 5.7000000000000005e-06, - "loss": 4.5756, + "loss": 4.677, "step": 98 }, { - "epoch": 0.06, + "epoch": 0.11, "learning_rate": 5.76e-06, - "loss": 4.9466, + "loss": 4.4746, "step": 99 }, { - "epoch": 0.06, - "learning_rate": 5.76e-06, - "loss": 4.8034, + "epoch": 0.11, + "learning_rate": 5.82e-06, + "loss": 4.4723, "step": 100 }, { - "epoch": 0.06, - "learning_rate": 5.82e-06, - "loss": 4.367, + "epoch": 0.11, + "learning_rate": 5.8800000000000005e-06, + "loss": 4.2406, "step": 101 }, { - "epoch": 0.06, - "learning_rate": 5.8800000000000005e-06, - "loss": 4.1602, + "epoch": 0.11, + "learning_rate": 5.940000000000001e-06, + "loss": 4.2459, "step": 102 }, { - "epoch": 0.06, - "learning_rate": 5.940000000000001e-06, - "loss": 4.2882, + "epoch": 0.12, + "learning_rate": 6e-06, + "loss": 4.3816, "step": 103 }, { - "epoch": 0.06, - "learning_rate": 6e-06, - "loss": 4.2509, + "epoch": 0.12, + "learning_rate": 6.0600000000000004e-06, + "loss": 4.2937, "step": 104 }, { - "epoch": 0.06, - "learning_rate": 6.0600000000000004e-06, - "loss": 4.3068, + "epoch": 0.12, + "learning_rate": 6.12e-06, + "loss": 4.117, "step": 105 }, { - "epoch": 0.06, - "learning_rate": 6.12e-06, - "loss": 4.2813, + "epoch": 0.12, + "learning_rate": 6.18e-06, + "loss": 4.3714, "step": 106 }, { - "epoch": 0.06, - "learning_rate": 6.18e-06, - "loss": 4.3237, + "epoch": 0.12, + "learning_rate": 6.2399999999999995e-06, + "loss": 4.3265, "step": 107 }, { - "epoch": 0.06, - "learning_rate": 6.2399999999999995e-06, - "loss": 4.4125, + "epoch": 0.12, + "learning_rate": 6.3e-06, + "loss": 4.4825, "step": 108 }, { - "epoch": 0.06, - "learning_rate": 6.3e-06, - "loss": 4.2929, + "epoch": 0.12, + "learning_rate": 6.36e-06, + "loss": 4.0523, "step": 109 }, { - "epoch": 0.06, - "learning_rate": 6.36e-06, - "loss": 4.3304, + "epoch": 0.12, + "learning_rate": 6.42e-06, + "loss": 4.2058, "step": 110 }, { - "epoch": 0.06, - "learning_rate": 6.42e-06, - "loss": 4.2666, + "epoch": 0.12, + "learning_rate": 6.48e-06, + "loss": 4.4007, "step": 111 }, { - "epoch": 0.06, - "learning_rate": 6.48e-06, - "loss": 4.1765, + "epoch": 0.13, + "learning_rate": 6.54e-06, + "loss": 4.2399, "step": 112 }, { - "epoch": 0.06, - "learning_rate": 6.54e-06, - "loss": 4.2945, + "epoch": 0.13, + "learning_rate": 6.6e-06, + "loss": 4.2235, "step": 113 }, { - "epoch": 0.06, - "learning_rate": 6.6e-06, - "loss": 4.0244, + "epoch": 0.13, + "learning_rate": 6.660000000000001e-06, + "loss": 4.1655, "step": 114 }, { - "epoch": 0.06, - "learning_rate": 6.660000000000001e-06, - "loss": 4.075, + "epoch": 0.13, + "learning_rate": 6.72e-06, + "loss": 4.2307, "step": 115 }, { - "epoch": 0.07, - "learning_rate": 6.72e-06, - "loss": 4.2312, + "epoch": 0.13, + "learning_rate": 6.78e-06, + "loss": 4.0141, "step": 116 }, { - "epoch": 0.07, - "learning_rate": 6.78e-06, - "loss": 4.3817, + "epoch": 0.13, + "learning_rate": 6.840000000000001e-06, + "loss": 4.3334, "step": 117 }, { - "epoch": 0.07, - "learning_rate": 6.840000000000001e-06, - "loss": 4.24, + "epoch": 0.13, + "learning_rate": 6.900000000000001e-06, + "loss": 4.1807, "step": 118 }, { - "epoch": 0.07, - "learning_rate": 6.900000000000001e-06, - "loss": 4.3153, + "epoch": 0.13, + "learning_rate": 6.96e-06, + "loss": 4.3577, "step": 119 }, { - "epoch": 0.07, - "learning_rate": 6.96e-06, - "loss": 4.3964, + "epoch": 0.13, + "learning_rate": 7.0200000000000006e-06, + "loss": 4.3768, "step": 120 }, { - "epoch": 0.07, - "learning_rate": 7.0200000000000006e-06, - "loss": 4.4129, + "epoch": 0.14, + "learning_rate": 7.08e-06, + "loss": 4.3085, "step": 121 }, { - "epoch": 0.07, - "learning_rate": 7.08e-06, - "loss": 4.2732, + "epoch": 0.14, + "learning_rate": 7.14e-06, + "loss": 4.2463, "step": 122 }, { - "epoch": 0.07, - "learning_rate": 7.14e-06, - "loss": 4.433, + "epoch": 0.14, + "learning_rate": 7.2e-06, + "loss": 4.1331, "step": 123 }, { - "epoch": 0.07, - "learning_rate": 7.2e-06, - "loss": 4.3277, + "epoch": 0.14, + "learning_rate": 7.26e-06, + "loss": 4.3671, "step": 124 }, { - "epoch": 0.07, - "learning_rate": 7.26e-06, - "loss": 4.0291, + "epoch": 0.14, + "learning_rate": 7.32e-06, + "loss": 4.3023, "step": 125 }, { - "epoch": 0.07, - "learning_rate": 7.32e-06, - "loss": 4.63, + "epoch": 0.14, + "learning_rate": 7.3800000000000005e-06, + "loss": 4.3423, "step": 126 }, { - "epoch": 0.07, - "learning_rate": 7.3800000000000005e-06, - "loss": 4.3313, + "epoch": 0.14, + "learning_rate": 7.44e-06, + "loss": 4.4363, "step": 127 }, { - "epoch": 0.07, - "learning_rate": 7.44e-06, - "loss": 4.2944, + "epoch": 0.14, + "learning_rate": 7.5e-06, + "loss": 4.2239, "step": 128 }, { - "epoch": 0.07, - "learning_rate": 7.5e-06, - "loss": 4.1793, + "epoch": 0.14, + "learning_rate": 7.5600000000000005e-06, + "loss": 4.1795, "step": 129 }, { - "epoch": 0.07, - "learning_rate": 7.5600000000000005e-06, - "loss": 4.2588, + "epoch": 0.15, + "learning_rate": 7.62e-06, + "loss": 4.3581, "step": 130 }, { - "epoch": 0.07, - "learning_rate": 7.62e-06, - "loss": 4.5454, + "epoch": 0.15, + "learning_rate": 7.680000000000001e-06, + "loss": 4.262, "step": 131 }, { - "epoch": 0.07, - "learning_rate": 7.680000000000001e-06, - "loss": 4.221, + "epoch": 0.15, + "learning_rate": 7.74e-06, + "loss": 4.3525, "step": 132 }, { - "epoch": 0.07, - "learning_rate": 7.74e-06, - "loss": 4.3223, + "epoch": 0.15, + "learning_rate": 7.8e-06, + "loss": 4.3125, "step": 133 }, { - "epoch": 0.08, - "learning_rate": 7.8e-06, - "loss": 4.2926, + "epoch": 0.15, + "learning_rate": 7.860000000000001e-06, + "loss": 4.3189, "step": 134 }, { - "epoch": 0.08, - "learning_rate": 7.860000000000001e-06, - "loss": 4.2339, + "epoch": 0.15, + "learning_rate": 7.92e-06, + "loss": 4.2083, "step": 135 }, { - "epoch": 0.08, - "learning_rate": 7.92e-06, - "loss": 4.3912, + "epoch": 0.15, + "learning_rate": 7.98e-06, + "loss": 4.2304, "step": 136 }, { - "epoch": 0.08, - "learning_rate": 7.98e-06, - "loss": 4.1663, + "epoch": 0.15, + "learning_rate": 8.040000000000001e-06, + "loss": 4.5801, "step": 137 }, { - "epoch": 0.08, - "learning_rate": 8.040000000000001e-06, - "loss": 4.3684, + "epoch": 0.15, + "learning_rate": 8.1e-06, + "loss": 4.3233, "step": 138 }, { - "epoch": 0.08, - "learning_rate": 8.1e-06, - "loss": 4.4916, + "epoch": 0.16, + "learning_rate": 8.160000000000001e-06, + "loss": 4.36, "step": 139 }, { - "epoch": 0.08, - "learning_rate": 8.160000000000001e-06, - "loss": 4.1525, + "epoch": 0.16, + "learning_rate": 8.220000000000001e-06, + "loss": 4.1527, "step": 140 }, { - "epoch": 0.08, - "learning_rate": 8.220000000000001e-06, - "loss": 4.37, + "epoch": 0.16, + "learning_rate": 8.28e-06, + "loss": 4.4406, "step": 141 }, { - "epoch": 0.08, - "learning_rate": 8.28e-06, - "loss": 4.6219, + "epoch": 0.16, + "learning_rate": 8.340000000000001e-06, + "loss": 4.5201, "step": 142 }, { - "epoch": 0.08, - "learning_rate": 8.340000000000001e-06, - "loss": 4.2804, + "epoch": 0.16, + "learning_rate": 8.400000000000001e-06, + "loss": 4.4419, "step": 143 }, { - "epoch": 0.08, - "learning_rate": 8.400000000000001e-06, - "loss": 4.4011, + "epoch": 0.16, + "learning_rate": 8.459999999999999e-06, + "loss": 4.351, "step": 144 }, { - "epoch": 0.08, - "learning_rate": 8.459999999999999e-06, - "loss": 4.6298, + "epoch": 0.16, + "learning_rate": 8.52e-06, + "loss": 4.4423, "step": 145 }, { - "epoch": 0.08, - "learning_rate": 8.52e-06, - "loss": 4.484, + "epoch": 0.16, + "learning_rate": 8.58e-06, + "loss": 4.4382, "step": 146 }, { - "epoch": 0.08, - "learning_rate": 8.58e-06, - "loss": 4.4453, + "epoch": 0.16, + "learning_rate": 8.64e-06, + "loss": 4.5181, "step": 147 }, { - "epoch": 0.08, - "learning_rate": 8.64e-06, - "loss": 4.5693, + "epoch": 0.17, + "learning_rate": 8.7e-06, + "loss": 4.2803, "step": 148 }, { - "epoch": 0.08, - "learning_rate": 8.7e-06, - "loss": 4.7461, + "epoch": 0.17, + "learning_rate": 8.759999999999999e-06, + "loss": 4.6252, "step": 149 }, { - "epoch": 0.08, - "learning_rate": 8.759999999999999e-06, - "loss": 4.2156, + "epoch": 0.17, + "learning_rate": 8.82e-06, + "loss": 4.2255, "step": 150 }, { - "epoch": 0.08, - "learning_rate": 8.82e-06, - "loss": 4.0355, + "epoch": 0.17, + "learning_rate": 8.88e-06, + "loss": 4.3498, "step": 151 }, { - "epoch": 0.09, - "learning_rate": 8.88e-06, - "loss": 4.3699, + "epoch": 0.17, + "learning_rate": 8.939999999999999e-06, + "loss": 4.2342, "step": 152 }, { - "epoch": 0.09, - "learning_rate": 8.939999999999999e-06, - "loss": 4.0072, + "epoch": 0.17, + "learning_rate": 9e-06, + "loss": 4.2839, "step": 153 }, { - "epoch": 0.09, - "learning_rate": 9e-06, - "loss": 4.4507, + "epoch": 0.17, + "learning_rate": 9.06e-06, + "loss": 4.1556, "step": 154 }, { - "epoch": 0.09, - "learning_rate": 9.06e-06, - "loss": 4.2933, + "epoch": 0.17, + "learning_rate": 9.12e-06, + "loss": 4.0843, "step": 155 }, { - "epoch": 0.09, - "learning_rate": 9.12e-06, - "loss": 4.5302, + "epoch": 0.17, + "learning_rate": 9.18e-06, + "loss": 4.2043, "step": 156 }, { - "epoch": 0.09, - "learning_rate": 9.18e-06, - "loss": 4.3209, + "epoch": 0.18, + "learning_rate": 9.24e-06, + "loss": 4.3446, "step": 157 }, { - "epoch": 0.09, - "learning_rate": 9.24e-06, - "loss": 4.429, + "epoch": 0.18, + "learning_rate": 9.3e-06, + "loss": 4.4503, "step": 158 }, { - "epoch": 0.09, - "learning_rate": 9.3e-06, - "loss": 4.3111, + "epoch": 0.18, + "learning_rate": 9.36e-06, + "loss": 4.4753, "step": 159 }, { - "epoch": 0.09, - "learning_rate": 9.36e-06, - "loss": 4.2651, + "epoch": 0.18, + "learning_rate": 9.42e-06, + "loss": 4.1959, "step": 160 }, { - "epoch": 0.09, - "learning_rate": 9.42e-06, - "loss": 4.3154, + "epoch": 0.18, + "learning_rate": 9.48e-06, + "loss": 4.2175, "step": 161 }, { - "epoch": 0.09, - "learning_rate": 9.48e-06, - "loss": 4.4383, + "epoch": 0.18, + "learning_rate": 9.54e-06, + "loss": 4.2721, "step": 162 }, { - "epoch": 0.09, - "learning_rate": 9.54e-06, - "loss": 4.2654, + "epoch": 0.18, + "learning_rate": 9.600000000000001e-06, + "loss": 4.2518, "step": 163 }, { - "epoch": 0.09, - "learning_rate": 9.600000000000001e-06, - "loss": 4.5006, + "epoch": 0.18, + "learning_rate": 9.66e-06, + "loss": 4.3758, "step": 164 }, { - "epoch": 0.09, - "learning_rate": 9.66e-06, - "loss": 4.287, + "epoch": 0.18, + "learning_rate": 9.72e-06, + "loss": 4.2946, "step": 165 }, { - "epoch": 0.09, - "learning_rate": 9.72e-06, - "loss": 4.4946, + "epoch": 0.19, + "learning_rate": 9.780000000000001e-06, + "loss": 4.3018, "step": 166 }, { - "epoch": 0.09, - "learning_rate": 9.780000000000001e-06, - "loss": 4.3238, + "epoch": 0.19, + "learning_rate": 9.84e-06, + "loss": 4.2654, "step": 167 }, { - "epoch": 0.09, - "learning_rate": 9.84e-06, - "loss": 4.0323, + "epoch": 0.19, + "learning_rate": 9.9e-06, + "loss": 4.0493, "step": 168 }, { - "epoch": 0.09, - "learning_rate": 9.9e-06, - "loss": 4.1272, + "epoch": 0.19, + "learning_rate": 9.960000000000001e-06, + "loss": 4.0436, "step": 169 }, { - "epoch": 0.1, - "learning_rate": 9.960000000000001e-06, - "loss": 4.2639, + "epoch": 0.19, + "learning_rate": 1.002e-05, + "loss": 4.2211, "step": 170 }, { - "epoch": 0.1, - "learning_rate": 1.002e-05, - "loss": 4.3443, + "epoch": 0.19, + "learning_rate": 1.008e-05, + "loss": 4.4244, "step": 171 }, { - "epoch": 0.1, - "learning_rate": 1.008e-05, - "loss": 4.2672, + "epoch": 0.19, + "learning_rate": 1.0140000000000001e-05, + "loss": 4.3368, "step": 172 }, { - "epoch": 0.1, - "learning_rate": 1.0140000000000001e-05, - "loss": 4.2944, + "epoch": 0.19, + "learning_rate": 1.02e-05, + "loss": 4.272, "step": 173 }, { - "epoch": 0.1, - "learning_rate": 1.02e-05, - "loss": 4.1786, + "epoch": 0.2, + "learning_rate": 1.0260000000000002e-05, + "loss": 4.1146, "step": 174 }, { - "epoch": 0.1, - "learning_rate": 1.0260000000000002e-05, - "loss": 4.429, + "epoch": 0.2, + "learning_rate": 1.032e-05, + "loss": 4.244, "step": 175 }, { - "epoch": 0.1, - "learning_rate": 1.032e-05, - "loss": 4.1529, + "epoch": 0.2, + "learning_rate": 1.0379999999999999e-05, + "loss": 4.0634, "step": 176 }, { - "epoch": 0.1, - "learning_rate": 1.0379999999999999e-05, - "loss": 4.205, + "epoch": 0.2, + "learning_rate": 1.044e-05, + "loss": 4.0988, "step": 177 }, { - "epoch": 0.1, - "learning_rate": 1.044e-05, - "loss": 4.1984, + "epoch": 0.2, + "learning_rate": 1.05e-05, + "loss": 4.1709, "step": 178 }, { - "epoch": 0.1, - "learning_rate": 1.05e-05, - "loss": 3.9722, + "epoch": 0.2, + "learning_rate": 1.0559999999999999e-05, + "loss": 4.2762, "step": 179 }, { - "epoch": 0.1, - "learning_rate": 1.0559999999999999e-05, - "loss": 4.4296, + "epoch": 0.2, + "learning_rate": 1.062e-05, + "loss": 4.2628, "step": 180 }, { - "epoch": 0.1, - "learning_rate": 1.062e-05, - "loss": 4.2805, + "epoch": 0.2, + "learning_rate": 1.068e-05, + "loss": 4.304, "step": 181 }, { - "epoch": 0.1, - "learning_rate": 1.068e-05, - "loss": 4.2324, + "epoch": 0.2, + "learning_rate": 1.074e-05, + "loss": 4.1144, "step": 182 }, { - "epoch": 0.1, - "learning_rate": 1.074e-05, - "loss": 4.3509, + "epoch": 0.21, + "learning_rate": 1.08e-05, + "loss": 4.3378, "step": 183 }, { - "epoch": 0.1, - "learning_rate": 1.08e-05, - "loss": 4.3776, + "epoch": 0.21, + "learning_rate": 1.086e-05, + "loss": 4.2838, "step": 184 }, { - "epoch": 0.1, - "learning_rate": 1.086e-05, - "loss": 4.3438, + "epoch": 0.21, + "learning_rate": 1.092e-05, + "loss": 4.3166, "step": 185 }, { - "epoch": 0.1, - "learning_rate": 1.092e-05, - "loss": 4.1843, + "epoch": 0.21, + "learning_rate": 1.098e-05, + "loss": 4.1774, "step": 186 }, { - "epoch": 0.1, - "learning_rate": 1.098e-05, - "loss": 4.242, + "epoch": 0.21, + "learning_rate": 1.104e-05, + "loss": 4.3944, "step": 187 }, { - "epoch": 0.11, - "learning_rate": 1.104e-05, - "loss": 4.4717, + "epoch": 0.21, + "learning_rate": 1.11e-05, + "loss": 4.1116, "step": 188 }, { - "epoch": 0.11, - "learning_rate": 1.11e-05, - "loss": 4.1776, + "epoch": 0.21, + "learning_rate": 1.116e-05, + "loss": 4.2198, "step": 189 }, { - "epoch": 0.11, - "learning_rate": 1.116e-05, - "loss": 4.208, + "epoch": 0.21, + "learning_rate": 1.1220000000000001e-05, + "loss": 4.3477, "step": 190 }, { - "epoch": 0.11, - "learning_rate": 1.1220000000000001e-05, - "loss": 4.4483, + "epoch": 0.21, + "learning_rate": 1.128e-05, + "loss": 4.3497, "step": 191 }, { - "epoch": 0.11, - "learning_rate": 1.128e-05, - "loss": 4.1557, + "epoch": 0.22, + "learning_rate": 1.134e-05, + "loss": 4.2842, "step": 192 }, { - "epoch": 0.11, - "learning_rate": 1.134e-05, - "loss": 4.0977, + "epoch": 0.22, + "learning_rate": 1.1400000000000001e-05, + "loss": 4.5254, "step": 193 }, { - "epoch": 0.11, - "learning_rate": 1.1400000000000001e-05, - "loss": 4.2601, + "epoch": 0.22, + "learning_rate": 1.146e-05, + "loss": 4.2371, "step": 194 }, { - "epoch": 0.11, - "learning_rate": 1.146e-05, - "loss": 4.5423, + "epoch": 0.22, + "learning_rate": 1.152e-05, + "loss": 4.2678, "step": 195 }, { - "epoch": 0.11, - "learning_rate": 1.152e-05, - "loss": 4.3749, + "epoch": 0.22, + "learning_rate": 1.1580000000000001e-05, + "loss": 4.5069, "step": 196 }, { - "epoch": 0.11, - "learning_rate": 1.1580000000000001e-05, - "loss": 4.3546, + "epoch": 0.22, + "learning_rate": 1.164e-05, + "loss": 4.4958, "step": 197 }, { - "epoch": 0.11, - "learning_rate": 1.164e-05, - "loss": 4.3407, + "epoch": 0.22, + "learning_rate": 1.1700000000000001e-05, + "loss": 4.4464, "step": 198 }, { - "epoch": 0.11, - "learning_rate": 1.1700000000000001e-05, - "loss": 4.4539, + "epoch": 0.22, + "learning_rate": 1.1760000000000001e-05, + "loss": 4.6181, "step": 199 }, { - "epoch": 0.11, - "learning_rate": 1.1760000000000001e-05, - "loss": 4.2983, + "epoch": 0.22, + "learning_rate": 1.182e-05, + "loss": 4.6803, "step": 200 }, { - "epoch": 0.11, - "learning_rate": 1.182e-05, - "loss": 4.2789, + "epoch": 0.23, + "learning_rate": 1.1880000000000001e-05, + "loss": 4.0903, "step": 201 }, { - "epoch": 0.11, - "learning_rate": 1.1880000000000001e-05, - "loss": 4.1983, + "epoch": 0.23, + "learning_rate": 1.1940000000000001e-05, + "loss": 4.2021, "step": 202 }, { - "epoch": 0.11, - "learning_rate": 1.1940000000000001e-05, - "loss": 4.3792, + "epoch": 0.23, + "learning_rate": 1.2e-05, + "loss": 4.0922, "step": 203 }, { - "epoch": 0.11, - "learning_rate": 1.2e-05, - "loss": 4.059, + "epoch": 0.23, + "learning_rate": 1.2060000000000001e-05, + "loss": 4.0967, "step": 204 }, { - "epoch": 0.11, - "learning_rate": 1.2060000000000001e-05, - "loss": 4.2062, + "epoch": 0.23, + "learning_rate": 1.2120000000000001e-05, + "loss": 4.0051, "step": 205 }, { - "epoch": 0.12, - "learning_rate": 1.2120000000000001e-05, - "loss": 4.3069, + "epoch": 0.23, + "learning_rate": 1.2180000000000002e-05, + "loss": 4.07, "step": 206 }, { - "epoch": 0.12, - "learning_rate": 1.2180000000000002e-05, - "loss": 4.3501, + "epoch": 0.23, + "learning_rate": 1.224e-05, + "loss": 4.1579, "step": 207 }, { - "epoch": 0.12, - "learning_rate": 1.224e-05, - "loss": 4.0598, + "epoch": 0.23, + "learning_rate": 1.2299999999999999e-05, + "loss": 3.9931, "step": 208 }, { - "epoch": 0.12, - "learning_rate": 1.2299999999999999e-05, - "loss": 4.1062, + "epoch": 0.23, + "learning_rate": 1.236e-05, + "loss": 4.2358, "step": 209 }, { - "epoch": 0.12, - "learning_rate": 1.236e-05, - "loss": 4.2752, + "epoch": 0.24, + "learning_rate": 1.242e-05, + "loss": 4.1737, "step": 210 }, { - "epoch": 0.12, - "learning_rate": 1.242e-05, - "loss": 4.2774, + "epoch": 0.24, + "learning_rate": 1.2479999999999999e-05, + "loss": 4.1183, "step": 211 }, { - "epoch": 0.12, - "learning_rate": 1.2479999999999999e-05, - "loss": 4.2148, + "epoch": 0.24, + "learning_rate": 1.254e-05, + "loss": 4.2407, "step": 212 }, { - "epoch": 0.12, - "learning_rate": 1.254e-05, - "loss": 4.1994, + "epoch": 0.24, + "learning_rate": 1.26e-05, + "loss": 4.1897, "step": 213 }, { - "epoch": 0.12, - "learning_rate": 1.26e-05, - "loss": 4.0105, + "epoch": 0.24, + "learning_rate": 1.2659999999999999e-05, + "loss": 4.2462, "step": 214 }, { - "epoch": 0.12, - "learning_rate": 1.2659999999999999e-05, - "loss": 4.4023, + "epoch": 0.24, + "learning_rate": 1.272e-05, + "loss": 4.1168, "step": 215 }, { - "epoch": 0.12, - "learning_rate": 1.272e-05, - "loss": 3.8713, + "epoch": 0.24, + "learning_rate": 1.278e-05, + "loss": 4.2695, "step": 216 }, { - "epoch": 0.12, - "learning_rate": 1.278e-05, - "loss": 4.1284, + "epoch": 0.24, + "learning_rate": 1.284e-05, + "loss": 4.2268, "step": 217 }, { - "epoch": 0.12, - "learning_rate": 1.284e-05, - "loss": 4.2426, + "epoch": 0.24, + "learning_rate": 1.29e-05, + "loss": 4.1414, "step": 218 }, { - "epoch": 0.12, - "learning_rate": 1.29e-05, - "loss": 4.1782, + "epoch": 0.25, + "learning_rate": 1.296e-05, + "loss": 4.2381, "step": 219 }, { - "epoch": 0.12, - "learning_rate": 1.296e-05, - "loss": 4.4739, + "epoch": 0.25, + "learning_rate": 1.302e-05, + "loss": 4.1982, "step": 220 }, { - "epoch": 0.12, - "learning_rate": 1.302e-05, - "loss": 4.3053, + "epoch": 0.25, + "learning_rate": 1.308e-05, + "loss": 4.1713, "step": 221 }, { - "epoch": 0.12, - "learning_rate": 1.308e-05, - "loss": 4.3529, + "epoch": 0.25, + "learning_rate": 1.314e-05, + "loss": 4.2545, "step": 222 }, { - "epoch": 0.12, - "learning_rate": 1.314e-05, - "loss": 4.1366, + "epoch": 0.25, + "learning_rate": 1.32e-05, + "loss": 4.2899, "step": 223 }, { - "epoch": 0.13, - "learning_rate": 1.32e-05, - "loss": 4.2963, + "epoch": 0.25, + "learning_rate": 1.326e-05, + "loss": 4.2193, "step": 224 }, { - "epoch": 0.13, - "learning_rate": 1.326e-05, - "loss": 4.3181, + "epoch": 0.25, + "learning_rate": 1.3320000000000001e-05, + "loss": 4.3013, "step": 225 }, { - "epoch": 0.13, - "learning_rate": 1.3320000000000001e-05, - "loss": 4.4328, + "epoch": 0.25, + "learning_rate": 1.338e-05, + "loss": 4.4583, "step": 226 }, { - "epoch": 0.13, - "learning_rate": 1.338e-05, - "loss": 4.3516, + "epoch": 0.25, + "learning_rate": 1.344e-05, + "loss": 4.212, "step": 227 }, { - "epoch": 0.13, - "learning_rate": 1.344e-05, - "loss": 4.0431, + "epoch": 0.26, + "learning_rate": 1.3500000000000001e-05, + "loss": 4.1256, "step": 228 }, { - "epoch": 0.13, - "learning_rate": 1.3500000000000001e-05, - "loss": 4.2322, + "epoch": 0.26, + "learning_rate": 1.356e-05, + "loss": 4.1234, "step": 229 }, { - "epoch": 0.13, - "learning_rate": 1.356e-05, - "loss": 4.2473, + "epoch": 0.26, + "learning_rate": 1.362e-05, + "loss": 4.204, "step": 230 }, { - "epoch": 0.13, - "learning_rate": 1.362e-05, - "loss": 4.2337, + "epoch": 0.26, + "learning_rate": 1.3680000000000001e-05, + "loss": 4.1709, "step": 231 }, { - "epoch": 0.13, - "learning_rate": 1.3680000000000001e-05, - "loss": 4.2574, + "epoch": 0.26, + "learning_rate": 1.374e-05, + "loss": 4.3005, "step": 232 }, { - "epoch": 0.13, - "learning_rate": 1.374e-05, - "loss": 4.3965, + "epoch": 0.26, + "learning_rate": 1.3800000000000002e-05, + "loss": 4.1864, "step": 233 }, { - "epoch": 0.13, - "learning_rate": 1.3800000000000002e-05, - "loss": 4.2948, + "epoch": 0.26, + "learning_rate": 1.3860000000000001e-05, + "loss": 4.2975, "step": 234 }, { - "epoch": 0.13, - "learning_rate": 1.3860000000000001e-05, - "loss": 4.121, + "epoch": 0.26, + "learning_rate": 1.392e-05, + "loss": 4.3514, "step": 235 }, { - "epoch": 0.13, - "learning_rate": 1.392e-05, - "loss": 4.6244, + "epoch": 0.26, + "learning_rate": 1.3980000000000002e-05, + "loss": 4.2039, "step": 236 }, { - "epoch": 0.13, - "learning_rate": 1.3980000000000002e-05, - "loss": 4.3004, + "epoch": 0.27, + "learning_rate": 1.4040000000000001e-05, + "loss": 4.2195, "step": 237 }, { - "epoch": 0.13, - "learning_rate": 1.4040000000000001e-05, - "loss": 4.2864, + "epoch": 0.27, + "learning_rate": 1.4099999999999999e-05, + "loss": 4.4241, "step": 238 }, { - "epoch": 0.13, - "learning_rate": 1.4099999999999999e-05, - "loss": 4.2729, + "epoch": 0.27, + "learning_rate": 1.416e-05, + "loss": 4.2328, "step": 239 }, { - "epoch": 0.13, - "learning_rate": 1.416e-05, - "loss": 4.2111, + "epoch": 0.27, + "learning_rate": 1.422e-05, + "loss": 4.3521, "step": 240 }, { - "epoch": 0.14, - "learning_rate": 1.422e-05, - "loss": 4.4951, + "epoch": 0.27, + "learning_rate": 1.428e-05, + "loss": 4.3209, "step": 241 }, { - "epoch": 0.14, - "learning_rate": 1.428e-05, - "loss": 4.2367, + "epoch": 0.27, + "learning_rate": 1.434e-05, + "loss": 4.4751, "step": 242 }, { - "epoch": 0.14, - "learning_rate": 1.434e-05, - "loss": 4.3396, + "epoch": 0.27, + "learning_rate": 1.44e-05, + "loss": 4.327, "step": 243 }, { - "epoch": 0.14, - "learning_rate": 1.44e-05, - "loss": 4.4672, + "epoch": 0.27, + "learning_rate": 1.446e-05, + "loss": 4.2864, "step": 244 }, { - "epoch": 0.14, - "learning_rate": 1.446e-05, - "loss": 4.6007, + "epoch": 0.27, + "learning_rate": 1.452e-05, + "loss": 4.4874, "step": 245 }, { - "epoch": 0.14, - "learning_rate": 1.452e-05, - "loss": 4.5437, + "epoch": 0.28, + "learning_rate": 1.458e-05, + "loss": 4.3515, "step": 246 }, { - "epoch": 0.14, - "learning_rate": 1.458e-05, - "loss": 4.3376, + "epoch": 0.28, + "learning_rate": 1.464e-05, + "loss": 4.7349, "step": 247 }, { - "epoch": 0.14, - "learning_rate": 1.464e-05, - "loss": 4.4152, + "epoch": 0.28, + "learning_rate": 1.47e-05, + "loss": 4.7552, "step": 248 }, { - "epoch": 0.14, - "learning_rate": 1.47e-05, - "loss": 4.2067, + "epoch": 0.28, + "learning_rate": 1.4760000000000001e-05, + "loss": 4.5734, "step": 249 }, { - "epoch": 0.14, - "learning_rate": 1.4760000000000001e-05, - "loss": 4.2647, + "epoch": 0.28, + "learning_rate": 1.482e-05, + "loss": 4.2063, "step": 250 }, { - "epoch": 0.14, - "learning_rate": 1.482e-05, - "loss": 4.1991, + "epoch": 0.28, + "learning_rate": 1.488e-05, + "loss": 4.3978, "step": 251 }, { - "epoch": 0.14, - "learning_rate": 1.488e-05, - "loss": 4.1397, + "epoch": 0.28, + "learning_rate": 1.4940000000000001e-05, + "loss": 4.0939, "step": 252 }, { - "epoch": 0.14, - "learning_rate": 1.4940000000000001e-05, - "loss": 4.1763, + "epoch": 0.28, + "learning_rate": 1.5e-05, + "loss": 4.1101, "step": 253 }, { - "epoch": 0.14, - "learning_rate": 1.5e-05, - "loss": 4.3289, + "epoch": 0.28, + "learning_rate": 1.506e-05, + "loss": 4.2572, "step": 254 }, { - "epoch": 0.14, - "learning_rate": 1.506e-05, - "loss": 4.2419, + "epoch": 0.29, + "learning_rate": 1.5120000000000001e-05, + "loss": 4.125, "step": 255 }, { - "epoch": 0.14, - "learning_rate": 1.5120000000000001e-05, - "loss": 4.1833, + "epoch": 0.29, + "learning_rate": 1.518e-05, + "loss": 4.1, "step": 256 }, { - "epoch": 0.14, - "learning_rate": 1.518e-05, - "loss": 4.2139, + "epoch": 0.29, + "learning_rate": 1.524e-05, + "loss": 4.2977, "step": 257 }, { - "epoch": 0.14, - "learning_rate": 1.524e-05, - "loss": 4.3484, + "epoch": 0.29, + "learning_rate": 1.53e-05, + "loss": 4.1329, "step": 258 }, { - "epoch": 0.15, - "learning_rate": 1.53e-05, - "loss": 4.4464, + "epoch": 0.29, + "learning_rate": 1.5360000000000002e-05, + "loss": 4.1073, "step": 259 }, { - "epoch": 0.15, - "learning_rate": 1.5360000000000002e-05, - "loss": 3.948, + "epoch": 0.29, + "learning_rate": 1.542e-05, + "loss": 4.2496, "step": 260 }, { - "epoch": 0.15, - "learning_rate": 1.542e-05, - "loss": 4.0849, + "epoch": 0.29, + "learning_rate": 1.548e-05, + "loss": 4.2685, "step": 261 }, { - "epoch": 0.15, - "learning_rate": 1.548e-05, - "loss": 4.3158, + "epoch": 0.29, + "learning_rate": 1.554e-05, + "loss": 4.0245, "step": 262 }, { - "epoch": 0.15, - "learning_rate": 1.554e-05, - "loss": 4.2621, + "epoch": 0.29, + "learning_rate": 1.56e-05, + "loss": 4.1365, "step": 263 }, { - "epoch": 0.15, - "learning_rate": 1.56e-05, - "loss": 4.0818, + "epoch": 0.3, + "learning_rate": 1.5660000000000003e-05, + "loss": 4.1303, "step": 264 }, { - "epoch": 0.15, - "learning_rate": 1.5660000000000003e-05, - "loss": 4.1697, + "epoch": 0.3, + "learning_rate": 1.5720000000000002e-05, + "loss": 4.2422, "step": 265 }, { - "epoch": 0.15, - "learning_rate": 1.5720000000000002e-05, - "loss": 4.0042, + "epoch": 0.3, + "learning_rate": 1.578e-05, + "loss": 4.307, "step": 266 }, { - "epoch": 0.15, - "learning_rate": 1.578e-05, - "loss": 4.3114, + "epoch": 0.3, + "learning_rate": 1.584e-05, + "loss": 4.1291, "step": 267 }, { - "epoch": 0.15, - "learning_rate": 1.584e-05, - "loss": 4.221, + "epoch": 0.3, + "learning_rate": 1.59e-05, + "loss": 4.4503, "step": 268 }, { - "epoch": 0.15, - "learning_rate": 1.59e-05, - "loss": 4.421, + "epoch": 0.3, + "learning_rate": 1.596e-05, + "loss": 4.1821, "step": 269 }, { - "epoch": 0.15, - "learning_rate": 1.596e-05, - "loss": 4.0347, + "epoch": 0.3, + "learning_rate": 1.6020000000000002e-05, + "loss": 4.3484, "step": 270 }, { - "epoch": 0.15, - "learning_rate": 1.6020000000000002e-05, - "loss": 4.3152, + "epoch": 0.3, + "learning_rate": 1.6080000000000002e-05, + "loss": 3.9466, "step": 271 }, { - "epoch": 0.15, - "learning_rate": 1.6080000000000002e-05, - "loss": 4.1378, + "epoch": 0.3, + "learning_rate": 1.614e-05, + "loss": 4.0963, "step": 272 }, { - "epoch": 0.15, - "learning_rate": 1.614e-05, - "loss": 4.0427, + "epoch": 0.31, + "learning_rate": 1.62e-05, + "loss": 4.1321, "step": 273 }, { - "epoch": 0.15, - "learning_rate": 1.62e-05, - "loss": 4.3118, + "epoch": 0.31, + "learning_rate": 1.626e-05, + "loss": 4.0624, "step": 274 }, { - "epoch": 0.15, - "learning_rate": 1.626e-05, - "loss": 4.2973, + "epoch": 0.31, + "learning_rate": 1.6320000000000003e-05, + "loss": 4.2895, "step": 275 }, { - "epoch": 0.15, - "learning_rate": 1.6320000000000003e-05, - "loss": 4.2771, + "epoch": 0.31, + "learning_rate": 1.6380000000000002e-05, + "loss": 4.225, "step": 276 }, { - "epoch": 0.16, - "learning_rate": 1.6380000000000002e-05, - "loss": 4.2649, + "epoch": 0.31, + "learning_rate": 1.6440000000000002e-05, + "loss": 4.1543, "step": 277 }, { - "epoch": 0.16, - "learning_rate": 1.6440000000000002e-05, - "loss": 4.3386, - "step": 278 + "epoch": 0.31, + "learning_rate": 1.65e-05, + "loss": 4.1254, + "step": 278 }, { - "epoch": 0.16, - "learning_rate": 1.65e-05, - "loss": 4.1876, + "epoch": 0.31, + "learning_rate": 1.656e-05, + "loss": 4.1338, "step": 279 }, { - "epoch": 0.16, - "learning_rate": 1.656e-05, - "loss": 4.201, + "epoch": 0.31, + "learning_rate": 1.6620000000000004e-05, + "loss": 4.2958, "step": 280 }, { - "epoch": 0.16, - "learning_rate": 1.6620000000000004e-05, - "loss": 4.3233, + "epoch": 0.32, + "learning_rate": 1.6680000000000003e-05, + "loss": 4.0743, "step": 281 }, { - "epoch": 0.16, - "learning_rate": 1.6680000000000003e-05, - "loss": 4.3703, + "epoch": 0.32, + "learning_rate": 1.6740000000000002e-05, + "loss": 4.217, "step": 282 }, { - "epoch": 0.16, - "learning_rate": 1.6740000000000002e-05, - "loss": 4.3515, + "epoch": 0.32, + "learning_rate": 1.6800000000000002e-05, + "loss": 3.9742, "step": 283 }, { - "epoch": 0.16, - "learning_rate": 1.6800000000000002e-05, - "loss": 4.1063, + "epoch": 0.32, + "learning_rate": 1.686e-05, + "loss": 4.1994, "step": 284 }, { - "epoch": 0.16, - "learning_rate": 1.686e-05, - "loss": 4.2593, + "epoch": 0.32, + "learning_rate": 1.6919999999999997e-05, + "loss": 4.2729, "step": 285 }, { - "epoch": 0.16, - "learning_rate": 1.6919999999999997e-05, - "loss": 4.0121, + "epoch": 0.32, + "learning_rate": 1.698e-05, + "loss": 4.1934, "step": 286 }, { - "epoch": 0.16, - "learning_rate": 1.698e-05, - "loss": 4.2617, + "epoch": 0.32, + "learning_rate": 1.704e-05, + "loss": 4.3364, "step": 287 }, { - "epoch": 0.16, - "learning_rate": 1.704e-05, - "loss": 4.4772, + "epoch": 0.32, + "learning_rate": 1.71e-05, + "loss": 4.2937, "step": 288 }, { - "epoch": 0.16, - "learning_rate": 1.71e-05, - "loss": 4.2036, + "epoch": 0.32, + "learning_rate": 1.716e-05, + "loss": 4.3273, "step": 289 }, { - "epoch": 0.16, - "learning_rate": 1.716e-05, - "loss": 4.291, + "epoch": 0.33, + "learning_rate": 1.7219999999999998e-05, + "loss": 4.302, "step": 290 }, { - "epoch": 0.16, - "learning_rate": 1.7219999999999998e-05, - "loss": 4.2281, + "epoch": 0.33, + "learning_rate": 1.728e-05, + "loss": 4.3626, "step": 291 }, { - "epoch": 0.16, - "learning_rate": 1.728e-05, - "loss": 4.6205, + "epoch": 0.33, + "learning_rate": 1.734e-05, + "loss": 4.2106, "step": 292 }, { - "epoch": 0.16, - "learning_rate": 1.734e-05, - "loss": 4.3975, + "epoch": 0.33, + "learning_rate": 1.74e-05, + "loss": 4.306, "step": 293 }, { - "epoch": 0.16, - "learning_rate": 1.74e-05, - "loss": 4.4876, + "epoch": 0.33, + "learning_rate": 1.746e-05, + "loss": 4.2628, "step": 294 }, { - "epoch": 0.17, - "learning_rate": 1.746e-05, - "loss": 4.3942, + "epoch": 0.33, + "learning_rate": 1.7519999999999998e-05, + "loss": 4.1744, "step": 295 }, { - "epoch": 0.17, - "learning_rate": 1.7519999999999998e-05, - "loss": 4.3009, + "epoch": 0.33, + "learning_rate": 1.758e-05, + "loss": 4.4221, "step": 296 }, { - "epoch": 0.17, - "learning_rate": 1.758e-05, - "loss": 4.2866, + "epoch": 0.33, + "learning_rate": 1.764e-05, + "loss": 4.5329, "step": 297 }, { - "epoch": 0.17, - "learning_rate": 1.764e-05, - "loss": 4.4495, + "epoch": 0.33, + "learning_rate": 1.77e-05, + "loss": 4.3351, "step": 298 }, { - "epoch": 0.17, - "learning_rate": 1.77e-05, - "loss": 4.3001, + "epoch": 0.34, + "learning_rate": 1.776e-05, + "loss": 4.738, "step": 299 }, { - "epoch": 0.17, - "learning_rate": 1.776e-05, - "loss": 4.5097, + "epoch": 0.34, + "learning_rate": 1.782e-05, + "loss": 4.7495, "step": 300 }, { - "epoch": 0.17, - "learning_rate": 1.782e-05, - "loss": 4.2911, + "epoch": 0.34, + "learning_rate": 1.7879999999999998e-05, + "loss": 4.0522, "step": 301 }, { - "epoch": 0.17, - "learning_rate": 1.7879999999999998e-05, - "loss": 4.1269, + "epoch": 0.34, + "learning_rate": 1.794e-05, + "loss": 4.1765, "step": 302 }, { - "epoch": 0.17, - "learning_rate": 1.794e-05, - "loss": 4.3113, + "epoch": 0.34, + "learning_rate": 1.8e-05, + "loss": 4.0936, "step": 303 }, { - "epoch": 0.17, - "learning_rate": 1.8e-05, - "loss": 4.1721, + "epoch": 0.34, + "learning_rate": 1.806e-05, + "loss": 4.2213, "step": 304 }, { - "epoch": 0.17, - "learning_rate": 1.806e-05, - "loss": 4.3154, + "epoch": 0.34, + "learning_rate": 1.812e-05, + "loss": 4.0573, "step": 305 }, { - "epoch": 0.17, - "learning_rate": 1.812e-05, - "loss": 4.1948, + "epoch": 0.34, + "learning_rate": 1.818e-05, + "loss": 4.122, "step": 306 }, { - "epoch": 0.17, - "learning_rate": 1.818e-05, - "loss": 4.3935, + "epoch": 0.34, + "learning_rate": 1.824e-05, + "loss": 4.1466, "step": 307 }, { - "epoch": 0.17, - "learning_rate": 1.824e-05, - "loss": 4.4485, + "epoch": 0.35, + "learning_rate": 1.83e-05, + "loss": 4.2568, "step": 308 }, { - "epoch": 0.17, - "learning_rate": 1.83e-05, - "loss": 4.4367, + "epoch": 0.35, + "learning_rate": 1.836e-05, + "loss": 4.1553, "step": 309 }, { - "epoch": 0.17, - "learning_rate": 1.836e-05, - "loss": 4.3007, + "epoch": 0.35, + "learning_rate": 1.842e-05, + "loss": 4.2805, "step": 310 }, { - "epoch": 0.17, - "learning_rate": 1.842e-05, - "loss": 4.301, + "epoch": 0.35, + "learning_rate": 1.848e-05, + "loss": 4.3665, "step": 311 }, { - "epoch": 0.17, - "learning_rate": 1.848e-05, - "loss": 4.1787, + "epoch": 0.35, + "learning_rate": 1.854e-05, + "loss": 4.2954, "step": 312 }, { - "epoch": 0.18, - "learning_rate": 1.854e-05, - "loss": 4.2407, + "epoch": 0.35, + "learning_rate": 1.86e-05, + "loss": 4.1888, "step": 313 }, { - "epoch": 0.18, - "learning_rate": 1.86e-05, - "loss": 4.3646, + "epoch": 0.35, + "learning_rate": 1.866e-05, + "loss": 4.2317, "step": 314 }, { - "epoch": 0.18, - "learning_rate": 1.866e-05, - "loss": 4.3755, + "epoch": 0.35, + "learning_rate": 1.872e-05, + "loss": 4.0812, "step": 315 }, { - "epoch": 0.18, - "learning_rate": 1.872e-05, - "loss": 4.1599, + "epoch": 0.35, + "learning_rate": 1.878e-05, + "loss": 4.2498, "step": 316 }, { - "epoch": 0.18, - "learning_rate": 1.878e-05, - "loss": 4.2004, + "epoch": 0.36, + "learning_rate": 1.884e-05, + "loss": 4.0964, "step": 317 }, { - "epoch": 0.18, - "learning_rate": 1.884e-05, - "loss": 3.9382, + "epoch": 0.36, + "learning_rate": 1.8900000000000002e-05, + "loss": 4.1132, "step": 318 }, { - "epoch": 0.18, - "learning_rate": 1.8900000000000002e-05, - "loss": 4.1481, + "epoch": 0.36, + "learning_rate": 1.896e-05, + "loss": 4.2153, "step": 319 }, { - "epoch": 0.18, - "learning_rate": 1.896e-05, - "loss": 4.2926, + "epoch": 0.36, + "learning_rate": 1.902e-05, + "loss": 4.1501, "step": 320 }, { - "epoch": 0.18, - "learning_rate": 1.902e-05, - "loss": 4.4841, + "epoch": 0.36, + "learning_rate": 1.908e-05, + "loss": 4.2813, "step": 321 }, { - "epoch": 0.18, - "learning_rate": 1.908e-05, - "loss": 4.3148, + "epoch": 0.36, + "learning_rate": 1.914e-05, + "loss": 4.1588, "step": 322 }, { - "epoch": 0.18, - "learning_rate": 1.914e-05, - "loss": 4.3228, + "epoch": 0.36, + "learning_rate": 1.9200000000000003e-05, + "loss": 4.0803, "step": 323 }, { - "epoch": 0.18, - "learning_rate": 1.9200000000000003e-05, - "loss": 4.2022, + "epoch": 0.36, + "learning_rate": 1.9260000000000002e-05, + "loss": 4.2773, "step": 324 }, { - "epoch": 0.18, - "learning_rate": 1.9260000000000002e-05, - "loss": 4.1847, + "epoch": 0.36, + "learning_rate": 1.932e-05, + "loss": 4.2866, "step": 325 }, { - "epoch": 0.18, - "learning_rate": 1.932e-05, - "loss": 4.1588, + "epoch": 0.37, + "learning_rate": 1.938e-05, + "loss": 4.2622, "step": 326 }, { - "epoch": 0.18, - "learning_rate": 1.938e-05, - "loss": 4.0783, + "epoch": 0.37, + "learning_rate": 1.944e-05, + "loss": 4.1441, "step": 327 }, { - "epoch": 0.18, - "learning_rate": 1.944e-05, - "loss": 4.2046, + "epoch": 0.37, + "learning_rate": 1.95e-05, + "loss": 4.1331, "step": 328 }, { - "epoch": 0.18, - "learning_rate": 1.95e-05, - "loss": 4.1779, + "epoch": 0.37, + "learning_rate": 1.9560000000000002e-05, + "loss": 4.3182, "step": 329 }, { - "epoch": 0.18, - "learning_rate": 1.9560000000000002e-05, - "loss": 4.4946, + "epoch": 0.37, + "learning_rate": 1.9620000000000002e-05, + "loss": 4.3532, "step": 330 }, { - "epoch": 0.19, - "learning_rate": 1.9620000000000002e-05, - "loss": 4.2943, + "epoch": 0.37, + "learning_rate": 1.968e-05, + "loss": 4.1673, "step": 331 }, { - "epoch": 0.19, - "learning_rate": 1.968e-05, - "loss": 4.3843, + "epoch": 0.37, + "learning_rate": 1.974e-05, + "loss": 4.2077, "step": 332 }, { - "epoch": 0.19, - "learning_rate": 1.974e-05, - "loss": 4.3714, + "epoch": 0.37, + "learning_rate": 1.98e-05, + "loss": 4.2068, "step": 333 }, { - "epoch": 0.19, - "learning_rate": 1.98e-05, - "loss": 4.3556, + "epoch": 0.37, + "learning_rate": 1.9860000000000003e-05, + "loss": 4.2471, "step": 334 }, { - "epoch": 0.19, - "learning_rate": 1.9860000000000003e-05, - "loss": 4.3798, + "epoch": 0.38, + "learning_rate": 1.9920000000000002e-05, + "loss": 4.3877, "step": 335 }, { - "epoch": 0.19, - "learning_rate": 1.9920000000000002e-05, - "loss": 4.152, + "epoch": 0.38, + "learning_rate": 1.9980000000000002e-05, + "loss": 4.1347, "step": 336 }, { - "epoch": 0.19, - "learning_rate": 1.9980000000000002e-05, - "loss": 4.2607, + "epoch": 0.38, + "learning_rate": 2.004e-05, + "loss": 4.0476, "step": 337 }, { - "epoch": 0.19, - "learning_rate": 2.004e-05, - "loss": 4.1062, + "epoch": 0.38, + "learning_rate": 2.01e-05, + "loss": 4.3133, "step": 338 }, { - "epoch": 0.19, - "learning_rate": 2.01e-05, - "loss": 4.2196, + "epoch": 0.38, + "learning_rate": 2.016e-05, + "loss": 4.1127, "step": 339 }, { - "epoch": 0.19, - "learning_rate": 2.016e-05, - "loss": 4.2091, + "epoch": 0.38, + "learning_rate": 2.0220000000000003e-05, + "loss": 4.2478, "step": 340 }, { - "epoch": 0.19, - "learning_rate": 2.0220000000000003e-05, - "loss": 4.1571, + "epoch": 0.38, + "learning_rate": 2.0280000000000002e-05, + "loss": 4.1574, "step": 341 }, { - "epoch": 0.19, - "learning_rate": 2.0280000000000002e-05, - "loss": 4.356, + "epoch": 0.38, + "learning_rate": 2.0340000000000002e-05, + "loss": 4.2871, "step": 342 }, { - "epoch": 0.19, - "learning_rate": 2.0340000000000002e-05, - "loss": 4.4906, + "epoch": 0.38, + "learning_rate": 2.04e-05, + "loss": 4.2293, "step": 343 }, { - "epoch": 0.19, - "learning_rate": 2.04e-05, - "loss": 4.0563, + "epoch": 0.39, + "learning_rate": 2.046e-05, + "loss": 4.3118, "step": 344 }, { - "epoch": 0.19, - "learning_rate": 2.046e-05, - "loss": 4.3361, + "epoch": 0.39, + "learning_rate": 2.0520000000000003e-05, + "loss": 4.3705, "step": 345 }, { - "epoch": 0.19, - "learning_rate": 2.0520000000000003e-05, - "loss": 4.1754, + "epoch": 0.39, + "learning_rate": 2.0580000000000003e-05, + "loss": 4.5409, "step": 346 }, { - "epoch": 0.19, - "learning_rate": 2.0580000000000003e-05, - "loss": 4.6586, + "epoch": 0.39, + "learning_rate": 2.064e-05, + "loss": 4.3818, "step": 347 }, { - "epoch": 0.2, - "learning_rate": 2.064e-05, - "loss": 4.605, + "epoch": 0.39, + "learning_rate": 2.07e-05, + "loss": 4.4166, "step": 348 }, { - "epoch": 0.2, - "learning_rate": 2.07e-05, - "loss": 4.2504, + "epoch": 0.39, + "learning_rate": 2.0759999999999998e-05, + "loss": 4.5959, "step": 349 }, { - "epoch": 0.2, - "learning_rate": 2.0759999999999998e-05, - "loss": 4.2302, + "epoch": 0.39, + "learning_rate": 2.082e-05, + "loss": 4.4276, "step": 350 }, { - "epoch": 0.2, - "learning_rate": 2.082e-05, - "loss": 4.3514, + "epoch": 0.39, + "learning_rate": 2.088e-05, + "loss": 4.1841, "step": 351 }, { - "epoch": 0.2, - "learning_rate": 2.088e-05, - "loss": 4.418, + "epoch": 0.39, + "learning_rate": 2.094e-05, + "loss": 4.2181, "step": 352 }, { - "epoch": 0.2, - "learning_rate": 2.094e-05, - "loss": 4.0628, + "epoch": 0.4, + "learning_rate": 2.1e-05, + "loss": 4.1374, "step": 353 }, { - "epoch": 0.2, - "learning_rate": 2.1e-05, - "loss": 4.1922, + "epoch": 0.4, + "learning_rate": 2.1059999999999998e-05, + "loss": 4.1394, "step": 354 }, { - "epoch": 0.2, - "learning_rate": 2.1059999999999998e-05, - "loss": 3.8978, + "epoch": 0.4, + "learning_rate": 2.1119999999999998e-05, + "loss": 4.2856, "step": 355 }, { - "epoch": 0.2, - "learning_rate": 2.1119999999999998e-05, - "loss": 4.1357, + "epoch": 0.4, + "learning_rate": 2.118e-05, + "loss": 4.0304, "step": 356 }, { - "epoch": 0.2, - "learning_rate": 2.118e-05, - "loss": 4.3353, + "epoch": 0.4, + "learning_rate": 2.124e-05, + "loss": 4.0897, "step": 357 }, { - "epoch": 0.2, - "learning_rate": 2.124e-05, - "loss": 4.3339, + "epoch": 0.4, + "learning_rate": 2.13e-05, + "loss": 4.1678, "step": 358 }, { - "epoch": 0.2, - "learning_rate": 2.13e-05, - "loss": 4.4832, + "epoch": 0.4, + "learning_rate": 2.136e-05, + "loss": 4.1266, "step": 359 }, { - "epoch": 0.2, - "learning_rate": 2.136e-05, - "loss": 4.1927, + "epoch": 0.4, + "learning_rate": 2.1419999999999998e-05, + "loss": 4.2845, "step": 360 }, { - "epoch": 0.2, - "learning_rate": 2.1419999999999998e-05, - "loss": 4.2226, + "epoch": 0.4, + "learning_rate": 2.148e-05, + "loss": 4.1321, "step": 361 }, { - "epoch": 0.2, - "learning_rate": 2.148e-05, - "loss": 4.0212, + "epoch": 0.41, + "learning_rate": 2.154e-05, + "loss": 4.1353, "step": 362 }, { - "epoch": 0.2, - "learning_rate": 2.154e-05, - "loss": 4.2884, + "epoch": 0.41, + "learning_rate": 2.16e-05, + "loss": 4.0673, "step": 363 }, { - "epoch": 0.2, - "learning_rate": 2.16e-05, - "loss": 4.2995, + "epoch": 0.41, + "learning_rate": 2.166e-05, + "loss": 4.2948, "step": 364 }, { - "epoch": 0.2, - "learning_rate": 2.166e-05, - "loss": 4.2459, + "epoch": 0.41, + "learning_rate": 2.172e-05, + "loss": 4.0433, "step": 365 }, { - "epoch": 0.21, - "learning_rate": 2.172e-05, - "loss": 4.3832, + "epoch": 0.41, + "learning_rate": 2.178e-05, + "loss": 4.1878, "step": 366 }, { - "epoch": 0.21, - "learning_rate": 2.178e-05, - "loss": 4.224, + "epoch": 0.41, + "learning_rate": 2.184e-05, + "loss": 4.0997, "step": 367 }, { - "epoch": 0.21, - "learning_rate": 2.184e-05, - "loss": 4.1484, + "epoch": 0.41, + "learning_rate": 2.19e-05, + "loss": 4.2037, "step": 368 }, { - "epoch": 0.21, - "learning_rate": 2.19e-05, - "loss": 4.0405, + "epoch": 0.41, + "learning_rate": 2.196e-05, + "loss": 4.2507, "step": 369 }, { - "epoch": 0.21, - "learning_rate": 2.196e-05, - "loss": 4.1106, + "epoch": 0.41, + "learning_rate": 2.202e-05, + "loss": 4.2778, "step": 370 }, { - "epoch": 0.21, - "learning_rate": 2.202e-05, - "loss": 4.1984, + "epoch": 0.42, + "learning_rate": 2.208e-05, + "loss": 4.0861, "step": 371 }, { - "epoch": 0.21, - "learning_rate": 2.208e-05, - "loss": 4.6375, + "epoch": 0.42, + "learning_rate": 2.214e-05, + "loss": 4.3873, "step": 372 }, { - "epoch": 0.21, - "learning_rate": 2.214e-05, - "loss": 4.2163, + "epoch": 0.42, + "learning_rate": 2.22e-05, + "loss": 4.1633, "step": 373 }, { - "epoch": 0.21, - "learning_rate": 2.22e-05, - "loss": 4.0386, + "epoch": 0.42, + "learning_rate": 2.226e-05, + "loss": 4.2557, "step": 374 }, { - "epoch": 0.21, - "learning_rate": 2.226e-05, - "loss": 4.0182, + "epoch": 0.42, + "learning_rate": 2.232e-05, + "loss": 4.105, "step": 375 }, { - "epoch": 0.21, - "learning_rate": 2.232e-05, - "loss": 4.0926, + "epoch": 0.42, + "learning_rate": 2.238e-05, + "loss": 4.1985, "step": 376 }, { - "epoch": 0.21, - "learning_rate": 2.238e-05, - "loss": 4.1242, + "epoch": 0.42, + "learning_rate": 2.2440000000000002e-05, + "loss": 3.9895, "step": 377 }, { - "epoch": 0.21, - "learning_rate": 2.2440000000000002e-05, - "loss": 3.9954, + "epoch": 0.42, + "learning_rate": 2.25e-05, + "loss": 4.3159, "step": 378 }, { - "epoch": 0.21, - "learning_rate": 2.25e-05, - "loss": 4.1984, + "epoch": 0.42, + "learning_rate": 2.256e-05, + "loss": 4.2099, "step": 379 }, { - "epoch": 0.21, - "learning_rate": 2.256e-05, - "loss": 4.3595, + "epoch": 0.43, + "learning_rate": 2.262e-05, + "loss": 4.3025, "step": 380 }, { - "epoch": 0.21, - "learning_rate": 2.262e-05, - "loss": 3.9519, + "epoch": 0.43, + "learning_rate": 2.268e-05, + "loss": 4.2785, "step": 381 }, { - "epoch": 0.21, - "learning_rate": 2.268e-05, - "loss": 4.1197, + "epoch": 0.43, + "learning_rate": 2.274e-05, + "loss": 4.1978, "step": 382 }, { - "epoch": 0.21, - "learning_rate": 2.274e-05, - "loss": 4.2409, + "epoch": 0.43, + "learning_rate": 2.2800000000000002e-05, + "loss": 4.2591, "step": 383 }, { - "epoch": 0.22, - "learning_rate": 2.2800000000000002e-05, - "loss": 4.1282, + "epoch": 0.43, + "learning_rate": 2.286e-05, + "loss": 4.2084, "step": 384 }, { - "epoch": 0.22, - "learning_rate": 2.286e-05, - "loss": 4.3583, + "epoch": 0.43, + "learning_rate": 2.292e-05, + "loss": 4.1318, "step": 385 }, { - "epoch": 0.22, - "learning_rate": 2.292e-05, - "loss": 4.318, + "epoch": 0.43, + "learning_rate": 2.298e-05, + "loss": 4.1975, "step": 386 }, { - "epoch": 0.22, - "learning_rate": 2.298e-05, - "loss": 4.3235, + "epoch": 0.43, + "learning_rate": 2.304e-05, + "loss": 4.1145, "step": 387 }, { - "epoch": 0.22, - "learning_rate": 2.304e-05, - "loss": 4.0547, + "epoch": 0.43, + "learning_rate": 2.3100000000000002e-05, + "loss": 4.007, "step": 388 }, { - "epoch": 0.22, - "learning_rate": 2.3100000000000002e-05, - "loss": 4.3362, + "epoch": 0.44, + "learning_rate": 2.3160000000000002e-05, + "loss": 4.153, "step": 389 }, { - "epoch": 0.22, - "learning_rate": 2.3160000000000002e-05, - "loss": 4.5647, + "epoch": 0.44, + "learning_rate": 2.322e-05, + "loss": 4.2659, "step": 390 }, { - "epoch": 0.22, - "learning_rate": 2.322e-05, - "loss": 4.1328, + "epoch": 0.44, + "learning_rate": 2.328e-05, + "loss": 4.3676, "step": 391 }, { - "epoch": 0.22, - "learning_rate": 2.328e-05, - "loss": 4.5629, + "epoch": 0.44, + "learning_rate": 2.334e-05, + "loss": 4.0016, "step": 392 }, { - "epoch": 0.22, - "learning_rate": 2.334e-05, - "loss": 4.5756, + "epoch": 0.44, + "learning_rate": 2.3400000000000003e-05, + "loss": 4.1861, "step": 393 }, { - "epoch": 0.22, - "learning_rate": 2.3400000000000003e-05, - "loss": 4.2939, + "epoch": 0.44, + "learning_rate": 2.3460000000000002e-05, + "loss": 4.325, "step": 394 }, { - "epoch": 0.22, - "learning_rate": 2.3460000000000002e-05, - "loss": 4.4157, + "epoch": 0.44, + "learning_rate": 2.3520000000000002e-05, + "loss": 4.5118, "step": 395 }, { - "epoch": 0.22, - "learning_rate": 2.3520000000000002e-05, - "loss": 4.3343, + "epoch": 0.44, + "learning_rate": 2.358e-05, + "loss": 4.2486, "step": 396 }, { - "epoch": 0.22, - "learning_rate": 2.358e-05, - "loss": 4.3268, + "epoch": 0.45, + "learning_rate": 2.364e-05, + "loss": 4.2231, "step": 397 }, { - "epoch": 0.22, - "learning_rate": 2.364e-05, - "loss": 4.7341, + "epoch": 0.45, + "learning_rate": 2.37e-05, + "loss": 4.6555, "step": 398 }, { - "epoch": 0.22, - "learning_rate": 2.37e-05, - "loss": 5.2309, + "epoch": 0.45, + "learning_rate": 2.3760000000000003e-05, + "loss": 4.428, "step": 399 }, { - "epoch": 0.22, - "learning_rate": 2.3760000000000003e-05, - "loss": 4.8813, + "epoch": 0.45, + "learning_rate": 2.3820000000000002e-05, + "loss": 4.5045, "step": 400 }, { - "epoch": 0.22, - "learning_rate": 2.3820000000000002e-05, - "loss": 4.3757, + "epoch": 0.45, + "learning_rate": 2.3880000000000002e-05, + "loss": 4.2833, "step": 401 }, { - "epoch": 0.23, - "learning_rate": 2.3880000000000002e-05, - "loss": 4.0931, + "epoch": 0.45, + "learning_rate": 2.394e-05, + "loss": 4.1244, "step": 402 }, { - "epoch": 0.23, - "learning_rate": 2.394e-05, - "loss": 4.1194, + "epoch": 0.45, + "learning_rate": 2.4e-05, + "loss": 4.235, "step": 403 }, { - "epoch": 0.23, - "learning_rate": 2.4e-05, - "loss": 4.3194, + "epoch": 0.45, + "learning_rate": 2.4060000000000003e-05, + "loss": 4.17, "step": 404 }, { - "epoch": 0.23, - "learning_rate": 2.4060000000000003e-05, - "loss": 4.0206, + "epoch": 0.45, + "learning_rate": 2.4120000000000003e-05, + "loss": 4.2928, "step": 405 }, { - "epoch": 0.23, - "learning_rate": 2.4120000000000003e-05, - "loss": 4.22, + "epoch": 0.46, + "learning_rate": 2.4180000000000002e-05, + "loss": 4.1763, "step": 406 }, { - "epoch": 0.23, - "learning_rate": 2.4180000000000002e-05, - "loss": 4.1186, + "epoch": 0.46, + "learning_rate": 2.4240000000000002e-05, + "loss": 4.1616, "step": 407 }, { - "epoch": 0.23, - "learning_rate": 2.4240000000000002e-05, - "loss": 4.24, + "epoch": 0.46, + "learning_rate": 2.43e-05, + "loss": 4.0906, "step": 408 }, { - "epoch": 0.23, - "learning_rate": 2.43e-05, - "loss": 4.3466, + "epoch": 0.46, + "learning_rate": 2.4360000000000004e-05, + "loss": 4.3761, "step": 409 }, { - "epoch": 0.23, - "learning_rate": 2.4360000000000004e-05, - "loss": 4.2623, + "epoch": 0.46, + "learning_rate": 2.442e-05, + "loss": 4.1907, "step": 410 }, { - "epoch": 0.23, - "learning_rate": 2.442e-05, - "loss": 4.2144, + "epoch": 0.46, + "learning_rate": 2.448e-05, + "loss": 4.0623, "step": 411 }, { - "epoch": 0.23, - "learning_rate": 2.448e-05, - "loss": 4.2472, + "epoch": 0.46, + "learning_rate": 2.454e-05, + "loss": 4.1642, "step": 412 }, { - "epoch": 0.23, - "learning_rate": 2.454e-05, - "loss": 4.268, + "epoch": 0.46, + "learning_rate": 2.4599999999999998e-05, + "loss": 4.2181, "step": 413 }, { - "epoch": 0.23, - "learning_rate": 2.4599999999999998e-05, - "loss": 4.2911, + "epoch": 0.46, + "learning_rate": 2.4659999999999998e-05, + "loss": 4.0526, "step": 414 }, { - "epoch": 0.23, - "learning_rate": 2.4659999999999998e-05, - "loss": 4.2753, + "epoch": 0.47, + "learning_rate": 2.472e-05, + "loss": 4.0498, "step": 415 }, { - "epoch": 0.23, - "learning_rate": 2.472e-05, - "loss": 4.1832, + "epoch": 0.47, + "learning_rate": 2.478e-05, + "loss": 4.2762, "step": 416 }, { - "epoch": 0.23, - "learning_rate": 2.478e-05, - "loss": 4.2735, + "epoch": 0.47, + "learning_rate": 2.484e-05, + "loss": 4.0794, "step": 417 }, { - "epoch": 0.23, - "learning_rate": 2.484e-05, - "loss": 4.2643, + "epoch": 0.47, + "learning_rate": 2.49e-05, + "loss": 4.2228, "step": 418 }, { - "epoch": 0.23, - "learning_rate": 2.49e-05, - "loss": 4.289, + "epoch": 0.47, + "learning_rate": 2.4959999999999998e-05, + "loss": 4.1585, "step": 419 }, { - "epoch": 0.24, - "learning_rate": 2.4959999999999998e-05, - "loss": 4.3005, + "epoch": 0.47, + "learning_rate": 2.502e-05, + "loss": 4.1818, "step": 420 }, { - "epoch": 0.24, - "learning_rate": 2.502e-05, - "loss": 4.1223, + "epoch": 0.47, + "learning_rate": 2.508e-05, + "loss": 4.2265, "step": 421 }, { - "epoch": 0.24, - "learning_rate": 2.508e-05, - "loss": 4.429, + "epoch": 0.47, + "learning_rate": 2.514e-05, + "loss": 4.2701, "step": 422 }, { - "epoch": 0.24, - "learning_rate": 2.514e-05, - "loss": 4.0601, + "epoch": 0.47, + "learning_rate": 2.52e-05, + "loss": 4.2883, "step": 423 }, { - "epoch": 0.24, - "learning_rate": 2.52e-05, - "loss": 4.5131, + "epoch": 0.48, + "learning_rate": 2.526e-05, + "loss": 4.1966, "step": 424 }, { - "epoch": 0.24, - "learning_rate": 2.526e-05, - "loss": 4.4075, + "epoch": 0.48, + "learning_rate": 2.5319999999999998e-05, + "loss": 4.2658, "step": 425 }, { - "epoch": 0.24, - "learning_rate": 2.5319999999999998e-05, - "loss": 4.2671, + "epoch": 0.48, + "learning_rate": 2.538e-05, + "loss": 4.286, "step": 426 }, { - "epoch": 0.24, - "learning_rate": 2.538e-05, - "loss": 4.1313, + "epoch": 0.48, + "learning_rate": 2.544e-05, + "loss": 4.1286, "step": 427 }, { - "epoch": 0.24, - "learning_rate": 2.544e-05, - "loss": 4.1743, + "epoch": 0.48, + "learning_rate": 2.55e-05, + "loss": 4.3255, "step": 428 }, { - "epoch": 0.24, - "learning_rate": 2.55e-05, - "loss": 4.1737, + "epoch": 0.48, + "learning_rate": 2.556e-05, + "loss": 4.1596, "step": 429 }, { - "epoch": 0.24, - "learning_rate": 2.556e-05, - "loss": 4.2877, + "epoch": 0.48, + "learning_rate": 2.562e-05, + "loss": 4.3055, "step": 430 }, { - "epoch": 0.24, - "learning_rate": 2.562e-05, - "loss": 4.1968, + "epoch": 0.48, + "learning_rate": 2.568e-05, + "loss": 4.3901, "step": 431 }, { - "epoch": 0.24, - "learning_rate": 2.568e-05, - "loss": 4.2805, + "epoch": 0.48, + "learning_rate": 2.574e-05, + "loss": 4.2223, "step": 432 }, { - "epoch": 0.24, - "learning_rate": 2.574e-05, - "loss": 4.2455, + "epoch": 0.49, + "learning_rate": 2.58e-05, + "loss": 3.9652, "step": 433 }, { - "epoch": 0.24, - "learning_rate": 2.58e-05, - "loss": 4.3769, + "epoch": 0.49, + "learning_rate": 2.586e-05, + "loss": 4.1636, "step": 434 }, { - "epoch": 0.24, - "learning_rate": 2.586e-05, - "loss": 4.3591, + "epoch": 0.49, + "learning_rate": 2.592e-05, + "loss": 4.1553, "step": 435 }, { - "epoch": 0.24, - "learning_rate": 2.592e-05, - "loss": 3.9896, + "epoch": 0.49, + "learning_rate": 2.5980000000000002e-05, + "loss": 4.3838, "step": 436 }, { - "epoch": 0.24, - "learning_rate": 2.5980000000000002e-05, - "loss": 4.1979, + "epoch": 0.49, + "learning_rate": 2.604e-05, + "loss": 4.3877, "step": 437 }, { - "epoch": 0.25, - "learning_rate": 2.604e-05, - "loss": 4.4241, + "epoch": 0.49, + "learning_rate": 2.61e-05, + "loss": 4.2501, "step": 438 }, { - "epoch": 0.25, - "learning_rate": 2.61e-05, - "loss": 4.1434, + "epoch": 0.49, + "learning_rate": 2.616e-05, + "loss": 4.0833, "step": 439 }, { - "epoch": 0.25, - "learning_rate": 2.616e-05, - "loss": 4.3756, + "epoch": 0.49, + "learning_rate": 2.622e-05, + "loss": 4.1429, "step": 440 }, { - "epoch": 0.25, - "learning_rate": 2.622e-05, - "loss": 4.36, + "epoch": 0.49, + "learning_rate": 2.628e-05, + "loss": 4.2947, "step": 441 }, { - "epoch": 0.25, - "learning_rate": 2.628e-05, - "loss": 4.3536, + "epoch": 0.5, + "learning_rate": 2.6340000000000002e-05, + "loss": 4.1863, "step": 442 }, { - "epoch": 0.25, - "learning_rate": 2.6340000000000002e-05, - "loss": 4.294, + "epoch": 0.5, + "learning_rate": 2.64e-05, + "loss": 4.2328, "step": 443 }, { - "epoch": 0.25, - "learning_rate": 2.64e-05, - "loss": 4.5094, + "epoch": 0.5, + "learning_rate": 2.646e-05, + "loss": 4.4469, "step": 444 }, { - "epoch": 0.25, - "learning_rate": 2.646e-05, - "loss": 4.3045, + "epoch": 0.5, + "learning_rate": 2.652e-05, + "loss": 4.2335, "step": 445 }, { - "epoch": 0.25, - "learning_rate": 2.652e-05, - "loss": 4.9616, + "epoch": 0.5, + "learning_rate": 2.658e-05, + "loss": 4.0421, "step": 446 }, { - "epoch": 0.25, - "learning_rate": 2.658e-05, - "loss": 4.705, + "epoch": 0.5, + "learning_rate": 2.6640000000000002e-05, + "loss": 4.2282, "step": 447 }, { - "epoch": 0.25, - "learning_rate": 2.6640000000000002e-05, - "loss": 5.1824, + "epoch": 0.5, + "learning_rate": 2.6700000000000002e-05, + "loss": 4.4805, "step": 448 }, { - "epoch": 0.25, - "learning_rate": 2.6700000000000002e-05, - "loss": 4.5262, - "step": 449 + "epoch": 0.5, + "learning_rate": 2.676e-05, + "loss": 4.3042, + "step": 449 }, { - "epoch": 0.25, - "learning_rate": 2.676e-05, - "loss": 4.516, + "epoch": 0.5, + "learning_rate": 2.682e-05, + "loss": 4.8199, "step": 450 }, { - "epoch": 0.25, - "learning_rate": 2.682e-05, - "loss": 3.9647, + "epoch": 0.51, + "learning_rate": 2.688e-05, + "loss": 4.2232, "step": 451 }, { - "epoch": 0.25, - "learning_rate": 2.688e-05, - "loss": 4.1498, + "epoch": 0.51, + "learning_rate": 2.6940000000000003e-05, + "loss": 4.2092, "step": 452 }, { - "epoch": 0.25, - "learning_rate": 2.6940000000000003e-05, - "loss": 4.2994, + "epoch": 0.51, + "learning_rate": 2.7000000000000002e-05, + "loss": 4.352, "step": 453 }, { - "epoch": 0.25, - "learning_rate": 2.7000000000000002e-05, - "loss": 3.9839, + "epoch": 0.51, + "learning_rate": 2.7060000000000002e-05, + "loss": 4.0295, "step": 454 }, { - "epoch": 0.26, - "learning_rate": 2.7060000000000002e-05, - "loss": 4.0633, + "epoch": 0.51, + "learning_rate": 2.712e-05, + "loss": 4.3023, "step": 455 }, { - "epoch": 0.26, - "learning_rate": 2.712e-05, - "loss": 4.1811, + "epoch": 0.51, + "learning_rate": 2.718e-05, + "loss": 4.2455, "step": 456 }, { - "epoch": 0.26, - "learning_rate": 2.718e-05, - "loss": 4.0188, + "epoch": 0.51, + "learning_rate": 2.724e-05, + "loss": 4.1829, "step": 457 }, { - "epoch": 0.26, - "learning_rate": 2.724e-05, - "loss": 4.007, + "epoch": 0.51, + "learning_rate": 2.7300000000000003e-05, + "loss": 4.1425, "step": 458 }, { - "epoch": 0.26, - "learning_rate": 2.7300000000000003e-05, - "loss": 3.9451, + "epoch": 0.51, + "learning_rate": 2.7360000000000002e-05, + "loss": 4.243, "step": 459 }, { - "epoch": 0.26, - "learning_rate": 2.7360000000000002e-05, - "loss": 4.1622, + "epoch": 0.52, + "learning_rate": 2.7420000000000002e-05, + "loss": 4.3147, "step": 460 }, { - "epoch": 0.26, - "learning_rate": 2.7420000000000002e-05, - "loss": 4.0804, + "epoch": 0.52, + "learning_rate": 2.748e-05, + "loss": 4.0656, "step": 461 }, { - "epoch": 0.26, - "learning_rate": 2.748e-05, - "loss": 4.0575, + "epoch": 0.52, + "learning_rate": 2.754e-05, + "loss": 4.103, "step": 462 }, { - "epoch": 0.26, - "learning_rate": 2.754e-05, - "loss": 4.1324, + "epoch": 0.52, + "learning_rate": 2.7600000000000003e-05, + "loss": 4.1576, "step": 463 }, { - "epoch": 0.26, - "learning_rate": 2.7600000000000003e-05, - "loss": 4.0888, + "epoch": 0.52, + "learning_rate": 2.7660000000000003e-05, + "loss": 4.0859, "step": 464 }, { - "epoch": 0.26, - "learning_rate": 2.7660000000000003e-05, - "loss": 4.1026, + "epoch": 0.52, + "learning_rate": 2.7720000000000002e-05, + "loss": 4.2193, "step": 465 }, { - "epoch": 0.26, - "learning_rate": 2.7720000000000002e-05, - "loss": 4.3026, + "epoch": 0.52, + "learning_rate": 2.778e-05, + "loss": 4.2194, "step": 466 }, { - "epoch": 0.26, - "learning_rate": 2.778e-05, - "loss": 4.1423, + "epoch": 0.52, + "learning_rate": 2.784e-05, + "loss": 4.1411, "step": 467 }, { - "epoch": 0.26, - "learning_rate": 2.784e-05, - "loss": 4.2504, + "epoch": 0.52, + "learning_rate": 2.79e-05, + "loss": 4.0721, "step": 468 }, { - "epoch": 0.26, - "learning_rate": 2.79e-05, - "loss": 4.065, + "epoch": 0.53, + "learning_rate": 2.7960000000000003e-05, + "loss": 4.1481, "step": 469 }, { - "epoch": 0.26, - "learning_rate": 2.7960000000000003e-05, - "loss": 4.1554, + "epoch": 0.53, + "learning_rate": 2.8020000000000003e-05, + "loss": 4.2244, "step": 470 }, { - "epoch": 0.26, - "learning_rate": 2.8020000000000003e-05, - "loss": 4.34, + "epoch": 0.53, + "learning_rate": 2.8080000000000002e-05, + "loss": 4.0628, "step": 471 }, { - "epoch": 0.26, - "learning_rate": 2.8080000000000002e-05, - "loss": 4.1225, + "epoch": 0.53, + "learning_rate": 2.8139999999999998e-05, + "loss": 4.0549, "step": 472 }, { - "epoch": 0.27, - "learning_rate": 2.8139999999999998e-05, - "loss": 4.1717, + "epoch": 0.53, + "learning_rate": 2.8199999999999998e-05, + "loss": 4.243, "step": 473 }, { - "epoch": 0.27, - "learning_rate": 2.8199999999999998e-05, - "loss": 4.261, + "epoch": 0.53, + "learning_rate": 2.826e-05, + "loss": 4.2145, "step": 474 }, { - "epoch": 0.27, - "learning_rate": 2.826e-05, - "loss": 4.2665, + "epoch": 0.53, + "learning_rate": 2.832e-05, + "loss": 4.1422, "step": 475 }, { - "epoch": 0.27, - "learning_rate": 2.832e-05, - "loss": 4.3772, + "epoch": 0.53, + "learning_rate": 2.838e-05, + "loss": 4.3287, "step": 476 }, { - "epoch": 0.27, - "learning_rate": 2.838e-05, - "loss": 4.3676, + "epoch": 0.53, + "learning_rate": 2.844e-05, + "loss": 4.3965, "step": 477 }, { - "epoch": 0.27, - "learning_rate": 2.844e-05, - "loss": 4.2325, + "epoch": 0.54, + "learning_rate": 2.8499999999999998e-05, + "loss": 4.0726, "step": 478 }, { - "epoch": 0.27, - "learning_rate": 2.8499999999999998e-05, - "loss": 4.3501, + "epoch": 0.54, + "learning_rate": 2.856e-05, + "loss": 3.9443, "step": 479 }, { - "epoch": 0.27, - "learning_rate": 2.856e-05, - "loss": 4.2706, + "epoch": 0.54, + "learning_rate": 2.862e-05, + "loss": 4.264, "step": 480 }, { - "epoch": 0.27, - "learning_rate": 2.862e-05, - "loss": 4.0092, + "epoch": 0.54, + "learning_rate": 2.868e-05, + "loss": 4.0944, "step": 481 }, { - "epoch": 0.27, - "learning_rate": 2.868e-05, - "loss": 4.3138, + "epoch": 0.54, + "learning_rate": 2.874e-05, + "loss": 4.0911, "step": 482 }, { - "epoch": 0.27, - "learning_rate": 2.874e-05, - "loss": 4.4792, + "epoch": 0.54, + "learning_rate": 2.88e-05, + "loss": 4.2576, "step": 483 }, { - "epoch": 0.27, - "learning_rate": 2.88e-05, - "loss": 4.0796, + "epoch": 0.54, + "learning_rate": 2.8859999999999998e-05, + "loss": 4.1463, "step": 484 }, { - "epoch": 0.27, - "learning_rate": 2.8859999999999998e-05, - "loss": 4.4669, + "epoch": 0.54, + "learning_rate": 2.892e-05, + "loss": 4.3166, "step": 485 }, { - "epoch": 0.27, - "learning_rate": 2.892e-05, - "loss": 4.3043, + "epoch": 0.54, + "learning_rate": 2.898e-05, + "loss": 4.0712, "step": 486 }, { - "epoch": 0.27, - "learning_rate": 2.898e-05, - "loss": 4.3602, + "epoch": 0.55, + "learning_rate": 2.904e-05, + "loss": 3.9972, "step": 487 }, { - "epoch": 0.27, - "learning_rate": 2.904e-05, - "loss": 4.4734, + "epoch": 0.55, + "learning_rate": 2.91e-05, + "loss": 4.127, "step": 488 }, { - "epoch": 0.27, - "learning_rate": 2.91e-05, - "loss": 4.4822, + "epoch": 0.55, + "learning_rate": 2.916e-05, + "loss": 4.1673, "step": 489 }, { - "epoch": 0.27, - "learning_rate": 2.916e-05, - "loss": 4.2915, + "epoch": 0.55, + "learning_rate": 2.922e-05, + "loss": 4.303, "step": 490 }, { - "epoch": 0.28, - "learning_rate": 2.922e-05, - "loss": 4.1764, + "epoch": 0.55, + "learning_rate": 2.928e-05, + "loss": 4.0828, "step": 491 }, { - "epoch": 0.28, - "learning_rate": 2.928e-05, - "loss": 4.4906, + "epoch": 0.55, + "learning_rate": 2.934e-05, + "loss": 4.3801, "step": 492 }, { - "epoch": 0.28, - "learning_rate": 2.934e-05, - "loss": 4.3887, + "epoch": 0.55, + "learning_rate": 2.94e-05, + "loss": 4.3874, "step": 493 }, { - "epoch": 0.28, - "learning_rate": 2.94e-05, - "loss": 4.4491, + "epoch": 0.55, + "learning_rate": 2.946e-05, + "loss": 4.1561, "step": 494 }, { - "epoch": 0.28, - "learning_rate": 2.946e-05, - "loss": 4.3901, + "epoch": 0.55, + "learning_rate": 2.9520000000000002e-05, + "loss": 4.2939, "step": 495 }, { - "epoch": 0.28, - "learning_rate": 2.9520000000000002e-05, - "loss": 4.3761, + "epoch": 0.56, + "learning_rate": 2.958e-05, + "loss": 4.0775, "step": 496 }, { - "epoch": 0.28, - "learning_rate": 2.958e-05, - "loss": 4.1813, + "epoch": 0.56, + "learning_rate": 2.964e-05, + "loss": 4.0271, "step": 497 }, { - "epoch": 0.28, - "learning_rate": 2.964e-05, - "loss": 4.6006, + "epoch": 0.56, + "learning_rate": 2.97e-05, + "loss": 4.7244, "step": 498 }, { - "epoch": 0.28, - "learning_rate": 2.97e-05, - "loss": 4.7313, + "epoch": 0.56, + "learning_rate": 2.976e-05, + "loss": 4.6151, "step": 499 }, { - "epoch": 0.28, - "learning_rate": 2.976e-05, - "loss": 4.114, + "epoch": 0.56, + "learning_rate": 2.982e-05, + "loss": 4.3507, "step": 500 }, { - "epoch": 0.28, - "eval_loss": 4.311306953430176, - "eval_runtime": 948.7803, - "eval_samples_per_second": 2.785, - "eval_steps_per_second": 0.349, - "eval_wer": 1.9471836572788577, + "epoch": 0.56, + "eval_loss": 4.249406337738037, + "eval_runtime": 954.2128, + "eval_samples_per_second": 2.769, + "eval_steps_per_second": 0.347, + "eval_wer": 1.931356604522015, "step": 500 }, { - "epoch": 0.28, - "learning_rate": 2.982e-05, - "loss": 4.3698, + "epoch": 0.56, + "learning_rate": 2.9880000000000002e-05, + "loss": 4.4717, "step": 501 }, { - "epoch": 0.28, - "learning_rate": 2.9880000000000002e-05, - "loss": 4.0329, + "epoch": 0.56, + "learning_rate": 2.994e-05, + "loss": 4.1174, "step": 502 }, { - "epoch": 0.28, - "learning_rate": 2.994e-05, - "loss": 4.1158, + "epoch": 0.56, + "learning_rate": 3e-05, + "loss": 4.3457, "step": 503 }, { - "epoch": 0.28, - "learning_rate": 3e-05, - "loss": 4.2736, + "epoch": 0.57, + "learning_rate": 2.9923469387755103e-05, + "loss": 4.1425, "step": 504 }, { - "epoch": 0.28, - "learning_rate": 2.9976635514018694e-05, - "loss": 4.1211, + "epoch": 0.57, + "learning_rate": 2.9846938775510208e-05, + "loss": 4.0916, "step": 505 }, { - "epoch": 0.28, - "learning_rate": 2.9953271028037384e-05, - "loss": 4.3151, + "epoch": 0.57, + "learning_rate": 2.9770408163265306e-05, + "loss": 4.23, "step": 506 }, { - "epoch": 0.28, - "learning_rate": 2.9929906542056074e-05, - "loss": 4.1362, + "epoch": 0.57, + "learning_rate": 2.969387755102041e-05, + "loss": 4.1879, "step": 507 }, { - "epoch": 0.28, - "learning_rate": 2.9906542056074768e-05, - "loss": 4.1223, + "epoch": 0.57, + "learning_rate": 2.961734693877551e-05, + "loss": 4.2693, "step": 508 }, { - "epoch": 0.29, - "learning_rate": 2.988317757009346e-05, - "loss": 4.3377, + "epoch": 0.57, + "learning_rate": 2.9540816326530616e-05, + "loss": 4.1314, "step": 509 }, { - "epoch": 0.29, - "learning_rate": 2.985981308411215e-05, - "loss": 4.4417, + "epoch": 0.57, + "learning_rate": 2.9464285714285714e-05, + "loss": 4.2637, "step": 510 }, { - "epoch": 0.29, - "learning_rate": 2.983644859813084e-05, - "loss": 4.299, + "epoch": 0.57, + "learning_rate": 2.9387755102040816e-05, + "loss": 4.2189, "step": 511 }, { - "epoch": 0.29, - "learning_rate": 2.9813084112149534e-05, - "loss": 4.1538, + "epoch": 0.57, + "learning_rate": 2.9311224489795918e-05, + "loss": 4.2026, "step": 512 }, { - "epoch": 0.29, - "learning_rate": 2.9789719626168224e-05, - "loss": 4.1437, + "epoch": 0.58, + "learning_rate": 2.9234693877551023e-05, + "loss": 4.4774, "step": 513 }, { - "epoch": 0.29, - "learning_rate": 2.9766355140186918e-05, - "loss": 4.0927, + "epoch": 0.58, + "learning_rate": 2.9158163265306125e-05, + "loss": 4.1445, "step": 514 }, { - "epoch": 0.29, - "learning_rate": 2.9742990654205608e-05, - "loss": 4.1283, + "epoch": 0.58, + "learning_rate": 2.9081632653061224e-05, + "loss": 4.1927, "step": 515 }, { - "epoch": 0.29, - "learning_rate": 2.97196261682243e-05, - "loss": 4.3935, + "epoch": 0.58, + "learning_rate": 2.9005102040816326e-05, + "loss": 4.0574, "step": 516 }, { - "epoch": 0.29, - "learning_rate": 2.969626168224299e-05, - "loss": 4.1754, + "epoch": 0.58, + "learning_rate": 2.892857142857143e-05, + "loss": 4.2716, "step": 517 }, { - "epoch": 0.29, - "learning_rate": 2.9672897196261685e-05, - "loss": 4.5719, + "epoch": 0.58, + "learning_rate": 2.8852040816326533e-05, + "loss": 4.1753, "step": 518 }, { - "epoch": 0.29, - "learning_rate": 2.9649532710280375e-05, - "loss": 4.3025, + "epoch": 0.58, + "learning_rate": 2.877551020408163e-05, + "loss": 4.2727, "step": 519 }, { - "epoch": 0.29, - "learning_rate": 2.9626168224299065e-05, - "loss": 4.0066, + "epoch": 0.58, + "learning_rate": 2.8698979591836733e-05, + "loss": 4.1709, "step": 520 }, { - "epoch": 0.29, - "learning_rate": 2.9602803738317758e-05, - "loss": 3.9659, + "epoch": 0.58, + "learning_rate": 2.862244897959184e-05, + "loss": 4.1588, "step": 521 }, { - "epoch": 0.29, - "learning_rate": 2.957943925233645e-05, - "loss": 4.1495, + "epoch": 0.59, + "learning_rate": 2.854591836734694e-05, + "loss": 4.104, "step": 522 }, { - "epoch": 0.29, - "learning_rate": 2.955607476635514e-05, - "loss": 4.1178, + "epoch": 0.59, + "learning_rate": 2.8469387755102042e-05, + "loss": 4.0468, "step": 523 }, { - "epoch": 0.29, - "learning_rate": 2.953271028037383e-05, - "loss": 4.3021, + "epoch": 0.59, + "learning_rate": 2.8392857142857144e-05, + "loss": 4.2316, "step": 524 }, { - "epoch": 0.29, - "learning_rate": 2.9509345794392525e-05, - "loss": 4.6051, + "epoch": 0.59, + "learning_rate": 2.8316326530612246e-05, + "loss": 4.3244, "step": 525 }, { - "epoch": 0.29, - "learning_rate": 2.9485981308411218e-05, - "loss": 4.289, + "epoch": 0.59, + "learning_rate": 2.8239795918367348e-05, + "loss": 4.0525, "step": 526 }, { - "epoch": 0.3, - "learning_rate": 2.9462616822429905e-05, - "loss": 4.425, + "epoch": 0.59, + "learning_rate": 2.816326530612245e-05, + "loss": 4.2284, "step": 527 }, { - "epoch": 0.3, - "learning_rate": 2.9439252336448598e-05, - "loss": 3.9979, + "epoch": 0.59, + "learning_rate": 2.8086734693877552e-05, + "loss": 4.1786, "step": 528 }, { - "epoch": 0.3, - "learning_rate": 2.941588785046729e-05, - "loss": 4.3364, + "epoch": 0.59, + "learning_rate": 2.8010204081632654e-05, + "loss": 4.14, "step": 529 }, { - "epoch": 0.3, - "learning_rate": 2.9392523364485985e-05, - "loss": 4.3657, + "epoch": 0.59, + "learning_rate": 2.7933673469387756e-05, + "loss": 4.4132, "step": 530 }, { - "epoch": 0.3, - "learning_rate": 2.936915887850467e-05, - "loss": 4.1333, + "epoch": 0.6, + "learning_rate": 2.7857142857142858e-05, + "loss": 4.273, "step": 531 }, { - "epoch": 0.3, - "learning_rate": 2.9345794392523365e-05, - "loss": 4.01, + "epoch": 0.6, + "learning_rate": 2.778061224489796e-05, + "loss": 3.9821, "step": 532 }, { - "epoch": 0.3, - "learning_rate": 2.932242990654206e-05, - "loss": 4.0705, + "epoch": 0.6, + "learning_rate": 2.770408163265306e-05, + "loss": 4.188, "step": 533 }, { - "epoch": 0.3, - "learning_rate": 2.929906542056075e-05, - "loss": 3.9721, + "epoch": 0.6, + "learning_rate": 2.7627551020408163e-05, + "loss": 4.0577, "step": 534 }, { - "epoch": 0.3, - "learning_rate": 2.927570093457944e-05, - "loss": 4.3871, + "epoch": 0.6, + "learning_rate": 2.7551020408163265e-05, + "loss": 4.2338, "step": 535 }, { - "epoch": 0.3, - "learning_rate": 2.9252336448598132e-05, - "loss": 4.2582, + "epoch": 0.6, + "learning_rate": 2.747448979591837e-05, + "loss": 4.2538, "step": 536 }, { - "epoch": 0.3, - "learning_rate": 2.9228971962616825e-05, - "loss": 4.1321, + "epoch": 0.6, + "learning_rate": 2.739795918367347e-05, + "loss": 4.3179, "step": 537 }, { - "epoch": 0.3, - "learning_rate": 2.9205607476635515e-05, - "loss": 4.4381, + "epoch": 0.6, + "learning_rate": 2.732142857142857e-05, + "loss": 4.3678, "step": 538 }, { - "epoch": 0.3, - "learning_rate": 2.9182242990654205e-05, - "loss": 4.2946, + "epoch": 0.6, + "learning_rate": 2.7244897959183673e-05, + "loss": 4.1675, "step": 539 }, { - "epoch": 0.3, - "learning_rate": 2.91588785046729e-05, - "loss": 4.3397, + "epoch": 0.61, + "learning_rate": 2.7168367346938778e-05, + "loss": 4.2616, "step": 540 }, { - "epoch": 0.3, - "learning_rate": 2.913551401869159e-05, - "loss": 4.5107, + "epoch": 0.61, + "learning_rate": 2.7091836734693877e-05, + "loss": 4.2889, "step": 541 }, { - "epoch": 0.3, - "learning_rate": 2.9112149532710282e-05, - "loss": 4.3734, + "epoch": 0.61, + "learning_rate": 2.701530612244898e-05, + "loss": 4.3385, "step": 542 }, { - "epoch": 0.3, - "learning_rate": 2.9088785046728972e-05, - "loss": 5.0012, + "epoch": 0.61, + "learning_rate": 2.6938775510204084e-05, + "loss": 4.4943, "step": 543 }, { - "epoch": 0.3, - "learning_rate": 2.9065420560747665e-05, - "loss": 4.3289, + "epoch": 0.61, + "learning_rate": 2.6862244897959186e-05, + "loss": 4.3588, "step": 544 }, { - "epoch": 0.31, - "learning_rate": 2.9042056074766355e-05, - "loss": 4.1193, + "epoch": 0.61, + "learning_rate": 2.6785714285714288e-05, + "loss": 4.386, "step": 545 }, { - "epoch": 0.31, - "learning_rate": 2.901869158878505e-05, - "loss": 4.4826, + "epoch": 0.61, + "learning_rate": 2.6709183673469386e-05, + "loss": 4.4699, "step": 546 }, { - "epoch": 0.31, - "learning_rate": 2.899532710280374e-05, - "loss": 4.7226, + "epoch": 0.61, + "learning_rate": 2.663265306122449e-05, + "loss": 4.3611, "step": 547 }, { - "epoch": 0.31, - "learning_rate": 2.897196261682243e-05, - "loss": 4.5186, + "epoch": 0.61, + "learning_rate": 2.6556122448979593e-05, + "loss": 4.6624, "step": 548 }, { - "epoch": 0.31, - "learning_rate": 2.8948598130841122e-05, - "loss": 4.7832, + "epoch": 0.62, + "learning_rate": 2.6479591836734695e-05, + "loss": 4.1314, "step": 549 }, { - "epoch": 0.31, - "learning_rate": 2.8925233644859816e-05, - "loss": 4.6424, + "epoch": 0.62, + "learning_rate": 2.6403061224489794e-05, + "loss": 5.1441, "step": 550 }, { - "epoch": 0.31, - "learning_rate": 2.8901869158878506e-05, - "loss": 4.6108, + "epoch": 0.62, + "learning_rate": 2.63265306122449e-05, + "loss": 4.1062, "step": 551 }, { - "epoch": 0.31, - "learning_rate": 2.8878504672897196e-05, - "loss": 4.484, + "epoch": 0.62, + "learning_rate": 2.625e-05, + "loss": 4.3416, "step": 552 }, { - "epoch": 0.31, - "learning_rate": 2.885514018691589e-05, - "loss": 4.4223, + "epoch": 0.62, + "learning_rate": 2.6173469387755103e-05, + "loss": 4.2483, "step": 553 }, { - "epoch": 0.31, - "learning_rate": 2.8831775700934582e-05, - "loss": 4.1946, + "epoch": 0.62, + "learning_rate": 2.6096938775510205e-05, + "loss": 4.1344, "step": 554 }, { - "epoch": 0.31, - "learning_rate": 2.8808411214953272e-05, - "loss": 4.105, + "epoch": 0.62, + "learning_rate": 2.6020408163265307e-05, + "loss": 4.1248, "step": 555 }, { - "epoch": 0.31, - "learning_rate": 2.8785046728971962e-05, - "loss": 4.2699, + "epoch": 0.62, + "learning_rate": 2.594387755102041e-05, + "loss": 4.1874, "step": 556 }, { - "epoch": 0.31, - "learning_rate": 2.8761682242990656e-05, - "loss": 4.3023, + "epoch": 0.62, + "learning_rate": 2.586734693877551e-05, + "loss": 4.0656, "step": 557 }, { - "epoch": 0.31, - "learning_rate": 2.873831775700935e-05, - "loss": 4.108, + "epoch": 0.63, + "learning_rate": 2.5790816326530613e-05, + "loss": 4.128, "step": 558 }, { - "epoch": 0.31, - "learning_rate": 2.8714953271028036e-05, - "loss": 4.065, + "epoch": 0.63, + "learning_rate": 2.5714285714285714e-05, + "loss": 4.1983, "step": 559 }, { - "epoch": 0.31, - "learning_rate": 2.869158878504673e-05, - "loss": 4.0726, + "epoch": 0.63, + "learning_rate": 2.5637755102040816e-05, + "loss": 4.0579, "step": 560 }, { - "epoch": 0.31, - "learning_rate": 2.8668224299065423e-05, - "loss": 4.3704, + "epoch": 0.63, + "learning_rate": 2.5561224489795918e-05, + "loss": 4.2611, "step": 561 }, { - "epoch": 0.32, - "learning_rate": 2.8644859813084116e-05, - "loss": 4.275, + "epoch": 0.63, + "learning_rate": 2.548469387755102e-05, + "loss": 4.094, "step": 562 }, { - "epoch": 0.32, - "learning_rate": 2.8621495327102803e-05, - "loss": 3.9922, + "epoch": 0.63, + "learning_rate": 2.5408163265306125e-05, + "loss": 4.0976, "step": 563 }, { - "epoch": 0.32, - "learning_rate": 2.8598130841121496e-05, - "loss": 4.6498, + "epoch": 0.63, + "learning_rate": 2.5331632653061224e-05, + "loss": 4.1275, "step": 564 }, { - "epoch": 0.32, - "learning_rate": 2.857476635514019e-05, - "loss": 4.4709, + "epoch": 0.63, + "learning_rate": 2.5255102040816326e-05, + "loss": 4.2178, "step": 565 }, { - "epoch": 0.32, - "learning_rate": 2.855140186915888e-05, - "loss": 4.1769, + "epoch": 0.63, + "learning_rate": 2.517857142857143e-05, + "loss": 4.2822, "step": 566 }, { - "epoch": 0.32, - "learning_rate": 2.852803738317757e-05, - "loss": 4.2284, + "epoch": 0.64, + "learning_rate": 2.5102040816326533e-05, + "loss": 4.1375, "step": 567 }, { - "epoch": 0.32, - "learning_rate": 2.8504672897196263e-05, - "loss": 4.4302, + "epoch": 0.64, + "learning_rate": 2.502551020408163e-05, + "loss": 4.0841, "step": 568 }, { - "epoch": 0.32, - "learning_rate": 2.8481308411214956e-05, - "loss": 4.2458, + "epoch": 0.64, + "learning_rate": 2.4948979591836734e-05, + "loss": 4.2421, "step": 569 }, { - "epoch": 0.32, - "learning_rate": 2.8457943925233646e-05, - "loss": 4.4577, + "epoch": 0.64, + "learning_rate": 2.487244897959184e-05, + "loss": 4.067, "step": 570 }, { - "epoch": 0.32, - "learning_rate": 2.8434579439252336e-05, - "loss": 4.1258, + "epoch": 0.64, + "learning_rate": 2.479591836734694e-05, + "loss": 4.0541, "step": 571 }, { - "epoch": 0.32, - "learning_rate": 2.841121495327103e-05, - "loss": 4.457, + "epoch": 0.64, + "learning_rate": 2.4719387755102043e-05, + "loss": 4.2383, "step": 572 }, { - "epoch": 0.32, - "learning_rate": 2.838785046728972e-05, - "loss": 4.1741, + "epoch": 0.64, + "learning_rate": 2.464285714285714e-05, + "loss": 4.0874, "step": 573 }, { - "epoch": 0.32, - "learning_rate": 2.8364485981308413e-05, - "loss": 4.0215, + "epoch": 0.64, + "learning_rate": 2.4566326530612246e-05, + "loss": 4.0371, "step": 574 }, { - "epoch": 0.32, - "learning_rate": 2.8341121495327103e-05, - "loss": 4.1338, + "epoch": 0.64, + "learning_rate": 2.448979591836735e-05, + "loss": 4.2899, "step": 575 }, { - "epoch": 0.32, - "learning_rate": 2.8317757009345797e-05, - "loss": 4.2283, + "epoch": 0.65, + "learning_rate": 2.441326530612245e-05, + "loss": 4.0796, "step": 576 }, { - "epoch": 0.32, - "learning_rate": 2.8294392523364487e-05, - "loss": 4.2055, + "epoch": 0.65, + "learning_rate": 2.433673469387755e-05, + "loss": 4.1797, "step": 577 }, { - "epoch": 0.32, - "learning_rate": 2.827102803738318e-05, - "loss": 4.0812, + "epoch": 0.65, + "learning_rate": 2.4260204081632654e-05, + "loss": 4.3384, "step": 578 }, { - "epoch": 0.32, - "learning_rate": 2.824766355140187e-05, - "loss": 4.1874, + "epoch": 0.65, + "learning_rate": 2.4183673469387756e-05, + "loss": 3.8088, "step": 579 }, { - "epoch": 0.33, - "learning_rate": 2.822429906542056e-05, - "loss": 4.4458, + "epoch": 0.65, + "learning_rate": 2.4107142857142858e-05, + "loss": 4.3933, "step": 580 }, { - "epoch": 0.33, - "learning_rate": 2.8200934579439253e-05, - "loss": 4.5588, + "epoch": 0.65, + "learning_rate": 2.403061224489796e-05, + "loss": 4.2551, "step": 581 }, { - "epoch": 0.33, - "learning_rate": 2.8177570093457947e-05, - "loss": 4.304, + "epoch": 0.65, + "learning_rate": 2.3954081632653062e-05, + "loss": 4.0488, "step": 582 }, { - "epoch": 0.33, - "learning_rate": 2.8154205607476637e-05, - "loss": 3.9939, + "epoch": 0.65, + "learning_rate": 2.3877551020408164e-05, + "loss": 4.1074, "step": 583 }, { - "epoch": 0.33, - "learning_rate": 2.8130841121495327e-05, - "loss": 4.4976, + "epoch": 0.65, + "learning_rate": 2.3801020408163266e-05, + "loss": 4.1928, "step": 584 }, { - "epoch": 0.33, - "learning_rate": 2.810747663551402e-05, - "loss": 4.4264, + "epoch": 0.66, + "learning_rate": 2.372448979591837e-05, + "loss": 4.3976, "step": 585 }, { - "epoch": 0.33, - "learning_rate": 2.8084112149532714e-05, - "loss": 4.218, + "epoch": 0.66, + "learning_rate": 2.364795918367347e-05, + "loss": 4.2446, "step": 586 }, { - "epoch": 0.33, - "learning_rate": 2.80607476635514e-05, - "loss": 4.4753, + "epoch": 0.66, + "learning_rate": 2.357142857142857e-05, + "loss": 4.255, "step": 587 }, { - "epoch": 0.33, - "learning_rate": 2.8037383177570094e-05, - "loss": 4.2927, + "epoch": 0.66, + "learning_rate": 2.3494897959183673e-05, + "loss": 4.2302, "step": 588 }, { - "epoch": 0.33, - "learning_rate": 2.8014018691588787e-05, - "loss": 4.3891, + "epoch": 0.66, + "learning_rate": 2.341836734693878e-05, + "loss": 3.9942, "step": 589 }, { - "epoch": 0.33, - "learning_rate": 2.799065420560748e-05, - "loss": 4.4293, + "epoch": 0.66, + "learning_rate": 2.3341836734693877e-05, + "loss": 4.1137, "step": 590 }, { - "epoch": 0.33, - "learning_rate": 2.7967289719626167e-05, - "loss": 4.1973, + "epoch": 0.66, + "learning_rate": 2.326530612244898e-05, + "loss": 4.2067, "step": 591 }, { - "epoch": 0.33, - "learning_rate": 2.794392523364486e-05, - "loss": 4.5021, + "epoch": 0.66, + "learning_rate": 2.318877551020408e-05, + "loss": 4.3063, "step": 592 }, { - "epoch": 0.33, - "learning_rate": 2.7920560747663554e-05, - "loss": 4.3199, + "epoch": 0.66, + "learning_rate": 2.3112244897959186e-05, + "loss": 4.4976, "step": 593 }, { - "epoch": 0.33, - "learning_rate": 2.7897196261682244e-05, - "loss": 4.4571, + "epoch": 0.67, + "learning_rate": 2.3035714285714288e-05, + "loss": 4.2891, "step": 594 }, { - "epoch": 0.33, - "learning_rate": 2.7873831775700934e-05, - "loss": 4.528, + "epoch": 0.67, + "learning_rate": 2.2959183673469387e-05, + "loss": 4.2596, "step": 595 }, { - "epoch": 0.33, - "learning_rate": 2.7850467289719627e-05, - "loss": 4.4157, + "epoch": 0.67, + "learning_rate": 2.288265306122449e-05, + "loss": 4.0942, "step": 596 }, { - "epoch": 0.33, - "learning_rate": 2.782710280373832e-05, - "loss": 4.5597, + "epoch": 0.67, + "learning_rate": 2.2806122448979594e-05, + "loss": 4.2176, "step": 597 }, { - "epoch": 0.34, - "learning_rate": 2.780373831775701e-05, - "loss": 4.9257, + "epoch": 0.67, + "learning_rate": 2.2729591836734696e-05, + "loss": 4.3389, "step": 598 }, { - "epoch": 0.34, - "learning_rate": 2.77803738317757e-05, - "loss": 4.7692, + "epoch": 0.67, + "learning_rate": 2.2653061224489794e-05, + "loss": 4.223, "step": 599 }, { - "epoch": 0.34, - "learning_rate": 2.7757009345794394e-05, - "loss": 4.8095, + "epoch": 0.67, + "learning_rate": 2.2576530612244896e-05, + "loss": 4.1904, "step": 600 }, { - "epoch": 0.34, - "learning_rate": 2.7733644859813084e-05, - "loss": 3.9404, + "epoch": 0.67, + "learning_rate": 2.25e-05, + "loss": 4.2231, "step": 601 }, { - "epoch": 0.34, - "learning_rate": 2.7710280373831777e-05, - "loss": 4.4654, + "epoch": 0.67, + "learning_rate": 2.2423469387755103e-05, + "loss": 4.0458, "step": 602 }, { - "epoch": 0.34, - "learning_rate": 2.7686915887850467e-05, - "loss": 4.1783, + "epoch": 0.68, + "learning_rate": 2.2346938775510205e-05, + "loss": 4.1966, "step": 603 }, { - "epoch": 0.34, - "learning_rate": 2.766355140186916e-05, - "loss": 4.2439, + "epoch": 0.68, + "learning_rate": 2.2270408163265307e-05, + "loss": 4.1522, "step": 604 }, { - "epoch": 0.34, - "learning_rate": 2.764018691588785e-05, - "loss": 4.072, + "epoch": 0.68, + "learning_rate": 2.219387755102041e-05, + "loss": 4.1061, "step": 605 }, { - "epoch": 0.34, - "learning_rate": 2.7616822429906544e-05, - "loss": 4.064, + "epoch": 0.68, + "learning_rate": 2.211734693877551e-05, + "loss": 4.1197, "step": 606 }, { - "epoch": 0.34, - "learning_rate": 2.7593457943925234e-05, - "loss": 4.1668, + "epoch": 0.68, + "learning_rate": 2.2040816326530613e-05, + "loss": 4.2229, "step": 607 }, { - "epoch": 0.34, - "learning_rate": 2.7570093457943924e-05, - "loss": 4.212, + "epoch": 0.68, + "learning_rate": 2.1964285714285715e-05, + "loss": 4.1381, "step": 608 }, { - "epoch": 0.34, - "learning_rate": 2.7546728971962618e-05, - "loss": 4.4345, + "epoch": 0.68, + "learning_rate": 2.1887755102040817e-05, + "loss": 4.1293, "step": 609 }, { - "epoch": 0.34, - "learning_rate": 2.752336448598131e-05, - "loss": 4.1837, + "epoch": 0.68, + "learning_rate": 2.181122448979592e-05, + "loss": 4.14, "step": 610 }, { - "epoch": 0.34, - "learning_rate": 2.75e-05, - "loss": 4.4643, + "epoch": 0.68, + "learning_rate": 2.173469387755102e-05, + "loss": 4.1027, "step": 611 }, { - "epoch": 0.34, - "learning_rate": 2.747663551401869e-05, - "loss": 4.1591, - "step": 612 + "epoch": 0.69, + "learning_rate": 2.1658163265306126e-05, + "loss": 4.2896, + "step": 612 }, { - "epoch": 0.34, - "learning_rate": 2.7453271028037384e-05, - "loss": 4.2991, + "epoch": 0.69, + "learning_rate": 2.1581632653061224e-05, + "loss": 4.1789, "step": 613 }, { - "epoch": 0.34, - "learning_rate": 2.7429906542056078e-05, - "loss": 4.1714, + "epoch": 0.69, + "learning_rate": 2.1505102040816326e-05, + "loss": 4.3021, "step": 614 }, { - "epoch": 0.34, - "learning_rate": 2.7406542056074764e-05, - "loss": 4.3276, + "epoch": 0.69, + "learning_rate": 2.1428571428571428e-05, + "loss": 4.1265, "step": 615 }, { - "epoch": 0.35, - "learning_rate": 2.7383177570093458e-05, - "loss": 4.0799, + "epoch": 0.69, + "learning_rate": 2.1352040816326533e-05, + "loss": 4.1473, "step": 616 }, { - "epoch": 0.35, - "learning_rate": 2.735981308411215e-05, - "loss": 4.1613, + "epoch": 0.69, + "learning_rate": 2.1275510204081632e-05, + "loss": 4.0719, "step": 617 }, { - "epoch": 0.35, - "learning_rate": 2.7336448598130845e-05, - "loss": 4.0844, + "epoch": 0.69, + "learning_rate": 2.1198979591836734e-05, + "loss": 4.1128, "step": 618 }, { - "epoch": 0.35, - "learning_rate": 2.731308411214953e-05, - "loss": 4.0714, + "epoch": 0.69, + "learning_rate": 2.1122448979591836e-05, + "loss": 4.0491, "step": 619 }, { - "epoch": 0.35, - "learning_rate": 2.7289719626168225e-05, - "loss": 4.3086, + "epoch": 0.7, + "learning_rate": 2.104591836734694e-05, + "loss": 4.1953, "step": 620 }, { - "epoch": 0.35, - "learning_rate": 2.7266355140186918e-05, - "loss": 4.2854, + "epoch": 0.7, + "learning_rate": 2.0969387755102043e-05, + "loss": 4.0822, "step": 621 }, { - "epoch": 0.35, - "learning_rate": 2.7242990654205608e-05, - "loss": 4.3431, + "epoch": 0.7, + "learning_rate": 2.089285714285714e-05, + "loss": 4.0029, "step": 622 }, { - "epoch": 0.35, - "learning_rate": 2.7219626168224298e-05, - "loss": 4.3078, + "epoch": 0.7, + "learning_rate": 2.0816326530612247e-05, + "loss": 4.2648, "step": 623 }, { - "epoch": 0.35, - "learning_rate": 2.719626168224299e-05, - "loss": 4.5547, + "epoch": 0.7, + "learning_rate": 2.073979591836735e-05, + "loss": 4.0931, "step": 624 }, { - "epoch": 0.35, - "learning_rate": 2.7172897196261685e-05, - "loss": 4.3998, + "epoch": 0.7, + "learning_rate": 2.066326530612245e-05, + "loss": 4.3205, "step": 625 }, { - "epoch": 0.35, - "learning_rate": 2.7149532710280375e-05, - "loss": 4.1948, + "epoch": 0.7, + "learning_rate": 2.058673469387755e-05, + "loss": 4.2901, "step": 626 }, { - "epoch": 0.35, - "learning_rate": 2.7126168224299065e-05, - "loss": 4.1507, + "epoch": 0.7, + "learning_rate": 2.0510204081632654e-05, + "loss": 4.181, "step": 627 }, { - "epoch": 0.35, - "learning_rate": 2.7102803738317758e-05, - "loss": 4.5187, + "epoch": 0.7, + "learning_rate": 2.0433673469387756e-05, + "loss": 4.1243, "step": 628 }, { - "epoch": 0.35, - "learning_rate": 2.7079439252336448e-05, - "loss": 4.82, + "epoch": 0.71, + "learning_rate": 2.0357142857142858e-05, + "loss": 4.1229, "step": 629 }, { - "epoch": 0.35, - "learning_rate": 2.705607476635514e-05, - "loss": 4.1649, + "epoch": 0.71, + "learning_rate": 2.0280612244897957e-05, + "loss": 4.2929, "step": 630 }, { - "epoch": 0.35, - "learning_rate": 2.703271028037383e-05, - "loss": 4.2138, + "epoch": 0.71, + "learning_rate": 2.0204081632653062e-05, + "loss": 4.2063, "step": 631 }, { - "epoch": 0.35, - "learning_rate": 2.7009345794392525e-05, - "loss": 4.2827, + "epoch": 0.71, + "learning_rate": 2.0127551020408164e-05, + "loss": 4.0951, "step": 632 }, { - "epoch": 0.35, - "learning_rate": 2.6985981308411215e-05, - "loss": 4.1455, + "epoch": 0.71, + "learning_rate": 2.0051020408163266e-05, + "loss": 4.1848, "step": 633 }, { - "epoch": 0.36, - "learning_rate": 2.696261682242991e-05, - "loss": 4.1618, + "epoch": 0.71, + "learning_rate": 1.9974489795918368e-05, + "loss": 4.1729, "step": 634 }, { - "epoch": 0.36, - "learning_rate": 2.69392523364486e-05, - "loss": 4.2469, + "epoch": 0.71, + "learning_rate": 1.989795918367347e-05, + "loss": 4.281, "step": 635 }, { - "epoch": 0.36, - "learning_rate": 2.691588785046729e-05, - "loss": 4.247, + "epoch": 0.71, + "learning_rate": 1.982142857142857e-05, + "loss": 4.2894, "step": 636 }, { - "epoch": 0.36, - "learning_rate": 2.6892523364485982e-05, - "loss": 4.2564, + "epoch": 0.71, + "learning_rate": 1.9744897959183674e-05, + "loss": 4.1699, "step": 637 }, { - "epoch": 0.36, - "learning_rate": 2.6869158878504675e-05, - "loss": 3.9329, + "epoch": 0.72, + "learning_rate": 1.9668367346938775e-05, + "loss": 4.0088, "step": 638 }, { - "epoch": 0.36, - "learning_rate": 2.6845794392523365e-05, - "loss": 4.3441, + "epoch": 0.72, + "learning_rate": 1.9591836734693877e-05, + "loss": 4.2616, "step": 639 }, { - "epoch": 0.36, - "learning_rate": 2.6822429906542055e-05, - "loss": 4.6779, + "epoch": 0.72, + "learning_rate": 1.951530612244898e-05, + "loss": 4.0165, "step": 640 }, { - "epoch": 0.36, - "learning_rate": 2.679906542056075e-05, - "loss": 4.4811, + "epoch": 0.72, + "learning_rate": 1.943877551020408e-05, + "loss": 4.1598, "step": 641 }, { - "epoch": 0.36, - "learning_rate": 2.6775700934579442e-05, - "loss": 4.4661, + "epoch": 0.72, + "learning_rate": 1.9362244897959186e-05, + "loss": 4.5368, "step": 642 }, { - "epoch": 0.36, - "learning_rate": 2.675233644859813e-05, - "loss": 4.1823, + "epoch": 0.72, + "learning_rate": 1.928571428571429e-05, + "loss": 4.3574, "step": 643 }, { - "epoch": 0.36, - "learning_rate": 2.6728971962616822e-05, - "loss": 4.3765, + "epoch": 0.72, + "learning_rate": 1.9209183673469387e-05, + "loss": 4.1234, "step": 644 }, { - "epoch": 0.36, - "learning_rate": 2.6705607476635515e-05, - "loss": 4.3031, + "epoch": 0.72, + "learning_rate": 1.913265306122449e-05, + "loss": 3.962, "step": 645 }, { - "epoch": 0.36, - "learning_rate": 2.668224299065421e-05, - "loss": 4.4316, + "epoch": 0.72, + "learning_rate": 1.9056122448979594e-05, + "loss": 4.2601, "step": 646 }, { - "epoch": 0.36, - "learning_rate": 2.6658878504672895e-05, - "loss": 5.1151, + "epoch": 0.73, + "learning_rate": 1.8979591836734696e-05, + "loss": 4.1748, "step": 647 }, { - "epoch": 0.36, - "learning_rate": 2.663551401869159e-05, - "loss": 5.3826, + "epoch": 0.73, + "learning_rate": 1.8903061224489795e-05, + "loss": 4.1067, "step": 648 }, { - "epoch": 0.36, - "learning_rate": 2.6612149532710282e-05, - "loss": 4.624, + "epoch": 0.73, + "learning_rate": 1.8826530612244896e-05, + "loss": 4.4876, "step": 649 }, { - "epoch": 0.36, - "learning_rate": 2.6588785046728972e-05, - "loss": 4.6994, + "epoch": 0.73, + "learning_rate": 1.8750000000000002e-05, + "loss": 4.2195, "step": 650 }, { - "epoch": 0.36, - "learning_rate": 2.6565420560747662e-05, - "loss": 4.1818, + "epoch": 0.73, + "learning_rate": 1.8673469387755104e-05, + "loss": 4.2406, "step": 651 }, { - "epoch": 0.37, - "learning_rate": 2.6542056074766356e-05, - "loss": 4.1108, + "epoch": 0.73, + "learning_rate": 1.8596938775510206e-05, + "loss": 4.2498, "step": 652 }, { - "epoch": 0.37, - "learning_rate": 2.651869158878505e-05, - "loss": 4.1831, + "epoch": 0.73, + "learning_rate": 1.8520408163265304e-05, + "loss": 4.2465, "step": 653 }, { - "epoch": 0.37, - "learning_rate": 2.649532710280374e-05, - "loss": 4.1909, + "epoch": 0.73, + "learning_rate": 1.844387755102041e-05, + "loss": 4.2118, "step": 654 }, { - "epoch": 0.37, - "learning_rate": 2.647196261682243e-05, - "loss": 4.1307, + "epoch": 0.73, + "learning_rate": 1.836734693877551e-05, + "loss": 4.1716, "step": 655 }, { - "epoch": 0.37, - "learning_rate": 2.6448598130841122e-05, - "loss": 4.1173, + "epoch": 0.74, + "learning_rate": 1.8290816326530613e-05, + "loss": 4.1002, "step": 656 }, { - "epoch": 0.37, - "learning_rate": 2.6425233644859816e-05, - "loss": 4.1902, + "epoch": 0.74, + "learning_rate": 1.8214285714285712e-05, + "loss": 4.1314, "step": 657 }, { - "epoch": 0.37, - "learning_rate": 2.6401869158878506e-05, - "loss": 4.5231, + "epoch": 0.74, + "learning_rate": 1.8137755102040817e-05, + "loss": 4.3314, "step": 658 }, { - "epoch": 0.37, - "learning_rate": 2.6378504672897196e-05, - "loss": 4.4093, + "epoch": 0.74, + "learning_rate": 1.806122448979592e-05, + "loss": 4.148, "step": 659 }, { - "epoch": 0.37, - "learning_rate": 2.635514018691589e-05, - "loss": 4.2605, + "epoch": 0.74, + "learning_rate": 1.798469387755102e-05, + "loss": 4.2266, "step": 660 }, { - "epoch": 0.37, - "learning_rate": 2.633177570093458e-05, - "loss": 4.2977, + "epoch": 0.74, + "learning_rate": 1.7908163265306126e-05, + "loss": 3.9961, "step": 661 }, { - "epoch": 0.37, - "learning_rate": 2.6308411214953273e-05, - "loss": 4.2694, + "epoch": 0.74, + "learning_rate": 1.7831632653061225e-05, + "loss": 3.9725, "step": 662 }, { - "epoch": 0.37, - "learning_rate": 2.6285046728971963e-05, - "loss": 4.3017, + "epoch": 0.74, + "learning_rate": 1.7755102040816327e-05, + "loss": 4.325, "step": 663 }, { - "epoch": 0.37, - "learning_rate": 2.6261682242990656e-05, - "loss": 4.386, + "epoch": 0.74, + "learning_rate": 1.767857142857143e-05, + "loss": 4.0709, "step": 664 }, { - "epoch": 0.37, - "learning_rate": 2.6238317757009346e-05, - "loss": 4.1274, + "epoch": 0.75, + "learning_rate": 1.7602040816326534e-05, + "loss": 4.0242, "step": 665 }, { - "epoch": 0.37, - "learning_rate": 2.621495327102804e-05, - "loss": 4.3211, + "epoch": 0.75, + "learning_rate": 1.7525510204081632e-05, + "loss": 4.2876, "step": 666 }, { - "epoch": 0.37, - "learning_rate": 2.619158878504673e-05, - "loss": 4.2019, + "epoch": 0.75, + "learning_rate": 1.7448979591836734e-05, + "loss": 4.0623, "step": 667 }, { - "epoch": 0.37, - "learning_rate": 2.616822429906542e-05, - "loss": 4.0505, + "epoch": 0.75, + "learning_rate": 1.7372448979591836e-05, + "loss": 4.2775, "step": 668 }, { - "epoch": 0.38, - "learning_rate": 2.6144859813084113e-05, - "loss": 4.0959, + "epoch": 0.75, + "learning_rate": 1.729591836734694e-05, + "loss": 4.237, "step": 669 }, { - "epoch": 0.38, - "learning_rate": 2.6121495327102806e-05, - "loss": 4.3859, + "epoch": 0.75, + "learning_rate": 1.7219387755102043e-05, + "loss": 4.2058, "step": 670 }, { - "epoch": 0.38, - "learning_rate": 2.6098130841121496e-05, - "loss": 4.3131, + "epoch": 0.75, + "learning_rate": 1.7142857142857142e-05, + "loss": 4.0794, "step": 671 }, { - "epoch": 0.38, - "learning_rate": 2.6074766355140186e-05, - "loss": 4.5609, + "epoch": 0.75, + "learning_rate": 1.7066326530612244e-05, + "loss": 3.9794, "step": 672 }, { - "epoch": 0.38, - "learning_rate": 2.605140186915888e-05, - "loss": 4.1302, + "epoch": 0.75, + "learning_rate": 1.698979591836735e-05, + "loss": 3.9593, "step": 673 }, { - "epoch": 0.38, - "learning_rate": 2.6028037383177573e-05, - "loss": 4.1409, + "epoch": 0.76, + "learning_rate": 1.691326530612245e-05, + "loss": 4.0232, "step": 674 }, { - "epoch": 0.38, - "learning_rate": 2.600467289719626e-05, - "loss": 4.0495, + "epoch": 0.76, + "learning_rate": 1.683673469387755e-05, + "loss": 4.0226, "step": 675 }, { - "epoch": 0.38, - "learning_rate": 2.5981308411214953e-05, - "loss": 4.2181, + "epoch": 0.76, + "learning_rate": 1.676020408163265e-05, + "loss": 4.1974, "step": 676 }, { - "epoch": 0.38, - "learning_rate": 2.5957943925233647e-05, - "loss": 4.1967, + "epoch": 0.76, + "learning_rate": 1.6683673469387757e-05, + "loss": 3.9916, "step": 677 }, { - "epoch": 0.38, - "learning_rate": 2.593457943925234e-05, - "loss": 4.1595, + "epoch": 0.76, + "learning_rate": 1.660714285714286e-05, + "loss": 3.9367, "step": 678 }, { - "epoch": 0.38, - "learning_rate": 2.5911214953271027e-05, - "loss": 4.2405, + "epoch": 0.76, + "learning_rate": 1.6530612244897957e-05, + "loss": 3.9962, "step": 679 }, { - "epoch": 0.38, - "learning_rate": 2.588785046728972e-05, - "loss": 4.1914, + "epoch": 0.76, + "learning_rate": 1.6454081632653062e-05, + "loss": 4.0845, "step": 680 }, { - "epoch": 0.38, - "learning_rate": 2.5864485981308413e-05, - "loss": 4.1382, + "epoch": 0.76, + "learning_rate": 1.6377551020408164e-05, + "loss": 4.3325, "step": 681 }, { - "epoch": 0.38, - "learning_rate": 2.5841121495327103e-05, - "loss": 4.3131, + "epoch": 0.76, + "learning_rate": 1.6301020408163266e-05, + "loss": 4.0054, "step": 682 }, { - "epoch": 0.38, - "learning_rate": 2.5817757009345793e-05, - "loss": 4.4054, + "epoch": 0.77, + "learning_rate": 1.6224489795918368e-05, + "loss": 4.1694, "step": 683 }, { - "epoch": 0.38, - "learning_rate": 2.5794392523364487e-05, - "loss": 4.3854, + "epoch": 0.77, + "learning_rate": 1.614795918367347e-05, + "loss": 4.1841, "step": 684 }, { - "epoch": 0.38, - "learning_rate": 2.577102803738318e-05, - "loss": 4.4064, + "epoch": 0.77, + "learning_rate": 1.6071428571428572e-05, + "loss": 4.2737, "step": 685 }, { - "epoch": 0.38, - "learning_rate": 2.574766355140187e-05, - "loss": 4.3073, + "epoch": 0.77, + "learning_rate": 1.5994897959183674e-05, + "loss": 4.1188, "step": 686 }, { - "epoch": 0.39, - "learning_rate": 2.572429906542056e-05, - "loss": 3.9777, + "epoch": 0.77, + "learning_rate": 1.5918367346938776e-05, + "loss": 4.1621, "step": 687 }, { - "epoch": 0.39, - "learning_rate": 2.5700934579439254e-05, - "loss": 4.0989, + "epoch": 0.77, + "learning_rate": 1.5841836734693878e-05, + "loss": 4.2523, "step": 688 }, { - "epoch": 0.39, - "learning_rate": 2.5677570093457944e-05, - "loss": 4.2607, + "epoch": 0.77, + "learning_rate": 1.576530612244898e-05, + "loss": 4.272, "step": 689 }, { - "epoch": 0.39, - "learning_rate": 2.5654205607476637e-05, - "loss": 4.333, + "epoch": 0.77, + "learning_rate": 1.568877551020408e-05, + "loss": 4.3658, "step": 690 }, { - "epoch": 0.39, - "learning_rate": 2.5630841121495327e-05, - "loss": 4.0871, + "epoch": 0.77, + "learning_rate": 1.5612244897959183e-05, + "loss": 4.0139, "step": 691 }, { - "epoch": 0.39, - "learning_rate": 2.560747663551402e-05, - "loss": 4.3324, + "epoch": 0.78, + "learning_rate": 1.553571428571429e-05, + "loss": 4.1947, "step": 692 }, { - "epoch": 0.39, - "learning_rate": 2.558411214953271e-05, - "loss": 4.3102, + "epoch": 0.78, + "learning_rate": 1.5459183673469387e-05, + "loss": 4.0925, "step": 693 }, { - "epoch": 0.39, - "learning_rate": 2.5560747663551404e-05, - "loss": 4.2393, + "epoch": 0.78, + "learning_rate": 1.538265306122449e-05, + "loss": 4.4659, "step": 694 }, { - "epoch": 0.39, - "learning_rate": 2.5537383177570094e-05, - "loss": 4.4242, + "epoch": 0.78, + "learning_rate": 1.530612244897959e-05, + "loss": 4.269, "step": 695 }, { - "epoch": 0.39, - "learning_rate": 2.5514018691588784e-05, - "loss": 4.5045, + "epoch": 0.78, + "learning_rate": 1.5229591836734695e-05, + "loss": 4.2835, "step": 696 }, { - "epoch": 0.39, - "learning_rate": 2.5490654205607477e-05, - "loss": 3.9591, + "epoch": 0.78, + "learning_rate": 1.5153061224489795e-05, + "loss": 4.3569, "step": 697 }, { - "epoch": 0.39, - "learning_rate": 2.546728971962617e-05, - "loss": 4.289, + "epoch": 0.78, + "learning_rate": 1.5076530612244897e-05, + "loss": 4.3019, "step": 698 }, { - "epoch": 0.39, - "learning_rate": 2.544392523364486e-05, - "loss": 4.2378, + "epoch": 0.78, + "learning_rate": 1.5e-05, + "loss": 4.3235, "step": 699 }, { - "epoch": 0.39, - "learning_rate": 2.542056074766355e-05, - "loss": 4.2691, + "epoch": 0.78, + "learning_rate": 1.4923469387755104e-05, + "loss": 4.4127, "step": 700 }, { - "epoch": 0.39, - "learning_rate": 2.5397196261682244e-05, - "loss": 4.0229, + "epoch": 0.79, + "learning_rate": 1.4846938775510204e-05, + "loss": 4.0454, "step": 701 }, { - "epoch": 0.39, - "learning_rate": 2.5373831775700937e-05, - "loss": 4.153, + "epoch": 0.79, + "learning_rate": 1.4770408163265308e-05, + "loss": 4.1379, "step": 702 }, { - "epoch": 0.39, - "learning_rate": 2.5350467289719624e-05, - "loss": 4.1632, + "epoch": 0.79, + "learning_rate": 1.4693877551020408e-05, + "loss": 4.0424, "step": 703 }, { - "epoch": 0.39, - "learning_rate": 2.5327102803738317e-05, - "loss": 4.4474, + "epoch": 0.79, + "learning_rate": 1.4617346938775512e-05, + "loss": 3.9108, "step": 704 }, { - "epoch": 0.4, - "learning_rate": 2.530373831775701e-05, - "loss": 4.2303, + "epoch": 0.79, + "learning_rate": 1.4540816326530612e-05, + "loss": 4.0812, "step": 705 }, { - "epoch": 0.4, - "learning_rate": 2.5280373831775704e-05, - "loss": 4.3282, + "epoch": 0.79, + "learning_rate": 1.4464285714285715e-05, + "loss": 4.2139, "step": 706 }, { - "epoch": 0.4, - "learning_rate": 2.525700934579439e-05, - "loss": 4.0947, + "epoch": 0.79, + "learning_rate": 1.4387755102040816e-05, + "loss": 4.23, "step": 707 }, { - "epoch": 0.4, - "learning_rate": 2.5233644859813084e-05, - "loss": 4.1924, + "epoch": 0.79, + "learning_rate": 1.431122448979592e-05, + "loss": 4.1661, "step": 708 }, { - "epoch": 0.4, - "learning_rate": 2.5210280373831778e-05, - "loss": 3.9936, + "epoch": 0.79, + "learning_rate": 1.4234693877551021e-05, + "loss": 4.0422, "step": 709 }, { - "epoch": 0.4, - "learning_rate": 2.5186915887850468e-05, - "loss": 4.2294, + "epoch": 0.8, + "learning_rate": 1.4158163265306123e-05, + "loss": 4.1259, "step": 710 }, { - "epoch": 0.4, - "learning_rate": 2.5163551401869158e-05, - "loss": 4.3738, + "epoch": 0.8, + "learning_rate": 1.4081632653061225e-05, + "loss": 4.0219, "step": 711 }, { - "epoch": 0.4, - "learning_rate": 2.514018691588785e-05, - "loss": 4.1016, + "epoch": 0.8, + "learning_rate": 1.4005102040816327e-05, + "loss": 4.0935, "step": 712 }, { - "epoch": 0.4, - "learning_rate": 2.5116822429906544e-05, - "loss": 4.1244, + "epoch": 0.8, + "learning_rate": 1.3928571428571429e-05, + "loss": 4.0343, "step": 713 }, { - "epoch": 0.4, - "learning_rate": 2.5093457943925234e-05, - "loss": 4.5749, + "epoch": 0.8, + "learning_rate": 1.385204081632653e-05, + "loss": 4.2476, "step": 714 }, { - "epoch": 0.4, - "learning_rate": 2.5070093457943924e-05, - "loss": 4.5455, + "epoch": 0.8, + "learning_rate": 1.3775510204081633e-05, + "loss": 4.2593, "step": 715 }, { - "epoch": 0.4, - "learning_rate": 2.5046728971962618e-05, - "loss": 4.0365, + "epoch": 0.8, + "learning_rate": 1.3698979591836735e-05, + "loss": 4.0613, "step": 716 }, { - "epoch": 0.4, - "learning_rate": 2.5023364485981308e-05, - "loss": 4.1837, + "epoch": 0.8, + "learning_rate": 1.3622448979591836e-05, + "loss": 4.2907, "step": 717 }, { - "epoch": 0.4, - "learning_rate": 2.5e-05, - "loss": 4.3674, + "epoch": 0.8, + "learning_rate": 1.3545918367346938e-05, + "loss": 4.057, "step": 718 }, { - "epoch": 0.4, - "learning_rate": 2.497663551401869e-05, - "loss": 4.068, + "epoch": 0.81, + "learning_rate": 1.3469387755102042e-05, + "loss": 3.8796, "step": 719 }, { - "epoch": 0.4, - "learning_rate": 2.4953271028037385e-05, - "loss": 4.2529, + "epoch": 0.81, + "learning_rate": 1.3392857142857144e-05, + "loss": 4.1651, "step": 720 }, { - "epoch": 0.4, - "learning_rate": 2.4929906542056075e-05, - "loss": 4.3839, + "epoch": 0.81, + "learning_rate": 1.3316326530612246e-05, + "loss": 3.965, "step": 721 }, { - "epoch": 0.4, - "learning_rate": 2.4906542056074768e-05, - "loss": 4.3385, + "epoch": 0.81, + "learning_rate": 1.3239795918367348e-05, + "loss": 4.017, "step": 722 }, { - "epoch": 0.41, - "learning_rate": 2.4883177570093458e-05, - "loss": 4.1595, + "epoch": 0.81, + "learning_rate": 1.316326530612245e-05, + "loss": 4.0912, "step": 723 }, { - "epoch": 0.41, - "learning_rate": 2.4859813084112148e-05, - "loss": 4.4662, + "epoch": 0.81, + "learning_rate": 1.3086734693877551e-05, + "loss": 4.1649, "step": 724 }, { - "epoch": 0.41, - "learning_rate": 2.483644859813084e-05, - "loss": 4.2813, + "epoch": 0.81, + "learning_rate": 1.3010204081632653e-05, + "loss": 4.0581, "step": 725 }, { - "epoch": 0.41, - "learning_rate": 2.4813084112149535e-05, - "loss": 4.188, + "epoch": 0.81, + "learning_rate": 1.2933673469387755e-05, + "loss": 4.1833, "step": 726 }, { - "epoch": 0.41, - "learning_rate": 2.4789719626168225e-05, - "loss": 3.8812, + "epoch": 0.82, + "learning_rate": 1.2857142857142857e-05, + "loss": 4.1175, "step": 727 }, { - "epoch": 0.41, - "learning_rate": 2.4766355140186915e-05, - "loss": 4.377, + "epoch": 0.82, + "learning_rate": 1.2780612244897959e-05, + "loss": 4.0868, "step": 728 }, { - "epoch": 0.41, - "learning_rate": 2.4742990654205608e-05, - "loss": 4.1591, + "epoch": 0.82, + "learning_rate": 1.2704081632653063e-05, + "loss": 3.9839, "step": 729 }, { - "epoch": 0.41, - "learning_rate": 2.47196261682243e-05, - "loss": 4.1232, + "epoch": 0.82, + "learning_rate": 1.2627551020408163e-05, + "loss": 4.2042, "step": 730 }, { - "epoch": 0.41, - "learning_rate": 2.4696261682242988e-05, - "loss": 4.2111, + "epoch": 0.82, + "learning_rate": 1.2551020408163267e-05, + "loss": 4.3366, "step": 731 }, { - "epoch": 0.41, - "learning_rate": 2.467289719626168e-05, - "loss": 4.195, + "epoch": 0.82, + "learning_rate": 1.2474489795918367e-05, + "loss": 4.1627, "step": 732 }, { - "epoch": 0.41, - "learning_rate": 2.4649532710280375e-05, - "loss": 4.5914, + "epoch": 0.82, + "learning_rate": 1.239795918367347e-05, + "loss": 4.0717, "step": 733 }, { - "epoch": 0.41, - "learning_rate": 2.462616822429907e-05, - "loss": 4.2196, + "epoch": 0.82, + "learning_rate": 1.232142857142857e-05, + "loss": 4.2349, "step": 734 }, { - "epoch": 0.41, - "learning_rate": 2.4602803738317755e-05, - "loss": 4.2522, + "epoch": 0.82, + "learning_rate": 1.2244897959183674e-05, + "loss": 3.9495, "step": 735 }, { - "epoch": 0.41, - "learning_rate": 2.457943925233645e-05, - "loss": 4.1694, + "epoch": 0.83, + "learning_rate": 1.2168367346938774e-05, + "loss": 4.072, "step": 736 }, { - "epoch": 0.41, - "learning_rate": 2.4556074766355142e-05, - "loss": 4.2373, + "epoch": 0.83, + "learning_rate": 1.2091836734693878e-05, + "loss": 4.2674, "step": 737 }, { - "epoch": 0.41, - "learning_rate": 2.4532710280373832e-05, - "loss": 3.9855, + "epoch": 0.83, + "learning_rate": 1.201530612244898e-05, + "loss": 4.1501, "step": 738 }, { - "epoch": 0.41, - "learning_rate": 2.4509345794392522e-05, - "loss": 4.2423, + "epoch": 0.83, + "learning_rate": 1.1938775510204082e-05, + "loss": 4.0789, "step": 739 }, { - "epoch": 0.41, - "learning_rate": 2.4485981308411215e-05, - "loss": 4.2095, + "epoch": 0.83, + "learning_rate": 1.1862244897959185e-05, + "loss": 4.2417, "step": 740 }, { - "epoch": 0.42, - "learning_rate": 2.446261682242991e-05, - "loss": 4.4789, + "epoch": 0.83, + "learning_rate": 1.1785714285714286e-05, + "loss": 4.2802, "step": 741 }, { - "epoch": 0.42, - "learning_rate": 2.44392523364486e-05, - "loss": 4.0323, + "epoch": 0.83, + "learning_rate": 1.170918367346939e-05, + "loss": 4.066, "step": 742 }, { - "epoch": 0.42, - "learning_rate": 2.441588785046729e-05, - "loss": 4.2841, + "epoch": 0.83, + "learning_rate": 1.163265306122449e-05, + "loss": 3.953, "step": 743 }, { - "epoch": 0.42, - "learning_rate": 2.4392523364485982e-05, - "loss": 4.2162, + "epoch": 0.83, + "learning_rate": 1.1556122448979593e-05, + "loss": 4.0866, "step": 744 }, { - "epoch": 0.42, - "learning_rate": 2.4369158878504672e-05, - "loss": 4.7925, + "epoch": 0.84, + "learning_rate": 1.1479591836734693e-05, + "loss": 4.2401, "step": 745 }, { - "epoch": 0.42, - "learning_rate": 2.4345794392523365e-05, - "loss": 4.2842, + "epoch": 0.84, + "learning_rate": 1.1403061224489797e-05, + "loss": 4.3716, "step": 746 }, { - "epoch": 0.42, - "learning_rate": 2.4322429906542055e-05, - "loss": 4.4468, + "epoch": 0.84, + "learning_rate": 1.1326530612244897e-05, + "loss": 4.2219, "step": 747 }, { - "epoch": 0.42, - "learning_rate": 2.429906542056075e-05, - "loss": 4.5483, + "epoch": 0.84, + "learning_rate": 1.125e-05, + "loss": 4.3632, "step": 748 }, { - "epoch": 0.42, - "learning_rate": 2.427570093457944e-05, - "loss": 4.1929, + "epoch": 0.84, + "learning_rate": 1.1173469387755103e-05, + "loss": 4.5002, "step": 749 }, { - "epoch": 0.42, - "learning_rate": 2.4252336448598132e-05, - "loss": 4.5784, + "epoch": 0.84, + "learning_rate": 1.1096938775510205e-05, + "loss": 4.2068, "step": 750 }, { - "epoch": 0.42, - "learning_rate": 2.4228971962616822e-05, - "loss": 4.2466, + "epoch": 0.84, + "learning_rate": 1.1020408163265306e-05, + "loss": 4.2042, "step": 751 }, { - "epoch": 0.42, - "learning_rate": 2.4205607476635516e-05, - "loss": 4.256, + "epoch": 0.84, + "learning_rate": 1.0943877551020408e-05, + "loss": 4.1899, "step": 752 }, { - "epoch": 0.42, - "learning_rate": 2.4182242990654206e-05, - "loss": 4.2916, + "epoch": 0.84, + "learning_rate": 1.086734693877551e-05, + "loss": 4.1472, "step": 753 }, { - "epoch": 0.42, - "learning_rate": 2.41588785046729e-05, - "loss": 4.0228, + "epoch": 0.85, + "learning_rate": 1.0790816326530612e-05, + "loss": 4.037, "step": 754 }, { - "epoch": 0.42, - "learning_rate": 2.413551401869159e-05, - "loss": 4.1207, + "epoch": 0.85, + "learning_rate": 1.0714285714285714e-05, + "loss": 4.1912, "step": 755 }, { - "epoch": 0.42, - "learning_rate": 2.411214953271028e-05, - "loss": 4.2923, + "epoch": 0.85, + "learning_rate": 1.0637755102040816e-05, + "loss": 4.2355, "step": 756 }, { - "epoch": 0.42, - "learning_rate": 2.4088785046728972e-05, - "loss": 4.0716, + "epoch": 0.85, + "learning_rate": 1.0561224489795918e-05, + "loss": 4.1236, "step": 757 }, { - "epoch": 0.42, - "learning_rate": 2.4065420560747666e-05, - "loss": 4.2561, + "epoch": 0.85, + "learning_rate": 1.0484693877551021e-05, + "loss": 4.2208, "step": 758 }, { - "epoch": 0.43, - "learning_rate": 2.4042056074766356e-05, - "loss": 4.2501, + "epoch": 0.85, + "learning_rate": 1.0408163265306123e-05, + "loss": 4.0635, "step": 759 }, { - "epoch": 0.43, - "learning_rate": 2.4018691588785046e-05, - "loss": 4.1544, + "epoch": 0.85, + "learning_rate": 1.0331632653061225e-05, + "loss": 4.1248, "step": 760 }, { - "epoch": 0.43, - "learning_rate": 2.399532710280374e-05, - "loss": 4.5181, + "epoch": 0.85, + "learning_rate": 1.0255102040816327e-05, + "loss": 4.0792, "step": 761 }, { - "epoch": 0.43, - "learning_rate": 2.3971962616822433e-05, - "loss": 4.1383, + "epoch": 0.85, + "learning_rate": 1.0178571428571429e-05, + "loss": 4.1373, "step": 762 }, { - "epoch": 0.43, - "learning_rate": 2.394859813084112e-05, - "loss": 4.2302, + "epoch": 0.86, + "learning_rate": 1.0102040816326531e-05, + "loss": 4.0515, "step": 763 }, { - "epoch": 0.43, - "learning_rate": 2.3925233644859813e-05, - "loss": 4.0103, + "epoch": 0.86, + "learning_rate": 1.0025510204081633e-05, + "loss": 4.0731, "step": 764 }, { - "epoch": 0.43, - "learning_rate": 2.3901869158878506e-05, - "loss": 3.9473, + "epoch": 0.86, + "learning_rate": 9.948979591836735e-06, + "loss": 4.1974, "step": 765 }, { - "epoch": 0.43, - "learning_rate": 2.38785046728972e-05, - "loss": 4.188, + "epoch": 0.86, + "learning_rate": 9.872448979591837e-06, + "loss": 4.0851, "step": 766 }, { - "epoch": 0.43, - "learning_rate": 2.3855140186915886e-05, - "loss": 4.3639, + "epoch": 0.86, + "learning_rate": 9.795918367346939e-06, + "loss": 4.0982, "step": 767 }, { - "epoch": 0.43, - "learning_rate": 2.383177570093458e-05, - "loss": 3.93, + "epoch": 0.86, + "learning_rate": 9.71938775510204e-06, + "loss": 4.1605, "step": 768 }, { - "epoch": 0.43, - "learning_rate": 2.3808411214953273e-05, - "loss": 4.5287, + "epoch": 0.86, + "learning_rate": 9.642857142857144e-06, + "loss": 4.1659, "step": 769 }, { - "epoch": 0.43, - "learning_rate": 2.3785046728971963e-05, - "loss": 4.2475, + "epoch": 0.86, + "learning_rate": 9.566326530612244e-06, + "loss": 4.0684, "step": 770 }, { - "epoch": 0.43, - "learning_rate": 2.3761682242990653e-05, - "loss": 4.1368, + "epoch": 0.86, + "learning_rate": 9.489795918367348e-06, + "loss": 4.0657, "step": 771 }, { - "epoch": 0.43, - "learning_rate": 2.3738317757009346e-05, - "loss": 4.1505, + "epoch": 0.87, + "learning_rate": 9.413265306122448e-06, + "loss": 4.238, "step": 772 }, { - "epoch": 0.43, - "learning_rate": 2.371495327102804e-05, - "loss": 4.0876, + "epoch": 0.87, + "learning_rate": 9.336734693877552e-06, + "loss": 4.2754, "step": 773 }, { - "epoch": 0.43, - "learning_rate": 2.369158878504673e-05, - "loss": 4.2193, + "epoch": 0.87, + "learning_rate": 9.260204081632652e-06, + "loss": 3.915, "step": 774 }, { - "epoch": 0.43, - "learning_rate": 2.366822429906542e-05, - "loss": 4.0992, + "epoch": 0.87, + "learning_rate": 9.183673469387756e-06, + "loss": 3.9955, "step": 775 }, { - "epoch": 0.43, - "learning_rate": 2.3644859813084113e-05, - "loss": 4.1377, + "epoch": 0.87, + "learning_rate": 9.107142857142856e-06, + "loss": 4.1319, "step": 776 }, { - "epoch": 0.44, - "learning_rate": 2.3621495327102803e-05, - "loss": 4.0301, + "epoch": 0.87, + "learning_rate": 9.03061224489796e-06, + "loss": 3.9204, "step": 777 }, { - "epoch": 0.44, - "learning_rate": 2.3598130841121497e-05, - "loss": 4.421, + "epoch": 0.87, + "learning_rate": 8.954081632653063e-06, + "loss": 4.1897, "step": 778 }, { - "epoch": 0.44, - "learning_rate": 2.3574766355140187e-05, - "loss": 4.3282, + "epoch": 0.87, + "learning_rate": 8.877551020408163e-06, + "loss": 4.0072, "step": 779 }, { - "epoch": 0.44, - "learning_rate": 2.355140186915888e-05, - "loss": 4.3547, + "epoch": 0.87, + "learning_rate": 8.801020408163267e-06, + "loss": 4.0125, "step": 780 }, { - "epoch": 0.44, - "learning_rate": 2.352803738317757e-05, - "loss": 4.4139, + "epoch": 0.88, + "learning_rate": 8.724489795918367e-06, + "loss": 3.9928, "step": 781 }, { - "epoch": 0.44, - "learning_rate": 2.3504672897196263e-05, - "loss": 4.347, + "epoch": 0.88, + "learning_rate": 8.64795918367347e-06, + "loss": 4.1519, "step": 782 }, { - "epoch": 0.44, - "learning_rate": 2.3481308411214953e-05, - "loss": 4.2355, + "epoch": 0.88, + "learning_rate": 8.571428571428571e-06, + "loss": 4.1489, "step": 783 }, { - "epoch": 0.44, - "learning_rate": 2.3457943925233643e-05, - "loss": 4.1498, + "epoch": 0.88, + "learning_rate": 8.494897959183675e-06, + "loss": 4.0451, "step": 784 }, { - "epoch": 0.44, - "learning_rate": 2.3434579439252337e-05, - "loss": 4.0922, + "epoch": 0.88, + "learning_rate": 8.418367346938775e-06, + "loss": 4.0635, "step": 785 }, { - "epoch": 0.44, - "learning_rate": 2.341121495327103e-05, - "loss": 4.478, + "epoch": 0.88, + "learning_rate": 8.341836734693878e-06, + "loss": 3.9523, "step": 786 }, { - "epoch": 0.44, - "learning_rate": 2.338785046728972e-05, - "loss": 3.8649, + "epoch": 0.88, + "learning_rate": 8.265306122448979e-06, + "loss": 4.1608, "step": 787 }, { - "epoch": 0.44, - "learning_rate": 2.336448598130841e-05, - "loss": 4.2899, + "epoch": 0.88, + "learning_rate": 8.188775510204082e-06, + "loss": 4.1575, "step": 788 }, { - "epoch": 0.44, - "learning_rate": 2.3341121495327104e-05, - "loss": 4.2538, + "epoch": 0.88, + "learning_rate": 8.112244897959184e-06, + "loss": 4.0074, "step": 789 }, { - "epoch": 0.44, - "learning_rate": 2.3317757009345797e-05, - "loss": 4.2685, + "epoch": 0.89, + "learning_rate": 8.035714285714286e-06, + "loss": 4.1326, "step": 790 }, { - "epoch": 0.44, - "learning_rate": 2.3294392523364484e-05, - "loss": 4.4079, + "epoch": 0.89, + "learning_rate": 7.959183673469388e-06, + "loss": 4.1211, "step": 791 }, { - "epoch": 0.44, - "learning_rate": 2.3271028037383177e-05, - "loss": 4.035, + "epoch": 0.89, + "learning_rate": 7.88265306122449e-06, + "loss": 3.9565, "step": 792 }, { - "epoch": 0.44, - "learning_rate": 2.324766355140187e-05, - "loss": 4.2569, + "epoch": 0.89, + "learning_rate": 7.806122448979592e-06, + "loss": 4.1213, "step": 793 }, { - "epoch": 0.45, - "learning_rate": 2.3224299065420564e-05, - "loss": 4.1586, + "epoch": 0.89, + "learning_rate": 7.729591836734694e-06, + "loss": 4.2507, "step": 794 }, { - "epoch": 0.45, - "learning_rate": 2.320093457943925e-05, - "loss": 4.5834, + "epoch": 0.89, + "learning_rate": 7.653061224489796e-06, + "loss": 4.2829, "step": 795 }, { - "epoch": 0.45, - "learning_rate": 2.3177570093457944e-05, - "loss": 4.2447, + "epoch": 0.89, + "learning_rate": 7.5765306122448974e-06, + "loss": 4.1432, "step": 796 }, { - "epoch": 0.45, - "learning_rate": 2.3154205607476637e-05, - "loss": 4.2122, + "epoch": 0.89, + "learning_rate": 7.5e-06, + "loss": 4.2565, "step": 797 }, { - "epoch": 0.45, - "learning_rate": 2.3130841121495327e-05, - "loss": 4.5336, + "epoch": 0.89, + "learning_rate": 7.423469387755102e-06, + "loss": 4.3092, "step": 798 }, { - "epoch": 0.45, - "learning_rate": 2.3107476635514017e-05, - "loss": 4.3484, + "epoch": 0.9, + "learning_rate": 7.346938775510204e-06, + "loss": 4.6345, "step": 799 }, { - "epoch": 0.45, - "learning_rate": 2.308411214953271e-05, - "loss": 4.4033, + "epoch": 0.9, + "learning_rate": 7.270408163265306e-06, + "loss": 4.5129, "step": 800 }, { - "epoch": 0.45, - "learning_rate": 2.3060747663551404e-05, - "loss": 4.2892, + "epoch": 0.9, + "learning_rate": 7.193877551020408e-06, + "loss": 4.1561, "step": 801 }, { - "epoch": 0.45, - "learning_rate": 2.3037383177570094e-05, - "loss": 4.292, + "epoch": 0.9, + "learning_rate": 7.117346938775511e-06, + "loss": 4.0536, "step": 802 }, { - "epoch": 0.45, - "learning_rate": 2.3014018691588784e-05, - "loss": 4.322, + "epoch": 0.9, + "learning_rate": 7.0408163265306125e-06, + "loss": 4.0601, "step": 803 }, { - "epoch": 0.45, - "learning_rate": 2.2990654205607477e-05, - "loss": 4.1898, + "epoch": 0.9, + "learning_rate": 6.964285714285714e-06, + "loss": 4.122, "step": 804 }, { - "epoch": 0.45, - "learning_rate": 2.2967289719626167e-05, - "loss": 4.2183, + "epoch": 0.9, + "learning_rate": 6.887755102040816e-06, + "loss": 3.9336, "step": 805 }, { - "epoch": 0.45, - "learning_rate": 2.294392523364486e-05, - "loss": 4.3549, + "epoch": 0.9, + "learning_rate": 6.811224489795918e-06, + "loss": 4.1128, "step": 806 }, { - "epoch": 0.45, - "learning_rate": 2.292056074766355e-05, - "loss": 4.2268, + "epoch": 0.9, + "learning_rate": 6.734693877551021e-06, + "loss": 4.183, "step": 807 }, { - "epoch": 0.45, - "learning_rate": 2.2897196261682244e-05, - "loss": 3.9229, + "epoch": 0.91, + "learning_rate": 6.658163265306123e-06, + "loss": 3.9993, "step": 808 }, { - "epoch": 0.45, - "learning_rate": 2.2873831775700934e-05, - "loss": 4.3157, + "epoch": 0.91, + "learning_rate": 6.581632653061225e-06, + "loss": 4.2294, "step": 809 }, { - "epoch": 0.45, - "learning_rate": 2.2850467289719628e-05, - "loss": 4.3893, + "epoch": 0.91, + "learning_rate": 6.505102040816327e-06, + "loss": 3.9998, "step": 810 }, { - "epoch": 0.45, - "learning_rate": 2.2827102803738318e-05, - "loss": 4.2611, + "epoch": 0.91, + "learning_rate": 6.428571428571429e-06, + "loss": 4.2926, "step": 811 }, { - "epoch": 0.46, - "learning_rate": 2.2803738317757008e-05, - "loss": 4.2447, + "epoch": 0.91, + "learning_rate": 6.352040816326531e-06, + "loss": 4.1365, "step": 812 }, { - "epoch": 0.46, - "learning_rate": 2.27803738317757e-05, - "loss": 4.2921, + "epoch": 0.91, + "learning_rate": 6.275510204081633e-06, + "loss": 3.9426, "step": 813 }, { - "epoch": 0.46, - "learning_rate": 2.2757009345794394e-05, - "loss": 4.0179, + "epoch": 0.91, + "learning_rate": 6.198979591836735e-06, + "loss": 4.0211, "step": 814 }, { - "epoch": 0.46, - "learning_rate": 2.2733644859813084e-05, - "loss": 4.1825, + "epoch": 0.91, + "learning_rate": 6.122448979591837e-06, + "loss": 4.0301, "step": 815 }, { - "epoch": 0.46, - "learning_rate": 2.2710280373831774e-05, - "loss": 4.1997, + "epoch": 0.91, + "learning_rate": 6.045918367346939e-06, + "loss": 4.017, "step": 816 }, { - "epoch": 0.46, - "learning_rate": 2.2686915887850468e-05, - "loss": 4.0681, + "epoch": 0.92, + "learning_rate": 5.969387755102041e-06, + "loss": 4.0372, "step": 817 }, { - "epoch": 0.46, - "learning_rate": 2.266355140186916e-05, - "loss": 4.3193, + "epoch": 0.92, + "learning_rate": 5.892857142857143e-06, + "loss": 4.0688, "step": 818 }, { - "epoch": 0.46, - "learning_rate": 2.2640186915887848e-05, - "loss": 4.1792, + "epoch": 0.92, + "learning_rate": 5.816326530612245e-06, + "loss": 4.0503, "step": 819 }, { - "epoch": 0.46, - "learning_rate": 2.261682242990654e-05, - "loss": 4.1559, + "epoch": 0.92, + "learning_rate": 5.739795918367347e-06, + "loss": 4.0608, "step": 820 }, { - "epoch": 0.46, - "learning_rate": 2.2593457943925235e-05, - "loss": 4.4142, + "epoch": 0.92, + "learning_rate": 5.6632653061224486e-06, + "loss": 3.9885, "step": 821 }, { - "epoch": 0.46, - "learning_rate": 2.2570093457943928e-05, - "loss": 4.1386, + "epoch": 0.92, + "learning_rate": 5.586734693877551e-06, + "loss": 4.0087, "step": 822 }, { - "epoch": 0.46, - "learning_rate": 2.2546728971962615e-05, - "loss": 4.5255, + "epoch": 0.92, + "learning_rate": 5.510204081632653e-06, + "loss": 3.9787, "step": 823 }, - { - "epoch": 0.46, - "learning_rate": 2.2523364485981308e-05, - "loss": 4.2801, - "step": 824 - }, - { - "epoch": 0.46, - "learning_rate": 2.25e-05, - "loss": 4.1778, - "step": 825 - }, - { - "epoch": 0.46, - "learning_rate": 2.247663551401869e-05, - "loss": 4.2408, - "step": 826 - }, - { - "epoch": 0.46, - "learning_rate": 2.2453271028037385e-05, - "loss": 4.4065, - "step": 827 - }, - { - "epoch": 0.46, - "learning_rate": 2.2429906542056075e-05, - "loss": 4.1209, - "step": 828 - }, - { - "epoch": 0.46, - "learning_rate": 2.2406542056074768e-05, - "loss": 4.1902, - "step": 829 - }, - { - "epoch": 0.47, - "learning_rate": 2.2383177570093458e-05, - "loss": 4.1085, - "step": 830 - }, - { - "epoch": 0.47, - "learning_rate": 2.235981308411215e-05, - "loss": 4.4849, - "step": 831 - }, - { - "epoch": 0.47, - "learning_rate": 2.233644859813084e-05, - "loss": 4.5035, - "step": 832 - }, - { - "epoch": 0.47, - "learning_rate": 2.231308411214953e-05, - "loss": 4.308, - "step": 833 - }, - { - "epoch": 0.47, - "learning_rate": 2.2289719626168225e-05, - "loss": 4.0138, - "step": 834 - }, - { - "epoch": 0.47, - "learning_rate": 2.226635514018692e-05, - "loss": 4.0107, - "step": 835 - }, - { - "epoch": 0.47, - "learning_rate": 2.224299065420561e-05, - "loss": 4.1838, - "step": 836 - }, - { - "epoch": 0.47, - "learning_rate": 2.22196261682243e-05, - "loss": 4.3247, - "step": 837 - }, - { - "epoch": 0.47, - "learning_rate": 2.2196261682242992e-05, - "loss": 4.2136, - "step": 838 - }, - { - "epoch": 0.47, - "learning_rate": 2.2172897196261685e-05, - "loss": 4.0103, - "step": 839 - }, - { - "epoch": 0.47, - "learning_rate": 2.2149532710280372e-05, - "loss": 4.2826, - "step": 840 - }, - { - "epoch": 0.47, - "learning_rate": 2.2126168224299065e-05, - "loss": 4.3545, - "step": 841 - }, - { - "epoch": 0.47, - "learning_rate": 2.210280373831776e-05, - "loss": 4.1326, - "step": 842 - }, - { - "epoch": 0.47, - "learning_rate": 2.2079439252336452e-05, - "loss": 4.1696, - "step": 843 - }, - { - "epoch": 0.47, - "learning_rate": 2.205607476635514e-05, - "loss": 4.2908, - "step": 844 - }, - { - "epoch": 0.47, - "learning_rate": 2.2032710280373832e-05, - "loss": 4.024, - "step": 845 - }, - { - "epoch": 0.47, - "learning_rate": 2.2009345794392525e-05, - "loss": 4.1375, - "step": 846 - }, - { - "epoch": 0.47, - "learning_rate": 2.1985981308411215e-05, - "loss": 3.9984, - "step": 847 - }, - { - "epoch": 0.48, - "learning_rate": 2.1962616822429905e-05, - "loss": 4.4272, - "step": 848 - }, - { - "epoch": 0.48, - "learning_rate": 2.19392523364486e-05, - "loss": 4.6845, - "step": 849 - }, - { - "epoch": 0.48, - "learning_rate": 2.1915887850467292e-05, - "loss": 4.6562, - "step": 850 - }, - { - "epoch": 0.48, - "learning_rate": 2.1892523364485982e-05, - "loss": 4.2803, - "step": 851 - }, - { - "epoch": 0.48, - "learning_rate": 2.1869158878504672e-05, - "loss": 4.2674, - "step": 852 - }, - { - "epoch": 0.48, - "learning_rate": 2.1845794392523366e-05, - "loss": 4.3411, - "step": 853 - }, - { - "epoch": 0.48, - "learning_rate": 2.182242990654206e-05, - "loss": 4.2781, - "step": 854 - }, - { - "epoch": 0.48, - "learning_rate": 2.179906542056075e-05, - "loss": 4.2525, - "step": 855 - }, - { - "epoch": 0.48, - "learning_rate": 2.177570093457944e-05, - "loss": 4.1371, - "step": 856 - }, - { - "epoch": 0.48, - "learning_rate": 2.1752336448598132e-05, - "loss": 4.1449, - "step": 857 - }, - { - "epoch": 0.48, - "learning_rate": 2.1728971962616822e-05, - "loss": 4.4371, - "step": 858 - }, - { - "epoch": 0.48, - "learning_rate": 2.1705607476635516e-05, - "loss": 4.2179, - "step": 859 - }, - { - "epoch": 0.48, - "learning_rate": 2.1682242990654206e-05, - "loss": 4.529, - "step": 860 - }, - { - "epoch": 0.48, - "learning_rate": 2.16588785046729e-05, - "loss": 3.8545, - "step": 861 - }, - { - "epoch": 0.48, - "learning_rate": 2.163551401869159e-05, - "loss": 4.1672, - "step": 862 - }, - { - "epoch": 0.48, - "learning_rate": 2.1612149532710283e-05, - "loss": 4.3361, - "step": 863 - }, - { - "epoch": 0.48, - "learning_rate": 2.1588785046728973e-05, - "loss": 4.0718, - "step": 864 - }, - { - "epoch": 0.48, - "learning_rate": 2.1565420560747663e-05, - "loss": 4.2156, - "step": 865 - }, - { - "epoch": 0.49, - "learning_rate": 2.1542056074766356e-05, - "loss": 4.272, - "step": 866 - }, - { - "epoch": 0.49, - "learning_rate": 2.151869158878505e-05, - "loss": 4.206, - "step": 867 - }, - { - "epoch": 0.49, - "learning_rate": 2.149532710280374e-05, - "loss": 4.2278, - "step": 868 - }, - { - "epoch": 0.49, - "learning_rate": 2.147196261682243e-05, - "loss": 4.2667, - "step": 869 - }, - { - "epoch": 0.49, - "learning_rate": 2.1448598130841123e-05, - "loss": 4.1348, - "step": 870 - }, - { - "epoch": 0.49, - "learning_rate": 2.1425233644859816e-05, - "loss": 4.1959, - "step": 871 - }, - { - "epoch": 0.49, - "learning_rate": 2.1401869158878503e-05, - "loss": 4.3323, - "step": 872 - }, - { - "epoch": 0.49, - "learning_rate": 2.1378504672897196e-05, - "loss": 4.2832, - "step": 873 - }, - { - "epoch": 0.49, - "learning_rate": 2.135514018691589e-05, - "loss": 4.109, - "step": 874 - }, - { - "epoch": 0.49, - "learning_rate": 2.1331775700934583e-05, - "loss": 4.4554, - "step": 875 - }, - { - "epoch": 0.49, - "learning_rate": 2.130841121495327e-05, - "loss": 4.2466, - "step": 876 - }, - { - "epoch": 0.49, - "learning_rate": 2.1285046728971963e-05, - "loss": 4.5604, - "step": 877 - }, - { - "epoch": 0.49, - "learning_rate": 2.1261682242990657e-05, - "loss": 4.2792, - "step": 878 - }, - { - "epoch": 0.49, - "learning_rate": 2.1238317757009347e-05, - "loss": 4.1452, - "step": 879 - }, - { - "epoch": 0.49, - "learning_rate": 2.1214953271028037e-05, - "loss": 4.3134, - "step": 880 - }, - { - "epoch": 0.49, - "learning_rate": 2.119158878504673e-05, - "loss": 4.4312, - "step": 881 - }, - { - "epoch": 0.49, - "learning_rate": 2.1168224299065423e-05, - "loss": 4.0045, - "step": 882 - }, - { - "epoch": 0.49, - "learning_rate": 2.1144859813084113e-05, - "loss": 4.3507, - "step": 883 - }, - { - "epoch": 0.5, - "learning_rate": 2.1121495327102803e-05, - "loss": 4.4745, - "step": 884 - }, - { - "epoch": 0.5, - "learning_rate": 2.1098130841121497e-05, - "loss": 4.441, - "step": 885 - }, - { - "epoch": 0.5, - "learning_rate": 2.1074766355140187e-05, - "loss": 4.3963, - "step": 886 - }, - { - "epoch": 0.5, - "learning_rate": 2.105140186915888e-05, - "loss": 4.691, - "step": 887 - }, - { - "epoch": 0.5, - "learning_rate": 2.102803738317757e-05, - "loss": 4.0693, - "step": 888 - }, - { - "epoch": 0.5, - "learning_rate": 2.1004672897196264e-05, - "loss": 4.0052, - "step": 889 - }, - { - "epoch": 0.5, - "learning_rate": 2.0981308411214954e-05, - "loss": 4.439, - "step": 890 - }, - { - "epoch": 0.5, - "learning_rate": 2.0957943925233647e-05, - "loss": 4.2542, - "step": 891 - }, - { - "epoch": 0.5, - "learning_rate": 2.0934579439252337e-05, - "loss": 4.3875, - "step": 892 - }, - { - "epoch": 0.5, - "learning_rate": 2.0911214953271027e-05, - "loss": 4.6864, - "step": 893 - }, - { - "epoch": 0.5, - "learning_rate": 2.088785046728972e-05, - "loss": 4.316, - "step": 894 - }, - { - "epoch": 0.5, - "learning_rate": 2.0864485981308414e-05, - "loss": 4.2591, - "step": 895 - }, - { - "epoch": 0.5, - "learning_rate": 2.0841121495327104e-05, - "loss": 4.0914, - "step": 896 - }, - { - "epoch": 0.5, - "learning_rate": 2.0817757009345794e-05, - "loss": 4.2464, - "step": 897 - }, - { - "epoch": 0.5, - "learning_rate": 2.0794392523364487e-05, - "loss": 4.9274, - "step": 898 - }, - { - "epoch": 0.5, - "learning_rate": 2.077102803738318e-05, - "loss": 4.7296, - "step": 899 - }, - { - "epoch": 0.5, - "learning_rate": 2.0747663551401867e-05, - "loss": 5.0117, - "step": 900 - }, - { - "epoch": 0.51, - "learning_rate": 2.072429906542056e-05, - "loss": 4.2921, - "step": 901 - }, - { - "epoch": 0.51, - "learning_rate": 2.0700934579439254e-05, - "loss": 4.0694, - "step": 902 - }, - { - "epoch": 0.51, - "learning_rate": 2.0677570093457947e-05, - "loss": 4.5726, - "step": 903 - }, - { - "epoch": 0.51, - "learning_rate": 2.0654205607476634e-05, - "loss": 4.1894, - "step": 904 - }, - { - "epoch": 0.51, - "learning_rate": 2.0630841121495327e-05, - "loss": 4.2496, - "step": 905 - }, - { - "epoch": 0.51, - "learning_rate": 2.060747663551402e-05, - "loss": 4.3837, - "step": 906 - }, - { - "epoch": 0.51, - "learning_rate": 2.058411214953271e-05, - "loss": 4.3154, - "step": 907 - }, - { - "epoch": 0.51, - "learning_rate": 2.05607476635514e-05, - "loss": 4.1458, - "step": 908 - }, - { - "epoch": 0.51, - "learning_rate": 2.0537383177570094e-05, - "loss": 4.2969, - "step": 909 - }, - { - "epoch": 0.51, - "learning_rate": 2.0514018691588788e-05, - "loss": 4.1434, - "step": 910 - }, - { - "epoch": 0.51, - "learning_rate": 2.0490654205607478e-05, - "loss": 4.2901, - "step": 911 - }, - { - "epoch": 0.51, - "learning_rate": 2.0467289719626168e-05, - "loss": 4.0777, - "step": 912 - }, - { - "epoch": 0.51, - "learning_rate": 2.044392523364486e-05, - "loss": 4.2098, - "step": 913 - }, - { - "epoch": 0.51, - "learning_rate": 2.042056074766355e-05, - "loss": 4.1676, - "step": 914 - }, - { - "epoch": 0.51, - "learning_rate": 2.0397196261682244e-05, - "loss": 4.3009, - "step": 915 - }, - { - "epoch": 0.51, - "learning_rate": 2.0373831775700934e-05, - "loss": 4.2482, - "step": 916 - }, - { - "epoch": 0.51, - "learning_rate": 2.0350467289719628e-05, - "loss": 4.0691, - "step": 917 - }, - { - "epoch": 0.51, - "learning_rate": 2.0327102803738318e-05, - "loss": 4.2496, - "step": 918 - }, - { - "epoch": 0.52, - "learning_rate": 2.030373831775701e-05, - "loss": 4.1904, - "step": 919 - }, - { - "epoch": 0.52, - "learning_rate": 2.02803738317757e-05, - "loss": 4.0056, - "step": 920 - }, - { - "epoch": 0.52, - "learning_rate": 2.025700934579439e-05, - "loss": 4.1408, - "step": 921 - }, - { - "epoch": 0.52, - "learning_rate": 2.0233644859813085e-05, - "loss": 3.9573, - "step": 922 - }, - { - "epoch": 0.52, - "learning_rate": 2.0210280373831778e-05, - "loss": 4.1, - "step": 923 - }, - { - "epoch": 0.52, - "learning_rate": 2.0186915887850468e-05, - "loss": 4.1254, - "step": 924 - }, - { - "epoch": 0.52, - "learning_rate": 2.0163551401869158e-05, - "loss": 4.2993, - "step": 925 - }, - { - "epoch": 0.52, - "learning_rate": 2.014018691588785e-05, - "loss": 4.3646, - "step": 926 - }, - { - "epoch": 0.52, - "learning_rate": 2.0116822429906545e-05, - "loss": 4.1477, - "step": 927 - }, - { - "epoch": 0.52, - "learning_rate": 2.009345794392523e-05, - "loss": 4.0236, - "step": 928 - }, - { - "epoch": 0.52, - "learning_rate": 2.0070093457943925e-05, - "loss": 4.1204, - "step": 929 - }, - { - "epoch": 0.52, - "learning_rate": 2.0046728971962618e-05, - "loss": 4.3844, - "step": 930 - }, - { - "epoch": 0.52, - "learning_rate": 2.002336448598131e-05, - "loss": 4.4103, - "step": 931 - }, - { - "epoch": 0.52, - "learning_rate": 1.9999999999999998e-05, - "loss": 4.1294, - "step": 932 - }, - { - "epoch": 0.52, - "learning_rate": 1.997663551401869e-05, - "loss": 4.2804, - "step": 933 - }, - { - "epoch": 0.52, - "learning_rate": 1.9953271028037385e-05, - "loss": 4.3015, - "step": 934 - }, - { - "epoch": 0.52, - "learning_rate": 1.9929906542056075e-05, - "loss": 4.4078, - "step": 935 - }, - { - "epoch": 0.52, - "learning_rate": 1.9906542056074765e-05, - "loss": 4.3072, - "step": 936 - }, - { - "epoch": 0.53, - "learning_rate": 1.988317757009346e-05, - "loss": 4.1355, - "step": 937 - }, - { - "epoch": 0.53, - "learning_rate": 1.9859813084112152e-05, - "loss": 3.9358, - "step": 938 - }, - { - "epoch": 0.53, - "learning_rate": 1.9836448598130842e-05, - "loss": 3.954, - "step": 939 - }, - { - "epoch": 0.53, - "learning_rate": 1.9813084112149532e-05, - "loss": 4.3639, - "step": 940 - }, - { - "epoch": 0.53, - "learning_rate": 1.9789719626168225e-05, - "loss": 4.2587, - "step": 941 - }, - { - "epoch": 0.53, - "learning_rate": 1.9766355140186915e-05, - "loss": 4.2508, - "step": 942 - }, - { - "epoch": 0.53, - "learning_rate": 1.974299065420561e-05, - "loss": 4.4542, - "step": 943 - }, - { - "epoch": 0.53, - "learning_rate": 1.97196261682243e-05, - "loss": 4.1811, - "step": 944 - }, - { - "epoch": 0.53, - "learning_rate": 1.9696261682242992e-05, - "loss": 4.3571, - "step": 945 - }, - { - "epoch": 0.53, - "learning_rate": 1.9672897196261682e-05, - "loss": 4.3206, - "step": 946 - }, - { - "epoch": 0.53, - "learning_rate": 1.9649532710280375e-05, - "loss": 3.8047, - "step": 947 - }, - { - "epoch": 0.53, - "learning_rate": 1.9626168224299065e-05, - "loss": 4.3799, - "step": 948 - }, - { - "epoch": 0.53, - "learning_rate": 1.960280373831776e-05, - "loss": 5.1039, - "step": 949 - }, - { - "epoch": 0.53, - "learning_rate": 1.957943925233645e-05, - "loss": 4.0424, - "step": 950 - }, - { - "epoch": 0.53, - "learning_rate": 1.9556074766355142e-05, - "loss": 4.163, - "step": 951 - }, - { - "epoch": 0.53, - "learning_rate": 1.9532710280373832e-05, - "loss": 4.2665, - "step": 952 - }, - { - "epoch": 0.53, - "learning_rate": 1.9509345794392522e-05, - "loss": 4.3006, - "step": 953 - }, - { - "epoch": 0.53, - "learning_rate": 1.9485981308411216e-05, - "loss": 3.9481, - "step": 954 - }, - { - "epoch": 0.54, - "learning_rate": 1.946261682242991e-05, - "loss": 4.3406, - "step": 955 - }, - { - "epoch": 0.54, - "learning_rate": 1.94392523364486e-05, - "loss": 4.1748, - "step": 956 - }, - { - "epoch": 0.54, - "learning_rate": 1.941588785046729e-05, - "loss": 4.2548, - "step": 957 - }, - { - "epoch": 0.54, - "learning_rate": 1.9392523364485982e-05, - "loss": 4.0172, - "step": 958 - }, - { - "epoch": 0.54, - "learning_rate": 1.9369158878504676e-05, - "loss": 4.5716, - "step": 959 - }, - { - "epoch": 0.54, - "learning_rate": 1.9345794392523363e-05, - "loss": 4.2715, - "step": 960 - }, - { - "epoch": 0.54, - "learning_rate": 1.9322429906542056e-05, - "loss": 4.111, - "step": 961 - }, - { - "epoch": 0.54, - "learning_rate": 1.929906542056075e-05, - "loss": 4.1517, - "step": 962 - }, - { - "epoch": 0.54, - "learning_rate": 1.9275700934579443e-05, - "loss": 4.1256, - "step": 963 - }, - { - "epoch": 0.54, - "learning_rate": 1.925233644859813e-05, - "loss": 4.2684, - "step": 964 - }, - { - "epoch": 0.54, - "learning_rate": 1.9228971962616823e-05, - "loss": 4.1267, - "step": 965 - }, - { - "epoch": 0.54, - "learning_rate": 1.9205607476635516e-05, - "loss": 4.0476, - "step": 966 - }, - { - "epoch": 0.54, - "learning_rate": 1.9182242990654206e-05, - "loss": 4.2444, - "step": 967 - }, - { - "epoch": 0.54, - "learning_rate": 1.9158878504672896e-05, - "loss": 4.1258, - "step": 968 - }, - { - "epoch": 0.54, - "learning_rate": 1.913551401869159e-05, - "loss": 4.1988, - "step": 969 - }, - { - "epoch": 0.54, - "learning_rate": 1.9112149532710283e-05, - "loss": 4.1309, - "step": 970 - }, - { - "epoch": 0.54, - "learning_rate": 1.9088785046728973e-05, - "loss": 4.2583, - "step": 971 - }, - { - "epoch": 0.54, - "learning_rate": 1.9065420560747663e-05, - "loss": 4.1522, - "step": 972 - }, - { - "epoch": 0.55, - "learning_rate": 1.9042056074766356e-05, - "loss": 4.221, - "step": 973 - }, - { - "epoch": 0.55, - "learning_rate": 1.9018691588785046e-05, - "loss": 4.4322, - "step": 974 - }, - { - "epoch": 0.55, - "learning_rate": 1.899532710280374e-05, - "loss": 4.3223, - "step": 975 - }, - { - "epoch": 0.55, - "learning_rate": 1.897196261682243e-05, - "loss": 4.2805, - "step": 976 - }, - { - "epoch": 0.55, - "learning_rate": 1.8948598130841123e-05, - "loss": 4.5155, - "step": 977 - }, - { - "epoch": 0.55, - "learning_rate": 1.8925233644859813e-05, - "loss": 4.1073, - "step": 978 - }, - { - "epoch": 0.55, - "learning_rate": 1.8901869158878507e-05, - "loss": 4.0066, - "step": 979 - }, - { - "epoch": 0.55, - "learning_rate": 1.8878504672897197e-05, - "loss": 4.0313, - "step": 980 - }, - { - "epoch": 0.55, - "learning_rate": 1.8855140186915887e-05, - "loss": 3.8829, - "step": 981 - }, - { - "epoch": 0.55, - "learning_rate": 1.883177570093458e-05, - "loss": 3.8576, - "step": 982 - }, - { - "epoch": 0.55, - "learning_rate": 1.8808411214953273e-05, - "loss": 4.3572, - "step": 983 - }, - { - "epoch": 0.55, - "learning_rate": 1.8785046728971963e-05, - "loss": 4.327, - "step": 984 - }, - { - "epoch": 0.55, - "learning_rate": 1.8761682242990653e-05, - "loss": 4.0703, - "step": 985 - }, - { - "epoch": 0.55, - "learning_rate": 1.8738317757009347e-05, - "loss": 4.1239, - "step": 986 - }, - { - "epoch": 0.55, - "learning_rate": 1.871495327102804e-05, - "loss": 4.3195, - "step": 987 - }, - { - "epoch": 0.55, - "learning_rate": 1.8691588785046727e-05, - "loss": 4.2369, - "step": 988 - }, - { - "epoch": 0.55, - "learning_rate": 1.866822429906542e-05, - "loss": 4.3695, - "step": 989 - }, - { - "epoch": 0.55, - "learning_rate": 1.8644859813084114e-05, - "loss": 4.0731, - "step": 990 - }, - { - "epoch": 0.56, - "learning_rate": 1.8621495327102807e-05, - "loss": 4.4181, - "step": 991 - }, - { - "epoch": 0.56, - "learning_rate": 1.8598130841121494e-05, - "loss": 4.585, - "step": 992 - }, - { - "epoch": 0.56, - "learning_rate": 1.8574766355140187e-05, - "loss": 4.335, - "step": 993 - }, - { - "epoch": 0.56, - "learning_rate": 1.855140186915888e-05, - "loss": 4.2939, - "step": 994 - }, - { - "epoch": 0.56, - "learning_rate": 1.852803738317757e-05, - "loss": 4.1693, - "step": 995 - }, - { - "epoch": 0.56, - "learning_rate": 1.850467289719626e-05, - "loss": 3.786, - "step": 996 - }, - { - "epoch": 0.56, - "learning_rate": 1.8481308411214954e-05, - "loss": 4.2314, - "step": 997 - }, - { - "epoch": 0.56, - "learning_rate": 1.8457943925233647e-05, - "loss": 4.8163, - "step": 998 - }, - { - "epoch": 0.56, - "learning_rate": 1.8434579439252337e-05, - "loss": 4.8442, - "step": 999 - }, - { - "epoch": 0.56, - "learning_rate": 1.8411214953271027e-05, - "loss": 4.3994, - "step": 1000 - }, - { - "epoch": 0.56, - "eval_loss": 4.276415824890137, - "eval_runtime": 941.3717, - "eval_samples_per_second": 2.807, - "eval_steps_per_second": 0.352, - "eval_wer": 1.960353034510115, - "step": 1000 - }, - { - "epoch": 0.56, - "learning_rate": 1.838785046728972e-05, - "loss": 4.3838, - "step": 1001 - }, - { - "epoch": 0.56, - "learning_rate": 1.836448598130841e-05, - "loss": 4.1807, - "step": 1002 - }, - { - "epoch": 0.56, - "learning_rate": 1.8341121495327104e-05, - "loss": 4.3417, - "step": 1003 - }, - { - "epoch": 0.56, - "learning_rate": 1.8317757009345794e-05, - "loss": 4.2058, - "step": 1004 - }, - { - "epoch": 0.56, - "learning_rate": 1.8294392523364487e-05, - "loss": 4.1705, - "step": 1005 - }, - { - "epoch": 0.56, - "learning_rate": 1.8271028037383177e-05, - "loss": 4.2186, - "step": 1006 - }, - { - "epoch": 0.56, - "learning_rate": 1.824766355140187e-05, - "loss": 4.1123, - "step": 1007 - }, - { - "epoch": 0.57, - "learning_rate": 1.822429906542056e-05, - "loss": 4.4482, - "step": 1008 - }, - { - "epoch": 0.57, - "learning_rate": 1.820093457943925e-05, - "loss": 4.3113, - "step": 1009 - }, - { - "epoch": 0.57, - "learning_rate": 1.8177570093457944e-05, - "loss": 4.0683, - "step": 1010 - }, - { - "epoch": 0.57, - "learning_rate": 1.8154205607476638e-05, - "loss": 4.2439, - "step": 1011 - }, - { - "epoch": 0.57, - "learning_rate": 1.8130841121495328e-05, - "loss": 4.1346, - "step": 1012 - }, - { - "epoch": 0.57, - "learning_rate": 1.8107476635514018e-05, - "loss": 4.2404, - "step": 1013 - }, - { - "epoch": 0.57, - "learning_rate": 1.808411214953271e-05, - "loss": 4.3736, - "step": 1014 - }, - { - "epoch": 0.57, - "learning_rate": 1.8060747663551404e-05, - "loss": 4.6737, - "step": 1015 - }, - { - "epoch": 0.57, - "learning_rate": 1.803738317757009e-05, - "loss": 4.2312, - "step": 1016 - }, - { - "epoch": 0.57, - "learning_rate": 1.8014018691588784e-05, - "loss": 4.2907, - "step": 1017 - }, - { - "epoch": 0.57, - "learning_rate": 1.7990654205607478e-05, - "loss": 4.1953, - "step": 1018 - }, - { - "epoch": 0.57, - "learning_rate": 1.796728971962617e-05, - "loss": 4.3023, - "step": 1019 - }, - { - "epoch": 0.57, - "learning_rate": 1.7943925233644858e-05, - "loss": 4.262, - "step": 1020 - }, - { - "epoch": 0.57, - "learning_rate": 1.792056074766355e-05, - "loss": 4.1887, - "step": 1021 - }, - { - "epoch": 0.57, - "learning_rate": 1.7897196261682245e-05, - "loss": 4.2461, - "step": 1022 - }, - { - "epoch": 0.57, - "learning_rate": 1.7873831775700935e-05, - "loss": 4.175, - "step": 1023 - }, - { - "epoch": 0.57, - "learning_rate": 1.7850467289719625e-05, - "loss": 4.2105, - "step": 1024 - }, - { - "epoch": 0.57, - "learning_rate": 1.7827102803738318e-05, - "loss": 4.245, - "step": 1025 - }, - { - "epoch": 0.58, - "learning_rate": 1.780373831775701e-05, - "loss": 3.8814, - "step": 1026 - }, - { - "epoch": 0.58, - "learning_rate": 1.77803738317757e-05, - "loss": 4.2568, - "step": 1027 - }, - { - "epoch": 0.58, - "learning_rate": 1.775700934579439e-05, - "loss": 4.4183, - "step": 1028 - }, - { - "epoch": 0.58, - "learning_rate": 1.7733644859813085e-05, - "loss": 4.1602, - "step": 1029 - }, - { - "epoch": 0.58, - "learning_rate": 1.7710280373831775e-05, - "loss": 4.2869, - "step": 1030 - }, - { - "epoch": 0.58, - "learning_rate": 1.7686915887850468e-05, - "loss": 4.2435, - "step": 1031 - }, - { - "epoch": 0.58, - "learning_rate": 1.7663551401869158e-05, - "loss": 4.066, - "step": 1032 - }, - { - "epoch": 0.58, - "learning_rate": 1.764018691588785e-05, - "loss": 4.4279, - "step": 1033 - }, - { - "epoch": 0.58, - "learning_rate": 1.761682242990654e-05, - "loss": 3.926, - "step": 1034 - }, - { - "epoch": 0.58, - "learning_rate": 1.7593457943925235e-05, - "loss": 4.1664, - "step": 1035 - }, - { - "epoch": 0.58, - "learning_rate": 1.7570093457943925e-05, - "loss": 4.3359, - "step": 1036 - }, - { - "epoch": 0.58, - "learning_rate": 1.7546728971962615e-05, - "loss": 4.2128, - "step": 1037 - }, - { - "epoch": 0.58, - "learning_rate": 1.752336448598131e-05, - "loss": 4.2249, - "step": 1038 - }, - { - "epoch": 0.58, - "learning_rate": 1.7500000000000002e-05, - "loss": 4.2303, - "step": 1039 - }, - { - "epoch": 0.58, - "learning_rate": 1.7476635514018692e-05, - "loss": 4.4989, - "step": 1040 - }, - { - "epoch": 0.58, - "learning_rate": 1.7453271028037382e-05, - "loss": 4.4136, - "step": 1041 - }, - { - "epoch": 0.58, - "learning_rate": 1.7429906542056075e-05, - "loss": 4.4061, - "step": 1042 - }, - { - "epoch": 0.58, - "learning_rate": 1.740654205607477e-05, - "loss": 4.4597, - "step": 1043 - }, - { - "epoch": 0.59, - "learning_rate": 1.738317757009346e-05, - "loss": 4.175, - "step": 1044 - }, - { - "epoch": 0.59, - "learning_rate": 1.735981308411215e-05, - "loss": 4.425, - "step": 1045 - }, - { - "epoch": 0.59, - "learning_rate": 1.7336448598130842e-05, - "loss": 4.3069, - "step": 1046 - }, - { - "epoch": 0.59, - "learning_rate": 1.7313084112149535e-05, - "loss": 4.2418, - "step": 1047 - }, - { - "epoch": 0.59, - "learning_rate": 1.7289719626168222e-05, - "loss": 4.1212, - "step": 1048 - }, - { - "epoch": 0.59, - "learning_rate": 1.7266355140186915e-05, - "loss": 4.3381, - "step": 1049 - }, - { - "epoch": 0.59, - "learning_rate": 1.724299065420561e-05, - "loss": 5.1032, - "step": 1050 - }, - { - "epoch": 0.59, - "learning_rate": 1.72196261682243e-05, - "loss": 4.6026, - "step": 1051 - }, - { - "epoch": 0.59, - "learning_rate": 1.719626168224299e-05, - "loss": 4.1127, - "step": 1052 - }, - { - "epoch": 0.59, - "learning_rate": 1.7172897196261682e-05, - "loss": 4.0776, - "step": 1053 - }, - { - "epoch": 0.59, - "learning_rate": 1.7149532710280376e-05, - "loss": 3.9369, - "step": 1054 - }, - { - "epoch": 0.59, - "learning_rate": 1.7126168224299066e-05, - "loss": 4.1443, - "step": 1055 - }, - { - "epoch": 0.59, - "learning_rate": 1.7102803738317756e-05, - "loss": 4.131, - "step": 1056 - }, - { - "epoch": 0.59, - "learning_rate": 1.707943925233645e-05, - "loss": 4.3994, - "step": 1057 - }, - { - "epoch": 0.59, - "learning_rate": 1.7056074766355142e-05, - "loss": 4.1317, - "step": 1058 - }, - { - "epoch": 0.59, - "learning_rate": 1.7032710280373833e-05, - "loss": 4.2785, - "step": 1059 - }, - { - "epoch": 0.59, - "learning_rate": 1.7009345794392523e-05, - "loss": 4.0166, - "step": 1060 - }, - { - "epoch": 0.59, - "learning_rate": 1.6985981308411216e-05, - "loss": 4.1581, - "step": 1061 - }, - { - "epoch": 0.6, - "learning_rate": 1.6962616822429906e-05, - "loss": 4.2306, - "step": 1062 - }, - { - "epoch": 0.6, - "learning_rate": 1.69392523364486e-05, - "loss": 4.5221, - "step": 1063 - }, - { - "epoch": 0.6, - "learning_rate": 1.691588785046729e-05, - "loss": 3.8312, - "step": 1064 - }, - { - "epoch": 0.6, - "learning_rate": 1.6892523364485983e-05, - "loss": 4.3187, - "step": 1065 - }, - { - "epoch": 0.6, - "learning_rate": 1.6869158878504673e-05, - "loss": 3.8238, - "step": 1066 - }, - { - "epoch": 0.6, - "learning_rate": 1.6845794392523366e-05, - "loss": 4.2842, - "step": 1067 - }, - { - "epoch": 0.6, - "learning_rate": 1.6822429906542056e-05, - "loss": 4.0101, - "step": 1068 - }, - { - "epoch": 0.6, - "learning_rate": 1.6799065420560746e-05, - "loss": 4.4147, - "step": 1069 - }, - { - "epoch": 0.6, - "learning_rate": 1.677570093457944e-05, - "loss": 3.924, - "step": 1070 - }, - { - "epoch": 0.6, - "learning_rate": 1.6752336448598133e-05, - "loss": 4.1558, - "step": 1071 - }, - { - "epoch": 0.6, - "learning_rate": 1.6728971962616823e-05, - "loss": 3.924, - "step": 1072 - }, - { - "epoch": 0.6, - "learning_rate": 1.6705607476635513e-05, - "loss": 3.9973, - "step": 1073 - }, - { - "epoch": 0.6, - "learning_rate": 1.6682242990654206e-05, - "loss": 4.3168, - "step": 1074 - }, - { - "epoch": 0.6, - "learning_rate": 1.66588785046729e-05, - "loss": 4.1304, - "step": 1075 - }, - { - "epoch": 0.6, - "learning_rate": 1.6635514018691586e-05, - "loss": 4.4585, - "step": 1076 - }, - { - "epoch": 0.6, - "learning_rate": 1.661214953271028e-05, - "loss": 4.1511, - "step": 1077 - }, - { - "epoch": 0.6, - "learning_rate": 1.6588785046728973e-05, - "loss": 4.1036, - "step": 1078 - }, - { - "epoch": 0.6, - "learning_rate": 1.6565420560747667e-05, - "loss": 4.1453, - "step": 1079 - }, - { - "epoch": 0.61, - "learning_rate": 1.6542056074766353e-05, - "loss": 4.2954, - "step": 1080 - }, - { - "epoch": 0.61, - "learning_rate": 1.6518691588785047e-05, - "loss": 4.1615, - "step": 1081 - }, - { - "epoch": 0.61, - "learning_rate": 1.649532710280374e-05, - "loss": 4.2027, - "step": 1082 - }, - { - "epoch": 0.61, - "learning_rate": 1.647196261682243e-05, - "loss": 3.9467, - "step": 1083 - }, - { - "epoch": 0.61, - "learning_rate": 1.644859813084112e-05, - "loss": 4.2254, - "step": 1084 - }, - { - "epoch": 0.61, - "learning_rate": 1.6425233644859813e-05, - "loss": 4.0942, - "step": 1085 - }, - { - "epoch": 0.61, - "learning_rate": 1.6401869158878507e-05, - "loss": 4.2585, - "step": 1086 - }, - { - "epoch": 0.61, - "learning_rate": 1.6378504672897197e-05, - "loss": 4.363, - "step": 1087 - }, - { - "epoch": 0.61, - "learning_rate": 1.6355140186915887e-05, - "loss": 4.246, - "step": 1088 - }, - { - "epoch": 0.61, - "learning_rate": 1.633177570093458e-05, - "loss": 4.3585, - "step": 1089 - }, - { - "epoch": 0.61, - "learning_rate": 1.630841121495327e-05, - "loss": 4.0166, - "step": 1090 - }, - { - "epoch": 0.61, - "learning_rate": 1.6285046728971964e-05, - "loss": 4.0657, - "step": 1091 - }, - { - "epoch": 0.61, - "learning_rate": 1.6261682242990654e-05, - "loss": 4.2611, - "step": 1092 - }, - { - "epoch": 0.61, - "learning_rate": 1.6238317757009347e-05, - "loss": 4.3694, - "step": 1093 - }, - { - "epoch": 0.61, - "learning_rate": 1.6214953271028037e-05, - "loss": 4.4972, - "step": 1094 - }, - { - "epoch": 0.61, - "learning_rate": 1.619158878504673e-05, - "loss": 4.3113, - "step": 1095 - }, - { - "epoch": 0.61, - "learning_rate": 1.616822429906542e-05, - "loss": 4.6755, - "step": 1096 - }, - { - "epoch": 0.61, - "learning_rate": 1.614485981308411e-05, - "loss": 4.4701, - "step": 1097 - }, - { - "epoch": 0.62, - "learning_rate": 1.6121495327102804e-05, - "loss": 5.0686, - "step": 1098 - }, - { - "epoch": 0.62, - "learning_rate": 1.6098130841121497e-05, - "loss": 4.3936, - "step": 1099 - }, - { - "epoch": 0.62, - "learning_rate": 1.6074766355140187e-05, - "loss": 4.7163, - "step": 1100 - }, - { - "epoch": 0.62, - "learning_rate": 1.6051401869158877e-05, - "loss": 3.8227, - "step": 1101 - }, - { - "epoch": 0.62, - "learning_rate": 1.602803738317757e-05, - "loss": 4.3251, - "step": 1102 - }, - { - "epoch": 0.62, - "learning_rate": 1.6004672897196264e-05, - "loss": 4.2298, - "step": 1103 - }, - { - "epoch": 0.62, - "learning_rate": 1.598130841121495e-05, - "loss": 3.9855, - "step": 1104 - }, - { - "epoch": 0.62, - "learning_rate": 1.5957943925233644e-05, - "loss": 3.9809, - "step": 1105 - }, - { - "epoch": 0.62, - "learning_rate": 1.5934579439252337e-05, - "loss": 4.0129, - "step": 1106 - }, - { - "epoch": 0.62, - "learning_rate": 1.591121495327103e-05, - "loss": 3.8364, - "step": 1107 - }, - { - "epoch": 0.62, - "learning_rate": 1.5887850467289717e-05, - "loss": 4.0056, - "step": 1108 - }, - { - "epoch": 0.62, - "learning_rate": 1.586448598130841e-05, - "loss": 4.2688, - "step": 1109 - }, - { - "epoch": 0.62, - "learning_rate": 1.5841121495327104e-05, - "loss": 4.2297, - "step": 1110 - }, - { - "epoch": 0.62, - "learning_rate": 1.5817757009345794e-05, - "loss": 3.9998, - "step": 1111 - }, - { - "epoch": 0.62, - "learning_rate": 1.5794392523364484e-05, - "loss": 4.3571, - "step": 1112 - }, - { - "epoch": 0.62, - "learning_rate": 1.5771028037383178e-05, - "loss": 3.7861, - "step": 1113 - }, - { - "epoch": 0.62, - "learning_rate": 1.574766355140187e-05, - "loss": 4.2606, - "step": 1114 - }, - { - "epoch": 0.62, - "learning_rate": 1.572429906542056e-05, - "loss": 4.4363, - "step": 1115 - }, - { - "epoch": 0.63, - "learning_rate": 1.570093457943925e-05, - "loss": 4.2552, - "step": 1116 - }, - { - "epoch": 0.63, - "learning_rate": 1.5677570093457944e-05, - "loss": 4.1592, - "step": 1117 - }, - { - "epoch": 0.63, - "learning_rate": 1.5654205607476634e-05, - "loss": 4.1694, - "step": 1118 - }, - { - "epoch": 0.63, - "learning_rate": 1.5630841121495328e-05, - "loss": 4.1184, - "step": 1119 - }, - { - "epoch": 0.63, - "learning_rate": 1.5607476635514018e-05, - "loss": 4.1001, - "step": 1120 - }, - { - "epoch": 0.63, - "learning_rate": 1.558411214953271e-05, - "loss": 4.214, - "step": 1121 - }, - { - "epoch": 0.63, - "learning_rate": 1.55607476635514e-05, - "loss": 4.131, - "step": 1122 - }, - { - "epoch": 0.63, - "learning_rate": 1.5537383177570095e-05, - "loss": 4.0021, - "step": 1123 - }, - { - "epoch": 0.63, - "learning_rate": 1.5514018691588785e-05, - "loss": 4.0581, - "step": 1124 - }, - { - "epoch": 0.63, - "learning_rate": 1.5490654205607475e-05, - "loss": 4.3702, - "step": 1125 - }, - { - "epoch": 0.63, - "learning_rate": 1.5467289719626168e-05, - "loss": 4.0275, - "step": 1126 - }, - { - "epoch": 0.63, - "learning_rate": 1.544392523364486e-05, - "loss": 4.0219, - "step": 1127 - }, - { - "epoch": 0.63, - "learning_rate": 1.542056074766355e-05, - "loss": 4.5626, - "step": 1128 - }, - { - "epoch": 0.63, - "learning_rate": 1.539719626168224e-05, - "loss": 3.7936, - "step": 1129 - }, - { - "epoch": 0.63, - "learning_rate": 1.5373831775700935e-05, - "loss": 4.47, - "step": 1130 - }, - { - "epoch": 0.63, - "learning_rate": 1.5350467289719628e-05, - "loss": 4.3354, - "step": 1131 - }, - { - "epoch": 0.63, - "learning_rate": 1.5327102803738315e-05, - "loss": 4.1541, - "step": 1132 - }, - { - "epoch": 0.64, - "learning_rate": 1.5303738317757008e-05, - "loss": 4.0681, - "step": 1133 - }, - { - "epoch": 0.64, - "learning_rate": 1.52803738317757e-05, - "loss": 4.2114, - "step": 1134 - }, - { - "epoch": 0.64, - "learning_rate": 1.5257009345794393e-05, - "loss": 4.4533, - "step": 1135 - }, - { - "epoch": 0.64, - "learning_rate": 1.5233644859813083e-05, - "loss": 4.1447, - "step": 1136 - }, - { - "epoch": 0.64, - "learning_rate": 1.5210280373831775e-05, - "loss": 4.2246, - "step": 1137 - }, - { - "epoch": 0.64, - "learning_rate": 1.5186915887850468e-05, - "loss": 4.3156, - "step": 1138 - }, - { - "epoch": 0.64, - "learning_rate": 1.516355140186916e-05, - "loss": 4.033, - "step": 1139 - }, - { - "epoch": 0.64, - "learning_rate": 1.5140186915887848e-05, - "loss": 4.2967, - "step": 1140 - }, - { - "epoch": 0.64, - "learning_rate": 1.5116822429906542e-05, - "loss": 4.1513, - "step": 1141 - }, - { - "epoch": 0.64, - "learning_rate": 1.5093457943925234e-05, - "loss": 4.5366, - "step": 1142 - }, - { - "epoch": 0.64, - "learning_rate": 1.5070093457943927e-05, - "loss": 4.2984, - "step": 1143 - }, - { - "epoch": 0.64, - "learning_rate": 1.5046728971962615e-05, - "loss": 4.515, - "step": 1144 - }, - { - "epoch": 0.64, - "learning_rate": 1.5023364485981309e-05, - "loss": 4.2453, - "step": 1145 - }, - { - "epoch": 0.64, - "learning_rate": 1.5e-05, - "loss": 4.3104, - "step": 1146 - }, - { - "epoch": 0.64, - "learning_rate": 1.4976635514018692e-05, - "loss": 4.2848, - "step": 1147 - }, - { - "epoch": 0.64, - "learning_rate": 1.4953271028037384e-05, - "loss": 4.132, - "step": 1148 - }, - { - "epoch": 0.64, - "learning_rate": 1.4929906542056075e-05, - "loss": 4.3078, - "step": 1149 - }, - { - "epoch": 0.64, - "learning_rate": 1.4906542056074767e-05, - "loss": 4.1989, - "step": 1150 - }, - { - "epoch": 0.65, - "learning_rate": 1.4883177570093459e-05, - "loss": 4.1369, - "step": 1151 - }, - { - "epoch": 0.65, - "learning_rate": 1.485981308411215e-05, - "loss": 4.2523, - "step": 1152 - }, - { - "epoch": 0.65, - "learning_rate": 1.4836448598130842e-05, - "loss": 4.2366, - "step": 1153 - }, - { - "epoch": 0.65, - "learning_rate": 1.4813084112149532e-05, - "loss": 4.2827, - "step": 1154 - }, - { - "epoch": 0.65, - "learning_rate": 1.4789719626168226e-05, - "loss": 4.2893, - "step": 1155 - }, - { - "epoch": 0.65, - "learning_rate": 1.4766355140186916e-05, - "loss": 4.2778, - "step": 1156 - }, - { - "epoch": 0.65, - "learning_rate": 1.4742990654205609e-05, - "loss": 4.1089, - "step": 1157 - }, - { - "epoch": 0.65, - "learning_rate": 1.4719626168224299e-05, - "loss": 4.1321, - "step": 1158 - }, - { - "epoch": 0.65, - "learning_rate": 1.4696261682242993e-05, - "loss": 4.0284, - "step": 1159 - }, - { - "epoch": 0.65, - "learning_rate": 1.4672897196261683e-05, - "loss": 3.9388, - "step": 1160 - }, - { - "epoch": 0.65, - "learning_rate": 1.4649532710280374e-05, - "loss": 4.3155, - "step": 1161 - }, - { - "epoch": 0.65, - "learning_rate": 1.4626168224299066e-05, - "loss": 4.1251, - "step": 1162 - }, - { - "epoch": 0.65, - "learning_rate": 1.4602803738317758e-05, - "loss": 4.1858, - "step": 1163 - }, - { - "epoch": 0.65, - "learning_rate": 1.457943925233645e-05, - "loss": 4.0236, - "step": 1164 - }, - { - "epoch": 0.65, - "learning_rate": 1.4556074766355141e-05, - "loss": 4.0506, - "step": 1165 - }, - { - "epoch": 0.65, - "learning_rate": 1.4532710280373833e-05, - "loss": 4.2977, - "step": 1166 - }, - { - "epoch": 0.65, - "learning_rate": 1.4509345794392524e-05, - "loss": 4.0952, - "step": 1167 - }, - { - "epoch": 0.65, - "learning_rate": 1.4485981308411214e-05, - "loss": 3.9324, - "step": 1168 - }, - { - "epoch": 0.66, - "learning_rate": 1.4462616822429908e-05, - "loss": 4.2742, - "step": 1169 - }, - { - "epoch": 0.66, - "learning_rate": 1.4439252336448598e-05, - "loss": 4.3295, - "step": 1170 - }, - { - "epoch": 0.66, - "learning_rate": 1.4415887850467291e-05, - "loss": 3.9502, - "step": 1171 - }, - { - "epoch": 0.66, - "learning_rate": 1.4392523364485981e-05, - "loss": 3.9991, - "step": 1172 - }, - { - "epoch": 0.66, - "learning_rate": 1.4369158878504675e-05, - "loss": 4.3436, - "step": 1173 - }, - { - "epoch": 0.66, - "learning_rate": 1.4345794392523365e-05, - "loss": 4.0077, - "step": 1174 - }, - { - "epoch": 0.66, - "learning_rate": 1.4322429906542058e-05, - "loss": 3.9805, - "step": 1175 - }, - { - "epoch": 0.66, - "learning_rate": 1.4299065420560748e-05, - "loss": 4.1969, - "step": 1176 - }, - { - "epoch": 0.66, - "learning_rate": 1.427570093457944e-05, - "loss": 4.2658, - "step": 1177 - }, - { - "epoch": 0.66, - "learning_rate": 1.4252336448598131e-05, - "loss": 4.2669, - "step": 1178 - }, - { - "epoch": 0.66, - "learning_rate": 1.4228971962616823e-05, - "loss": 3.7971, - "step": 1179 - }, - { - "epoch": 0.66, - "learning_rate": 1.4205607476635515e-05, - "loss": 4.1124, - "step": 1180 - }, - { - "epoch": 0.66, - "learning_rate": 1.4182242990654207e-05, - "loss": 4.0168, - "step": 1181 - }, - { - "epoch": 0.66, - "learning_rate": 1.4158878504672898e-05, - "loss": 4.1873, - "step": 1182 - }, - { - "epoch": 0.66, - "learning_rate": 1.413551401869159e-05, - "loss": 3.976, - "step": 1183 - }, - { - "epoch": 0.66, - "learning_rate": 1.411214953271028e-05, - "loss": 4.0415, - "step": 1184 - }, - { - "epoch": 0.66, - "learning_rate": 1.4088785046728973e-05, - "loss": 4.5128, - "step": 1185 - }, - { - "epoch": 0.66, - "learning_rate": 1.4065420560747663e-05, - "loss": 3.9412, - "step": 1186 - }, - { - "epoch": 0.67, - "learning_rate": 1.4042056074766357e-05, - "loss": 4.4509, - "step": 1187 - }, - { - "epoch": 0.67, - "learning_rate": 1.4018691588785047e-05, - "loss": 4.0894, - "step": 1188 - }, - { - "epoch": 0.67, - "learning_rate": 1.399532710280374e-05, - "loss": 3.9693, - "step": 1189 - }, - { - "epoch": 0.67, - "learning_rate": 1.397196261682243e-05, - "loss": 4.1943, - "step": 1190 - }, - { - "epoch": 0.67, - "learning_rate": 1.3948598130841122e-05, - "loss": 4.2143, - "step": 1191 - }, - { - "epoch": 0.67, - "learning_rate": 1.3925233644859814e-05, - "loss": 4.3757, - "step": 1192 - }, - { - "epoch": 0.67, - "learning_rate": 1.3901869158878505e-05, - "loss": 4.6904, - "step": 1193 - }, - { - "epoch": 0.67, - "learning_rate": 1.3878504672897197e-05, - "loss": 4.1949, - "step": 1194 - }, - { - "epoch": 0.67, - "learning_rate": 1.3855140186915889e-05, - "loss": 4.4187, - "step": 1195 - }, - { - "epoch": 0.67, - "learning_rate": 1.383177570093458e-05, - "loss": 4.0573, - "step": 1196 - }, - { - "epoch": 0.67, - "learning_rate": 1.3808411214953272e-05, - "loss": 4.0778, - "step": 1197 - }, - { - "epoch": 0.67, - "learning_rate": 1.3785046728971962e-05, - "loss": 4.4602, - "step": 1198 - }, - { - "epoch": 0.67, - "learning_rate": 1.3761682242990655e-05, - "loss": 4.364, - "step": 1199 - }, - { - "epoch": 0.67, - "learning_rate": 1.3738317757009345e-05, - "loss": 4.0137, - "step": 1200 - }, - { - "epoch": 0.67, - "learning_rate": 1.3714953271028039e-05, - "loss": 4.063, - "step": 1201 - }, - { - "epoch": 0.67, - "learning_rate": 1.3691588785046729e-05, - "loss": 4.0722, - "step": 1202 - }, - { - "epoch": 0.67, - "learning_rate": 1.3668224299065422e-05, - "loss": 4.0259, - "step": 1203 - }, - { - "epoch": 0.67, - "learning_rate": 1.3644859813084112e-05, - "loss": 3.9311, - "step": 1204 - }, - { - "epoch": 0.68, - "learning_rate": 1.3621495327102804e-05, - "loss": 3.8923, - "step": 1205 - }, - { - "epoch": 0.68, - "learning_rate": 1.3598130841121496e-05, - "loss": 4.2806, - "step": 1206 - }, - { - "epoch": 0.68, - "learning_rate": 1.3574766355140187e-05, - "loss": 4.4994, - "step": 1207 - }, - { - "epoch": 0.68, - "learning_rate": 1.3551401869158879e-05, - "loss": 4.1264, - "step": 1208 - }, - { - "epoch": 0.68, - "learning_rate": 1.352803738317757e-05, - "loss": 4.0846, - "step": 1209 - }, - { - "epoch": 0.68, - "learning_rate": 1.3504672897196263e-05, - "loss": 3.9229, - "step": 1210 - }, - { - "epoch": 0.68, - "learning_rate": 1.3481308411214954e-05, - "loss": 4.2628, - "step": 1211 - }, - { - "epoch": 0.68, - "learning_rate": 1.3457943925233644e-05, - "loss": 4.3152, - "step": 1212 - }, - { - "epoch": 0.68, - "learning_rate": 1.3434579439252338e-05, - "loss": 4.2379, - "step": 1213 - }, - { - "epoch": 0.68, - "learning_rate": 1.3411214953271028e-05, - "loss": 4.0713, - "step": 1214 - }, - { - "epoch": 0.68, - "learning_rate": 1.3387850467289721e-05, - "loss": 4.1032, - "step": 1215 - }, - { - "epoch": 0.68, - "learning_rate": 1.3364485981308411e-05, - "loss": 4.1874, - "step": 1216 - }, - { - "epoch": 0.68, - "learning_rate": 1.3341121495327104e-05, - "loss": 4.0298, - "step": 1217 - }, - { - "epoch": 0.68, - "learning_rate": 1.3317757009345794e-05, - "loss": 3.9534, - "step": 1218 - }, - { - "epoch": 0.68, - "learning_rate": 1.3294392523364486e-05, - "loss": 3.9305, - "step": 1219 - }, - { - "epoch": 0.68, - "learning_rate": 1.3271028037383178e-05, - "loss": 4.2816, - "step": 1220 - }, - { - "epoch": 0.68, - "learning_rate": 1.324766355140187e-05, - "loss": 4.1107, - "step": 1221 - }, - { - "epoch": 0.68, - "learning_rate": 1.3224299065420561e-05, - "loss": 3.8697, - "step": 1222 - }, - { - "epoch": 0.69, - "learning_rate": 1.3200934579439253e-05, - "loss": 4.2744, - "step": 1223 - }, - { - "epoch": 0.69, - "learning_rate": 1.3177570093457945e-05, - "loss": 4.1552, - "step": 1224 - }, - { - "epoch": 0.69, - "learning_rate": 1.3154205607476636e-05, - "loss": 4.2383, - "step": 1225 - }, - { - "epoch": 0.69, - "learning_rate": 1.3130841121495328e-05, - "loss": 4.2343, - "step": 1226 - }, - { - "epoch": 0.69, - "learning_rate": 1.310747663551402e-05, - "loss": 4.0618, - "step": 1227 - }, - { - "epoch": 0.69, - "learning_rate": 1.308411214953271e-05, - "loss": 4.0713, - "step": 1228 - }, - { - "epoch": 0.69, - "learning_rate": 1.3060747663551403e-05, - "loss": 4.0311, - "step": 1229 - }, - { - "epoch": 0.69, - "learning_rate": 1.3037383177570093e-05, - "loss": 4.2391, - "step": 1230 - }, - { - "epoch": 0.69, - "learning_rate": 1.3014018691588787e-05, - "loss": 4.1531, - "step": 1231 - }, - { - "epoch": 0.69, - "learning_rate": 1.2990654205607477e-05, - "loss": 4.0725, - "step": 1232 - }, - { - "epoch": 0.69, - "learning_rate": 1.296728971962617e-05, - "loss": 4.0772, - "step": 1233 - }, - { - "epoch": 0.69, - "learning_rate": 1.294392523364486e-05, - "loss": 4.0475, - "step": 1234 - }, - { - "epoch": 0.69, - "learning_rate": 1.2920560747663552e-05, - "loss": 4.4437, - "step": 1235 - }, - { - "epoch": 0.69, - "learning_rate": 1.2897196261682243e-05, - "loss": 4.154, - "step": 1236 - }, - { - "epoch": 0.69, - "learning_rate": 1.2873831775700935e-05, - "loss": 4.0849, - "step": 1237 - }, - { - "epoch": 0.69, - "learning_rate": 1.2850467289719627e-05, - "loss": 4.0024, - "step": 1238 - }, - { - "epoch": 0.69, - "learning_rate": 1.2827102803738318e-05, - "loss": 4.1889, - "step": 1239 - }, - { - "epoch": 0.7, - "learning_rate": 1.280373831775701e-05, - "loss": 4.2619, - "step": 1240 - }, - { - "epoch": 0.7, - "learning_rate": 1.2780373831775702e-05, - "loss": 4.2038, - "step": 1241 - }, - { - "epoch": 0.7, - "learning_rate": 1.2757009345794392e-05, - "loss": 4.5605, - "step": 1242 - }, - { - "epoch": 0.7, - "learning_rate": 1.2733644859813085e-05, - "loss": 4.277, - "step": 1243 - }, - { - "epoch": 0.7, - "learning_rate": 1.2710280373831775e-05, - "loss": 4.2266, - "step": 1244 - }, - { - "epoch": 0.7, - "learning_rate": 1.2686915887850469e-05, - "loss": 3.9489, - "step": 1245 - }, - { - "epoch": 0.7, - "learning_rate": 1.2663551401869159e-05, - "loss": 4.3945, - "step": 1246 - }, - { - "epoch": 0.7, - "learning_rate": 1.2640186915887852e-05, - "loss": 4.4063, - "step": 1247 - }, - { - "epoch": 0.7, - "learning_rate": 1.2616822429906542e-05, - "loss": 4.1918, - "step": 1248 - }, - { - "epoch": 0.7, - "learning_rate": 1.2593457943925234e-05, - "loss": 4.4201, - "step": 1249 - }, - { - "epoch": 0.7, - "learning_rate": 1.2570093457943925e-05, - "loss": 4.5105, - "step": 1250 - }, - { - "epoch": 0.7, - "learning_rate": 1.2546728971962617e-05, - "loss": 4.2447, - "step": 1251 - }, - { - "epoch": 0.7, - "learning_rate": 1.2523364485981309e-05, - "loss": 4.1262, - "step": 1252 - }, - { - "epoch": 0.7, - "learning_rate": 1.25e-05, - "loss": 4.1989, - "step": 1253 - }, - { - "epoch": 0.7, - "learning_rate": 1.2476635514018692e-05, - "loss": 4.292, - "step": 1254 - }, - { - "epoch": 0.7, - "learning_rate": 1.2453271028037384e-05, - "loss": 4.3866, - "step": 1255 - }, - { - "epoch": 0.7, - "learning_rate": 1.2429906542056074e-05, - "loss": 4.0586, - "step": 1256 - }, - { - "epoch": 0.7, - "learning_rate": 1.2406542056074767e-05, - "loss": 4.0311, - "step": 1257 - }, - { - "epoch": 0.71, - "learning_rate": 1.2383177570093457e-05, - "loss": 3.9164, - "step": 1258 - }, - { - "epoch": 0.71, - "learning_rate": 1.235981308411215e-05, - "loss": 4.1645, - "step": 1259 - }, - { - "epoch": 0.71, - "learning_rate": 1.233644859813084e-05, - "loss": 4.4957, - "step": 1260 - }, - { - "epoch": 0.71, - "learning_rate": 1.2313084112149534e-05, - "loss": 3.9794, - "step": 1261 - }, - { - "epoch": 0.71, - "learning_rate": 1.2289719626168224e-05, - "loss": 4.2441, - "step": 1262 - }, - { - "epoch": 0.71, - "learning_rate": 1.2266355140186916e-05, - "loss": 4.2401, - "step": 1263 - }, - { - "epoch": 0.71, - "learning_rate": 1.2242990654205608e-05, - "loss": 4.1159, - "step": 1264 - }, - { - "epoch": 0.71, - "learning_rate": 1.22196261682243e-05, - "loss": 4.317, - "step": 1265 - }, - { - "epoch": 0.71, - "learning_rate": 1.2196261682242991e-05, - "loss": 4.2399, - "step": 1266 - }, - { - "epoch": 0.71, - "learning_rate": 1.2172897196261683e-05, - "loss": 4.0619, - "step": 1267 - }, - { - "epoch": 0.71, - "learning_rate": 1.2149532710280374e-05, - "loss": 4.0484, - "step": 1268 - }, - { - "epoch": 0.71, - "learning_rate": 1.2126168224299066e-05, - "loss": 4.2496, - "step": 1269 - }, - { - "epoch": 0.71, - "learning_rate": 1.2102803738317758e-05, - "loss": 4.0288, - "step": 1270 - }, - { - "epoch": 0.71, - "learning_rate": 1.207943925233645e-05, - "loss": 4.0943, - "step": 1271 - }, - { - "epoch": 0.71, - "learning_rate": 1.205607476635514e-05, - "loss": 4.1295, - "step": 1272 - }, - { - "epoch": 0.71, - "learning_rate": 1.2032710280373833e-05, - "loss": 4.3845, - "step": 1273 - }, - { - "epoch": 0.71, - "learning_rate": 1.2009345794392523e-05, - "loss": 4.0699, - "step": 1274 - }, - { - "epoch": 0.71, - "learning_rate": 1.1985981308411216e-05, - "loss": 4.1361, - "step": 1275 - }, - { - "epoch": 0.72, - "learning_rate": 1.1962616822429906e-05, - "loss": 4.4625, - "step": 1276 - }, - { - "epoch": 0.72, - "learning_rate": 1.19392523364486e-05, - "loss": 4.2566, - "step": 1277 - }, - { - "epoch": 0.72, - "learning_rate": 1.191588785046729e-05, - "loss": 4.1578, - "step": 1278 - }, - { - "epoch": 0.72, - "learning_rate": 1.1892523364485981e-05, - "loss": 4.3413, - "step": 1279 - }, - { - "epoch": 0.72, - "learning_rate": 1.1869158878504673e-05, - "loss": 4.28, - "step": 1280 - }, - { - "epoch": 0.72, - "learning_rate": 1.1845794392523365e-05, - "loss": 4.2081, - "step": 1281 - }, - { - "epoch": 0.72, - "learning_rate": 1.1822429906542057e-05, - "loss": 4.0978, - "step": 1282 - }, - { - "epoch": 0.72, - "learning_rate": 1.1799065420560748e-05, - "loss": 4.3259, - "step": 1283 - }, - { - "epoch": 0.72, - "learning_rate": 1.177570093457944e-05, - "loss": 4.1025, - "step": 1284 - }, - { - "epoch": 0.72, - "learning_rate": 1.1752336448598132e-05, - "loss": 4.1345, - "step": 1285 - }, - { - "epoch": 0.72, - "learning_rate": 1.1728971962616822e-05, - "loss": 4.2033, - "step": 1286 - }, - { - "epoch": 0.72, - "learning_rate": 1.1705607476635515e-05, - "loss": 4.1905, - "step": 1287 - }, - { - "epoch": 0.72, - "learning_rate": 1.1682242990654205e-05, - "loss": 4.1205, - "step": 1288 - }, - { - "epoch": 0.72, - "learning_rate": 1.1658878504672898e-05, - "loss": 4.1419, - "step": 1289 - }, - { - "epoch": 0.72, - "learning_rate": 1.1635514018691588e-05, - "loss": 3.8481, - "step": 1290 - }, - { - "epoch": 0.72, - "learning_rate": 1.1612149532710282e-05, - "loss": 4.2781, - "step": 1291 - }, - { - "epoch": 0.72, - "learning_rate": 1.1588785046728972e-05, - "loss": 4.3232, - "step": 1292 - }, - { - "epoch": 0.72, - "learning_rate": 1.1565420560747664e-05, - "loss": 4.6789, - "step": 1293 - }, - { - "epoch": 0.73, - "learning_rate": 1.1542056074766355e-05, - "loss": 4.0168, - "step": 1294 - }, - { - "epoch": 0.73, - "learning_rate": 1.1518691588785047e-05, - "loss": 4.0139, - "step": 1295 - }, - { - "epoch": 0.73, - "learning_rate": 1.1495327102803739e-05, - "loss": 3.9468, - "step": 1296 - }, - { - "epoch": 0.73, - "learning_rate": 1.147196261682243e-05, - "loss": 4.0327, - "step": 1297 - }, - { - "epoch": 0.73, - "learning_rate": 1.1448598130841122e-05, - "loss": 3.8068, - "step": 1298 - }, - { - "epoch": 0.73, - "learning_rate": 1.1425233644859814e-05, - "loss": 4.6063, - "step": 1299 - }, - { - "epoch": 0.73, - "learning_rate": 1.1401869158878504e-05, - "loss": 3.9934, - "step": 1300 - }, - { - "epoch": 0.73, - "learning_rate": 1.1378504672897197e-05, - "loss": 4.2126, - "step": 1301 - }, - { - "epoch": 0.73, - "learning_rate": 1.1355140186915887e-05, - "loss": 4.3675, - "step": 1302 - }, - { - "epoch": 0.73, - "learning_rate": 1.133177570093458e-05, - "loss": 4.1626, - "step": 1303 - }, - { - "epoch": 0.73, - "learning_rate": 1.130841121495327e-05, - "loss": 4.1996, - "step": 1304 - }, - { - "epoch": 0.73, - "learning_rate": 1.1285046728971964e-05, - "loss": 4.0115, - "step": 1305 - }, - { - "epoch": 0.73, - "learning_rate": 1.1261682242990654e-05, - "loss": 4.1536, - "step": 1306 - }, - { - "epoch": 0.73, - "learning_rate": 1.1238317757009346e-05, - "loss": 4.0911, - "step": 1307 - }, - { - "epoch": 0.73, - "learning_rate": 1.1214953271028037e-05, - "loss": 3.9302, - "step": 1308 - }, - { - "epoch": 0.73, - "learning_rate": 1.1191588785046729e-05, - "loss": 4.1187, - "step": 1309 - }, - { - "epoch": 0.73, - "learning_rate": 1.116822429906542e-05, - "loss": 3.9579, - "step": 1310 - }, - { - "epoch": 0.73, - "learning_rate": 1.1144859813084113e-05, - "loss": 4.0662, - "step": 1311 - }, - { - "epoch": 0.74, - "learning_rate": 1.1121495327102804e-05, - "loss": 4.2141, - "step": 1312 - }, - { - "epoch": 0.74, - "learning_rate": 1.1098130841121496e-05, - "loss": 4.1037, - "step": 1313 - }, - { - "epoch": 0.74, - "learning_rate": 1.1074766355140186e-05, - "loss": 3.9829, - "step": 1314 - }, - { - "epoch": 0.74, - "learning_rate": 1.105140186915888e-05, - "loss": 4.3324, - "step": 1315 - }, - { - "epoch": 0.74, - "learning_rate": 1.102803738317757e-05, - "loss": 4.1306, - "step": 1316 - }, - { - "epoch": 0.74, - "learning_rate": 1.1004672897196263e-05, - "loss": 4.0159, - "step": 1317 - }, - { - "epoch": 0.74, - "learning_rate": 1.0981308411214953e-05, - "loss": 4.567, - "step": 1318 - }, - { - "epoch": 0.74, - "learning_rate": 1.0957943925233646e-05, - "loss": 4.0622, - "step": 1319 - }, - { - "epoch": 0.74, - "learning_rate": 1.0934579439252336e-05, - "loss": 3.8993, - "step": 1320 - }, - { - "epoch": 0.74, - "learning_rate": 1.091121495327103e-05, - "loss": 3.8566, - "step": 1321 - }, - { - "epoch": 0.74, - "learning_rate": 1.088785046728972e-05, - "loss": 4.09, - "step": 1322 - }, - { - "epoch": 0.74, - "learning_rate": 1.0864485981308411e-05, - "loss": 4.0313, - "step": 1323 - }, - { - "epoch": 0.74, - "learning_rate": 1.0841121495327103e-05, - "loss": 4.0637, - "step": 1324 - }, - { - "epoch": 0.74, - "learning_rate": 1.0817757009345795e-05, - "loss": 4.1181, - "step": 1325 - }, - { - "epoch": 0.74, - "learning_rate": 1.0794392523364486e-05, - "loss": 3.9657, - "step": 1326 - }, - { - "epoch": 0.74, - "learning_rate": 1.0771028037383178e-05, - "loss": 3.8988, - "step": 1327 - }, - { - "epoch": 0.74, - "learning_rate": 1.074766355140187e-05, - "loss": 4.0866, - "step": 1328 - }, - { - "epoch": 0.74, - "learning_rate": 1.0724299065420561e-05, - "loss": 4.2302, - "step": 1329 - }, - { - "epoch": 0.75, - "learning_rate": 1.0700934579439251e-05, - "loss": 4.2435, - "step": 1330 - }, - { - "epoch": 0.75, - "learning_rate": 1.0677570093457945e-05, - "loss": 3.9233, - "step": 1331 - }, - { - "epoch": 0.75, - "learning_rate": 1.0654205607476635e-05, - "loss": 4.1559, - "step": 1332 - }, - { - "epoch": 0.75, - "learning_rate": 1.0630841121495328e-05, - "loss": 4.1301, - "step": 1333 - }, - { - "epoch": 0.75, - "learning_rate": 1.0607476635514018e-05, - "loss": 4.2697, - "step": 1334 - }, - { - "epoch": 0.75, - "learning_rate": 1.0584112149532712e-05, - "loss": 3.9397, - "step": 1335 - }, - { - "epoch": 0.75, - "learning_rate": 1.0560747663551402e-05, - "loss": 4.0593, - "step": 1336 - }, - { - "epoch": 0.75, - "learning_rate": 1.0537383177570093e-05, - "loss": 4.2725, - "step": 1337 - }, - { - "epoch": 0.75, - "learning_rate": 1.0514018691588785e-05, - "loss": 4.2991, - "step": 1338 - }, - { - "epoch": 0.75, - "learning_rate": 1.0490654205607477e-05, - "loss": 4.4362, - "step": 1339 - }, - { - "epoch": 0.75, - "learning_rate": 1.0467289719626168e-05, - "loss": 4.5216, - "step": 1340 - }, - { - "epoch": 0.75, - "learning_rate": 1.044392523364486e-05, - "loss": 4.3957, - "step": 1341 - }, - { - "epoch": 0.75, - "learning_rate": 1.0420560747663552e-05, - "loss": 4.1009, - "step": 1342 - }, - { - "epoch": 0.75, - "learning_rate": 1.0397196261682244e-05, - "loss": 4.3266, - "step": 1343 - }, - { - "epoch": 0.75, - "learning_rate": 1.0373831775700934e-05, - "loss": 4.087, - "step": 1344 - }, - { - "epoch": 0.75, - "learning_rate": 1.0350467289719627e-05, - "loss": 4.2562, - "step": 1345 - }, - { - "epoch": 0.75, - "learning_rate": 1.0327102803738317e-05, - "loss": 4.2976, - "step": 1346 - }, - { - "epoch": 0.76, - "learning_rate": 1.030373831775701e-05, - "loss": 4.6248, - "step": 1347 - }, - { - "epoch": 0.76, - "learning_rate": 1.02803738317757e-05, - "loss": 4.1296, - "step": 1348 - }, - { - "epoch": 0.76, - "learning_rate": 1.0257009345794394e-05, - "loss": 4.2328, - "step": 1349 - }, - { - "epoch": 0.76, - "learning_rate": 1.0233644859813084e-05, - "loss": 4.5745, - "step": 1350 - }, - { - "epoch": 0.76, - "learning_rate": 1.0210280373831776e-05, - "loss": 4.3112, - "step": 1351 - }, - { - "epoch": 0.76, - "learning_rate": 1.0186915887850467e-05, - "loss": 4.2557, - "step": 1352 - }, - { - "epoch": 0.76, - "learning_rate": 1.0163551401869159e-05, - "loss": 4.2107, - "step": 1353 - }, - { - "epoch": 0.76, - "learning_rate": 1.014018691588785e-05, - "loss": 4.2676, - "step": 1354 - }, - { - "epoch": 0.76, - "learning_rate": 1.0116822429906542e-05, - "loss": 4.0928, - "step": 1355 - }, - { - "epoch": 0.76, - "learning_rate": 1.0093457943925234e-05, - "loss": 4.4405, - "step": 1356 - }, - { - "epoch": 0.76, - "learning_rate": 1.0070093457943926e-05, - "loss": 3.9069, - "step": 1357 - }, - { - "epoch": 0.76, - "learning_rate": 1.0046728971962616e-05, - "loss": 4.2085, - "step": 1358 - }, - { - "epoch": 0.76, - "learning_rate": 1.0023364485981309e-05, - "loss": 4.364, - "step": 1359 - }, - { - "epoch": 0.76, - "learning_rate": 9.999999999999999e-06, - "loss": 4.6242, - "step": 1360 - }, - { - "epoch": 0.76, - "learning_rate": 9.976635514018693e-06, - "loss": 4.115, - "step": 1361 - }, - { - "epoch": 0.76, - "learning_rate": 9.953271028037383e-06, - "loss": 3.9853, - "step": 1362 - }, - { - "epoch": 0.76, - "learning_rate": 9.929906542056076e-06, - "loss": 4.0141, - "step": 1363 - }, - { - "epoch": 0.76, - "learning_rate": 9.906542056074766e-06, - "loss": 4.4776, - "step": 1364 - }, - { - "epoch": 0.77, - "learning_rate": 9.883177570093458e-06, - "loss": 4.1173, - "step": 1365 - }, - { - "epoch": 0.77, - "learning_rate": 9.85981308411215e-06, - "loss": 3.8906, - "step": 1366 - }, - { - "epoch": 0.77, - "learning_rate": 9.836448598130841e-06, - "loss": 4.1557, - "step": 1367 - }, - { - "epoch": 0.77, - "learning_rate": 9.813084112149533e-06, - "loss": 3.9848, - "step": 1368 - }, - { - "epoch": 0.77, - "learning_rate": 9.789719626168224e-06, - "loss": 4.3488, - "step": 1369 - }, - { - "epoch": 0.77, - "learning_rate": 9.766355140186916e-06, - "loss": 4.376, - "step": 1370 - }, - { - "epoch": 0.77, - "learning_rate": 9.742990654205608e-06, - "loss": 4.28, - "step": 1371 - }, - { - "epoch": 0.77, - "learning_rate": 9.7196261682243e-06, - "loss": 3.9477, - "step": 1372 - }, - { - "epoch": 0.77, - "learning_rate": 9.696261682242991e-06, - "loss": 4.1401, - "step": 1373 - }, - { - "epoch": 0.77, - "learning_rate": 9.672897196261681e-06, - "loss": 4.2706, - "step": 1374 - }, - { - "epoch": 0.77, - "learning_rate": 9.649532710280375e-06, - "loss": 3.8855, - "step": 1375 - }, - { - "epoch": 0.77, - "learning_rate": 9.626168224299065e-06, - "loss": 3.8925, - "step": 1376 - }, - { - "epoch": 0.77, - "learning_rate": 9.602803738317758e-06, - "loss": 4.0841, - "step": 1377 - }, - { - "epoch": 0.77, - "learning_rate": 9.579439252336448e-06, - "loss": 4.1056, - "step": 1378 - }, - { - "epoch": 0.77, - "learning_rate": 9.556074766355141e-06, - "loss": 4.1274, - "step": 1379 - }, - { - "epoch": 0.77, - "learning_rate": 9.532710280373831e-06, - "loss": 4.0996, - "step": 1380 - }, - { - "epoch": 0.77, - "learning_rate": 9.509345794392523e-06, - "loss": 3.7845, - "step": 1381 - }, - { - "epoch": 0.77, - "learning_rate": 9.485981308411215e-06, - "loss": 4.3562, - "step": 1382 - }, - { - "epoch": 0.78, - "learning_rate": 9.462616822429907e-06, - "loss": 4.0043, - "step": 1383 - }, - { - "epoch": 0.78, - "learning_rate": 9.439252336448598e-06, - "loss": 4.2826, - "step": 1384 - }, - { - "epoch": 0.78, - "learning_rate": 9.41588785046729e-06, - "loss": 4.442, - "step": 1385 - }, - { - "epoch": 0.78, - "learning_rate": 9.392523364485982e-06, - "loss": 4.2274, - "step": 1386 - }, - { - "epoch": 0.78, - "learning_rate": 9.369158878504673e-06, - "loss": 4.2576, - "step": 1387 - }, - { - "epoch": 0.78, - "learning_rate": 9.345794392523363e-06, - "loss": 4.2564, - "step": 1388 - }, - { - "epoch": 0.78, - "learning_rate": 9.322429906542057e-06, - "loss": 4.1399, - "step": 1389 - }, - { - "epoch": 0.78, - "learning_rate": 9.299065420560747e-06, - "loss": 4.205, - "step": 1390 - }, - { - "epoch": 0.78, - "learning_rate": 9.27570093457944e-06, - "loss": 3.9346, - "step": 1391 - }, - { - "epoch": 0.78, - "learning_rate": 9.25233644859813e-06, - "loss": 4.0359, - "step": 1392 - }, - { - "epoch": 0.78, - "learning_rate": 9.228971962616824e-06, - "loss": 4.2414, - "step": 1393 - }, - { - "epoch": 0.78, - "learning_rate": 9.205607476635514e-06, - "loss": 4.3723, - "step": 1394 - }, - { - "epoch": 0.78, - "learning_rate": 9.182242990654205e-06, - "loss": 4.2651, - "step": 1395 - }, - { - "epoch": 0.78, - "learning_rate": 9.158878504672897e-06, - "loss": 4.1683, - "step": 1396 - }, - { - "epoch": 0.78, - "learning_rate": 9.135514018691589e-06, - "loss": 4.34, - "step": 1397 - }, - { - "epoch": 0.78, - "learning_rate": 9.11214953271028e-06, - "loss": 4.1625, - "step": 1398 - }, - { - "epoch": 0.78, - "learning_rate": 9.088785046728972e-06, - "loss": 4.3213, - "step": 1399 - }, - { - "epoch": 0.78, - "learning_rate": 9.065420560747664e-06, - "loss": 4.4264, - "step": 1400 - }, - { - "epoch": 0.79, - "learning_rate": 9.042056074766356e-06, - "loss": 3.9775, - "step": 1401 - }, - { - "epoch": 0.79, - "learning_rate": 9.018691588785046e-06, - "loss": 4.1745, - "step": 1402 - }, - { - "epoch": 0.79, - "learning_rate": 8.995327102803739e-06, - "loss": 4.0301, - "step": 1403 - }, - { - "epoch": 0.79, - "learning_rate": 8.971962616822429e-06, - "loss": 3.9575, - "step": 1404 - }, - { - "epoch": 0.79, - "learning_rate": 8.948598130841122e-06, - "loss": 3.9296, - "step": 1405 - }, - { - "epoch": 0.79, - "learning_rate": 8.925233644859812e-06, - "loss": 4.4694, - "step": 1406 - }, - { - "epoch": 0.79, - "learning_rate": 8.901869158878506e-06, - "loss": 4.2067, - "step": 1407 - }, - { - "epoch": 0.79, - "learning_rate": 8.878504672897196e-06, - "loss": 3.9505, - "step": 1408 - }, - { - "epoch": 0.79, - "learning_rate": 8.855140186915887e-06, - "loss": 4.0672, - "step": 1409 - }, - { - "epoch": 0.79, - "learning_rate": 8.831775700934579e-06, - "loss": 3.8908, - "step": 1410 - }, - { - "epoch": 0.79, - "learning_rate": 8.80841121495327e-06, - "loss": 3.9355, - "step": 1411 - }, - { - "epoch": 0.79, - "learning_rate": 8.785046728971963e-06, - "loss": 3.9802, - "step": 1412 - }, - { - "epoch": 0.79, - "learning_rate": 8.761682242990654e-06, - "loss": 4.402, - "step": 1413 - }, - { - "epoch": 0.79, - "learning_rate": 8.738317757009346e-06, - "loss": 4.2815, - "step": 1414 - }, - { - "epoch": 0.79, - "learning_rate": 8.714953271028038e-06, - "loss": 4.2071, - "step": 1415 - }, - { - "epoch": 0.79, - "learning_rate": 8.69158878504673e-06, - "loss": 4.3293, - "step": 1416 - }, - { - "epoch": 0.79, - "learning_rate": 8.668224299065421e-06, - "loss": 4.1243, - "step": 1417 - }, - { - "epoch": 0.79, - "learning_rate": 8.644859813084111e-06, - "loss": 4.1596, - "step": 1418 - }, - { - "epoch": 0.8, - "learning_rate": 8.621495327102804e-06, - "loss": 4.1897, - "step": 1419 - }, - { - "epoch": 0.8, - "learning_rate": 8.598130841121494e-06, - "loss": 3.9652, - "step": 1420 - }, - { - "epoch": 0.8, - "learning_rate": 8.574766355140188e-06, - "loss": 4.1442, - "step": 1421 - }, - { - "epoch": 0.8, - "learning_rate": 8.551401869158878e-06, - "loss": 4.2353, - "step": 1422 - }, - { - "epoch": 0.8, - "learning_rate": 8.528037383177571e-06, - "loss": 4.1577, - "step": 1423 - }, - { - "epoch": 0.8, - "learning_rate": 8.504672897196261e-06, - "loss": 4.0383, - "step": 1424 - }, - { - "epoch": 0.8, - "learning_rate": 8.481308411214953e-06, - "loss": 4.2949, - "step": 1425 - }, - { - "epoch": 0.8, - "learning_rate": 8.457943925233645e-06, - "loss": 4.2969, - "step": 1426 - }, - { - "epoch": 0.8, - "learning_rate": 8.434579439252336e-06, - "loss": 4.0419, - "step": 1427 - }, - { - "epoch": 0.8, - "learning_rate": 8.411214953271028e-06, - "loss": 4.0483, - "step": 1428 - }, - { - "epoch": 0.8, - "learning_rate": 8.38785046728972e-06, - "loss": 4.2055, - "step": 1429 - }, - { - "epoch": 0.8, - "learning_rate": 8.364485981308411e-06, - "loss": 4.332, - "step": 1430 - }, - { - "epoch": 0.8, - "learning_rate": 8.341121495327103e-06, - "loss": 4.2456, - "step": 1431 - }, - { - "epoch": 0.8, - "learning_rate": 8.317757009345793e-06, - "loss": 4.2195, - "step": 1432 - }, - { - "epoch": 0.8, - "learning_rate": 8.294392523364487e-06, - "loss": 4.4525, - "step": 1433 - }, - { - "epoch": 0.8, - "learning_rate": 8.271028037383177e-06, - "loss": 4.0624, - "step": 1434 - }, - { - "epoch": 0.8, - "learning_rate": 8.24766355140187e-06, - "loss": 3.9296, - "step": 1435 - }, - { - "epoch": 0.8, - "learning_rate": 8.22429906542056e-06, - "loss": 3.9153, - "step": 1436 - }, - { - "epoch": 0.81, - "learning_rate": 8.200934579439253e-06, - "loss": 4.3315, - "step": 1437 - }, - { - "epoch": 0.81, - "learning_rate": 8.177570093457943e-06, - "loss": 4.0011, - "step": 1438 - }, - { - "epoch": 0.81, - "learning_rate": 8.154205607476635e-06, - "loss": 4.2285, - "step": 1439 - }, - { - "epoch": 0.81, - "learning_rate": 8.130841121495327e-06, - "loss": 4.1969, - "step": 1440 - }, - { - "epoch": 0.81, - "learning_rate": 8.107476635514018e-06, - "loss": 4.3416, - "step": 1441 - }, - { - "epoch": 0.81, - "learning_rate": 8.08411214953271e-06, - "loss": 4.0149, - "step": 1442 - }, - { - "epoch": 0.81, - "learning_rate": 8.060747663551402e-06, - "loss": 4.0146, - "step": 1443 - }, - { - "epoch": 0.81, - "learning_rate": 8.037383177570094e-06, - "loss": 4.1167, - "step": 1444 - }, - { - "epoch": 0.81, - "learning_rate": 8.014018691588785e-06, - "loss": 4.2799, - "step": 1445 - }, - { - "epoch": 0.81, - "learning_rate": 7.990654205607475e-06, - "loss": 4.3641, - "step": 1446 - }, - { - "epoch": 0.81, - "learning_rate": 7.967289719626169e-06, - "loss": 4.2311, - "step": 1447 - }, - { - "epoch": 0.81, - "learning_rate": 7.943925233644859e-06, - "loss": 4.4094, - "step": 1448 - }, - { - "epoch": 0.81, - "learning_rate": 7.920560747663552e-06, - "loss": 4.5298, - "step": 1449 - }, - { - "epoch": 0.81, - "learning_rate": 7.897196261682242e-06, - "loss": 4.159, - "step": 1450 - }, - { - "epoch": 0.81, - "learning_rate": 7.873831775700936e-06, - "loss": 4.0792, - "step": 1451 - }, - { - "epoch": 0.81, - "learning_rate": 7.850467289719626e-06, - "loss": 4.1228, - "step": 1452 - }, - { - "epoch": 0.81, - "learning_rate": 7.827102803738317e-06, - "loss": 3.9725, - "step": 1453 - }, - { - "epoch": 0.82, - "learning_rate": 7.803738317757009e-06, - "loss": 3.9233, - "step": 1454 - }, - { - "epoch": 0.82, - "learning_rate": 7.7803738317757e-06, - "loss": 4.0491, - "step": 1455 - }, - { - "epoch": 0.82, - "learning_rate": 7.757009345794392e-06, - "loss": 4.0949, - "step": 1456 - }, - { - "epoch": 0.82, - "learning_rate": 7.733644859813084e-06, - "loss": 4.3294, - "step": 1457 - }, - { - "epoch": 0.82, - "learning_rate": 7.710280373831776e-06, - "loss": 4.2094, - "step": 1458 - }, - { - "epoch": 0.82, - "learning_rate": 7.686915887850467e-06, - "loss": 3.9979, - "step": 1459 - }, - { - "epoch": 0.82, - "learning_rate": 7.663551401869157e-06, - "loss": 4.189, - "step": 1460 - }, - { - "epoch": 0.82, - "learning_rate": 7.64018691588785e-06, - "loss": 4.1713, - "step": 1461 - }, - { - "epoch": 0.82, - "learning_rate": 7.616822429906542e-06, - "loss": 4.2261, - "step": 1462 - }, - { - "epoch": 0.82, - "learning_rate": 7.593457943925234e-06, - "loss": 4.154, - "step": 1463 - }, - { - "epoch": 0.82, - "learning_rate": 7.570093457943924e-06, - "loss": 4.2109, - "step": 1464 - }, - { - "epoch": 0.82, - "learning_rate": 7.546728971962617e-06, - "loss": 4.3885, - "step": 1465 - }, - { - "epoch": 0.82, - "learning_rate": 7.523364485981308e-06, - "loss": 3.8577, - "step": 1466 - }, - { - "epoch": 0.82, - "learning_rate": 7.5e-06, - "loss": 4.0928, - "step": 1467 - }, - { - "epoch": 0.82, - "learning_rate": 7.476635514018692e-06, - "loss": 4.1424, - "step": 1468 - }, - { - "epoch": 0.82, - "learning_rate": 7.453271028037384e-06, - "loss": 3.821, - "step": 1469 - }, - { - "epoch": 0.82, - "learning_rate": 7.429906542056075e-06, - "loss": 3.8844, - "step": 1470 - }, - { - "epoch": 0.82, - "learning_rate": 7.406542056074766e-06, - "loss": 4.113, - "step": 1471 - }, - { - "epoch": 0.83, - "learning_rate": 7.383177570093458e-06, - "loss": 3.8827, - "step": 1472 - }, - { - "epoch": 0.83, - "learning_rate": 7.3598130841121496e-06, - "loss": 4.0513, - "step": 1473 - }, - { - "epoch": 0.83, - "learning_rate": 7.336448598130841e-06, - "loss": 3.9866, - "step": 1474 - }, - { - "epoch": 0.83, - "learning_rate": 7.313084112149533e-06, - "loss": 4.1726, - "step": 1475 - }, - { - "epoch": 0.83, - "learning_rate": 7.289719626168225e-06, - "loss": 4.1526, - "step": 1476 - }, - { - "epoch": 0.83, - "learning_rate": 7.266355140186916e-06, - "loss": 4.2075, - "step": 1477 - }, - { - "epoch": 0.83, - "learning_rate": 7.242990654205607e-06, - "loss": 3.8465, - "step": 1478 - }, - { - "epoch": 0.83, - "learning_rate": 7.219626168224299e-06, - "loss": 4.1633, - "step": 1479 - }, - { - "epoch": 0.83, - "learning_rate": 7.196261682242991e-06, - "loss": 3.9722, - "step": 1480 - }, - { - "epoch": 0.83, - "learning_rate": 7.172897196261682e-06, - "loss": 4.0622, - "step": 1481 - }, - { - "epoch": 0.83, - "learning_rate": 7.149532710280374e-06, - "loss": 4.1813, - "step": 1482 - }, - { - "epoch": 0.83, - "learning_rate": 7.126168224299066e-06, - "loss": 4.1506, - "step": 1483 - }, - { - "epoch": 0.83, - "learning_rate": 7.1028037383177574e-06, - "loss": 3.7479, - "step": 1484 - }, - { - "epoch": 0.83, - "learning_rate": 7.079439252336449e-06, - "loss": 4.0951, - "step": 1485 - }, - { - "epoch": 0.83, - "learning_rate": 7.05607476635514e-06, - "loss": 4.239, - "step": 1486 - }, - { - "epoch": 0.83, - "learning_rate": 7.032710280373832e-06, - "loss": 4.2861, - "step": 1487 - }, - { - "epoch": 0.83, - "learning_rate": 7.009345794392523e-06, - "loss": 4.242, - "step": 1488 - }, - { - "epoch": 0.83, - "learning_rate": 6.985981308411215e-06, - "loss": 4.1746, - "step": 1489 - }, - { - "epoch": 0.84, - "learning_rate": 6.962616822429907e-06, - "loss": 3.9778, - "step": 1490 - }, - { - "epoch": 0.84, - "learning_rate": 6.9392523364485985e-06, - "loss": 4.5709, - "step": 1491 - }, - { - "epoch": 0.84, - "learning_rate": 6.91588785046729e-06, - "loss": 4.1187, - "step": 1492 - }, - { - "epoch": 0.84, - "learning_rate": 6.892523364485981e-06, - "loss": 3.7873, - "step": 1493 - }, - { - "epoch": 0.84, - "learning_rate": 6.869158878504673e-06, - "loss": 3.9793, - "step": 1494 - }, - { - "epoch": 0.84, - "learning_rate": 6.8457943925233644e-06, - "loss": 4.111, - "step": 1495 - }, - { - "epoch": 0.84, - "learning_rate": 6.822429906542056e-06, - "loss": 4.5429, - "step": 1496 - }, - { - "epoch": 0.84, - "learning_rate": 6.799065420560748e-06, - "loss": 4.3431, - "step": 1497 - }, - { - "epoch": 0.84, - "learning_rate": 6.7757009345794396e-06, - "loss": 4.5138, - "step": 1498 - }, - { - "epoch": 0.84, - "learning_rate": 6.752336448598131e-06, - "loss": 4.0346, - "step": 1499 - }, - { - "epoch": 0.84, - "learning_rate": 6.728971962616822e-06, - "loss": 4.3001, - "step": 1500 - }, - { - "epoch": 0.84, - "eval_loss": 4.2075958251953125, - "eval_runtime": 947.3446, - "eval_samples_per_second": 2.789, - "eval_steps_per_second": 0.349, - "eval_wer": 1.9032526775089251, - "step": 1500 - }, - { - "epoch": 0.84, - "learning_rate": 6.705607476635514e-06, - "loss": 4.2039, - "step": 1501 - }, - { - "epoch": 0.84, - "learning_rate": 6.6822429906542055e-06, - "loss": 4.1607, - "step": 1502 - }, - { - "epoch": 0.84, - "learning_rate": 6.658878504672897e-06, - "loss": 4.2332, - "step": 1503 - }, - { - "epoch": 0.84, - "learning_rate": 6.635514018691589e-06, - "loss": 4.108, - "step": 1504 - }, - { - "epoch": 0.84, - "learning_rate": 6.612149532710281e-06, - "loss": 4.2319, - "step": 1505 - }, - { - "epoch": 0.84, - "learning_rate": 6.588785046728972e-06, - "loss": 4.055, - "step": 1506 - }, - { - "epoch": 0.84, - "learning_rate": 6.565420560747664e-06, - "loss": 4.0099, - "step": 1507 - }, - { - "epoch": 0.85, - "learning_rate": 6.542056074766355e-06, - "loss": 4.0962, - "step": 1508 - }, - { - "epoch": 0.85, - "learning_rate": 6.518691588785047e-06, - "loss": 4.1213, - "step": 1509 - }, - { - "epoch": 0.85, - "learning_rate": 6.495327102803738e-06, - "loss": 4.0894, - "step": 1510 - }, - { - "epoch": 0.85, - "learning_rate": 6.47196261682243e-06, - "loss": 3.9209, - "step": 1511 - }, - { - "epoch": 0.85, - "learning_rate": 6.448598130841122e-06, - "loss": 4.2152, - "step": 1512 - }, - { - "epoch": 0.85, - "learning_rate": 6.425233644859813e-06, - "loss": 3.8835, - "step": 1513 - }, - { - "epoch": 0.85, - "learning_rate": 6.401869158878505e-06, - "loss": 4.0394, - "step": 1514 - }, - { - "epoch": 0.85, - "learning_rate": 6.378504672897196e-06, - "loss": 4.0479, - "step": 1515 - }, - { - "epoch": 0.85, - "learning_rate": 6.355140186915888e-06, - "loss": 4.4315, - "step": 1516 - }, - { - "epoch": 0.85, - "learning_rate": 6.331775700934579e-06, - "loss": 4.0129, - "step": 1517 - }, - { - "epoch": 0.85, - "learning_rate": 6.308411214953271e-06, - "loss": 4.2214, - "step": 1518 - }, - { - "epoch": 0.85, - "learning_rate": 6.285046728971963e-06, - "loss": 4.3242, - "step": 1519 - }, - { - "epoch": 0.85, - "learning_rate": 6.2616822429906544e-06, - "loss": 4.0849, - "step": 1520 - }, - { - "epoch": 0.85, - "learning_rate": 6.238317757009346e-06, - "loss": 4.0637, - "step": 1521 - }, - { - "epoch": 0.85, - "learning_rate": 6.214953271028037e-06, - "loss": 3.9984, - "step": 1522 - }, - { - "epoch": 0.85, - "learning_rate": 6.191588785046729e-06, - "loss": 4.3311, - "step": 1523 - }, - { - "epoch": 0.85, - "learning_rate": 6.16822429906542e-06, - "loss": 3.9591, - "step": 1524 - }, - { - "epoch": 0.85, - "learning_rate": 6.144859813084112e-06, - "loss": 4.0905, - "step": 1525 - }, - { - "epoch": 0.86, - "learning_rate": 6.121495327102804e-06, - "loss": 3.9547, - "step": 1526 - }, - { - "epoch": 0.86, - "learning_rate": 6.0981308411214955e-06, - "loss": 4.0565, - "step": 1527 - }, - { - "epoch": 0.86, - "learning_rate": 6.074766355140187e-06, - "loss": 4.1071, - "step": 1528 - }, - { - "epoch": 0.86, - "learning_rate": 6.051401869158879e-06, - "loss": 3.6556, - "step": 1529 - }, - { - "epoch": 0.86, - "learning_rate": 6.02803738317757e-06, - "loss": 3.9483, - "step": 1530 - }, - { - "epoch": 0.86, - "learning_rate": 6.0046728971962615e-06, - "loss": 3.9194, - "step": 1531 - }, - { - "epoch": 0.86, - "learning_rate": 5.981308411214953e-06, - "loss": 3.8927, - "step": 1532 - }, - { - "epoch": 0.86, - "learning_rate": 5.957943925233645e-06, - "loss": 4.0501, - "step": 1533 - }, - { - "epoch": 0.86, - "learning_rate": 5.934579439252337e-06, - "loss": 4.0536, - "step": 1534 - }, - { - "epoch": 0.86, - "learning_rate": 5.911214953271028e-06, - "loss": 4.0931, - "step": 1535 - }, - { - "epoch": 0.86, - "learning_rate": 5.88785046728972e-06, - "loss": 3.8412, - "step": 1536 - }, - { - "epoch": 0.86, - "learning_rate": 5.864485981308411e-06, - "loss": 4.1066, - "step": 1537 - }, - { - "epoch": 0.86, - "learning_rate": 5.8411214953271025e-06, - "loss": 4.3525, - "step": 1538 - }, - { - "epoch": 0.86, - "learning_rate": 5.817757009345794e-06, - "loss": 4.0533, - "step": 1539 - }, - { - "epoch": 0.86, - "learning_rate": 5.794392523364486e-06, - "loss": 4.1529, - "step": 1540 - }, - { - "epoch": 0.86, - "learning_rate": 5.771028037383178e-06, - "loss": 4.2804, - "step": 1541 - }, - { - "epoch": 0.86, - "learning_rate": 5.747663551401869e-06, - "loss": 3.8271, - "step": 1542 - }, - { - "epoch": 0.86, - "learning_rate": 5.724299065420561e-06, - "loss": 4.2896, - "step": 1543 - }, - { - "epoch": 0.87, - "learning_rate": 5.700934579439252e-06, - "loss": 4.2828, - "step": 1544 - }, - { - "epoch": 0.87, - "learning_rate": 5.677570093457944e-06, - "loss": 4.6037, - "step": 1545 - }, - { - "epoch": 0.87, - "learning_rate": 5.654205607476635e-06, - "loss": 3.9111, - "step": 1546 - }, - { - "epoch": 0.87, - "learning_rate": 5.630841121495327e-06, - "loss": 4.2947, - "step": 1547 - }, - { - "epoch": 0.87, - "learning_rate": 5.607476635514019e-06, - "loss": 4.2559, - "step": 1548 - }, - { - "epoch": 0.87, - "learning_rate": 5.58411214953271e-06, - "loss": 4.4756, - "step": 1549 - }, - { - "epoch": 0.87, - "learning_rate": 5.560747663551402e-06, - "loss": 3.9216, - "step": 1550 - }, - { - "epoch": 0.87, - "learning_rate": 5.537383177570093e-06, - "loss": 4.3611, - "step": 1551 - }, - { - "epoch": 0.87, - "learning_rate": 5.514018691588785e-06, - "loss": 4.1302, - "step": 1552 - }, - { - "epoch": 0.87, - "learning_rate": 5.490654205607476e-06, - "loss": 4.0421, - "step": 1553 - }, - { - "epoch": 0.87, - "learning_rate": 5.467289719626168e-06, - "loss": 3.9994, - "step": 1554 - }, - { - "epoch": 0.87, - "learning_rate": 5.44392523364486e-06, - "loss": 4.3427, - "step": 1555 - }, - { - "epoch": 0.87, - "learning_rate": 5.4205607476635515e-06, - "loss": 4.5006, - "step": 1556 - }, - { - "epoch": 0.87, - "learning_rate": 5.397196261682243e-06, - "loss": 4.18, - "step": 1557 - }, - { - "epoch": 0.87, - "learning_rate": 5.373831775700935e-06, - "loss": 4.2818, - "step": 1558 - }, - { - "epoch": 0.87, - "learning_rate": 5.350467289719626e-06, - "loss": 3.9262, - "step": 1559 - }, - { - "epoch": 0.87, - "learning_rate": 5.3271028037383174e-06, - "loss": 4.2223, - "step": 1560 - }, - { - "epoch": 0.88, - "learning_rate": 5.303738317757009e-06, - "loss": 4.1897, - "step": 1561 - }, - { - "epoch": 0.88, - "learning_rate": 5.280373831775701e-06, - "loss": 4.1049, - "step": 1562 - }, - { - "epoch": 0.88, - "learning_rate": 5.2570093457943925e-06, - "loss": 4.2849, - "step": 1563 - }, - { - "epoch": 0.88, - "learning_rate": 5.233644859813084e-06, - "loss": 4.0656, - "step": 1564 - }, - { - "epoch": 0.88, - "learning_rate": 5.210280373831776e-06, - "loss": 4.1985, - "step": 1565 - }, - { - "epoch": 0.88, - "learning_rate": 5.186915887850467e-06, - "loss": 3.958, - "step": 1566 - }, - { - "epoch": 0.88, - "learning_rate": 5.1635514018691585e-06, - "loss": 3.8778, - "step": 1567 - }, - { - "epoch": 0.88, - "learning_rate": 5.14018691588785e-06, - "loss": 4.1992, - "step": 1568 - }, - { - "epoch": 0.88, - "learning_rate": 5.116822429906542e-06, - "loss": 4.1769, - "step": 1569 - }, - { - "epoch": 0.88, - "learning_rate": 5.093457943925234e-06, - "loss": 3.9919, - "step": 1570 - }, - { - "epoch": 0.88, - "learning_rate": 5.070093457943925e-06, - "loss": 4.0812, - "step": 1571 - }, - { - "epoch": 0.88, - "learning_rate": 5.046728971962617e-06, - "loss": 4.3101, - "step": 1572 - }, - { - "epoch": 0.88, - "learning_rate": 5.023364485981308e-06, - "loss": 4.2435, - "step": 1573 - }, - { - "epoch": 0.88, - "learning_rate": 4.9999999999999996e-06, - "loss": 3.9456, - "step": 1574 - }, - { - "epoch": 0.88, - "learning_rate": 4.976635514018691e-06, - "loss": 3.8921, - "step": 1575 - }, - { - "epoch": 0.88, - "learning_rate": 4.953271028037383e-06, - "loss": 4.1622, - "step": 1576 - }, - { - "epoch": 0.88, - "learning_rate": 4.929906542056075e-06, - "loss": 3.9452, - "step": 1577 - }, - { - "epoch": 0.88, - "learning_rate": 4.906542056074766e-06, - "loss": 4.215, - "step": 1578 - }, - { - "epoch": 0.89, - "learning_rate": 4.883177570093458e-06, - "loss": 4.184, - "step": 1579 - }, - { - "epoch": 0.89, - "learning_rate": 4.85981308411215e-06, - "loss": 4.3362, - "step": 1580 - }, - { - "epoch": 0.89, - "learning_rate": 4.836448598130841e-06, - "loss": 4.2123, - "step": 1581 - }, - { - "epoch": 0.89, - "learning_rate": 4.813084112149532e-06, - "loss": 4.2072, - "step": 1582 - }, - { - "epoch": 0.89, - "learning_rate": 4.789719626168224e-06, - "loss": 4.107, - "step": 1583 - }, - { - "epoch": 0.89, - "learning_rate": 4.766355140186916e-06, - "loss": 4.0697, - "step": 1584 - }, - { - "epoch": 0.89, - "learning_rate": 4.7429906542056074e-06, - "loss": 3.9829, - "step": 1585 - }, - { - "epoch": 0.89, - "learning_rate": 4.719626168224299e-06, - "loss": 4.0668, - "step": 1586 - }, - { - "epoch": 0.89, - "learning_rate": 4.696261682242991e-06, - "loss": 4.008, - "step": 1587 - }, - { - "epoch": 0.89, - "learning_rate": 4.672897196261682e-06, - "loss": 4.2659, - "step": 1588 - }, - { - "epoch": 0.89, - "learning_rate": 4.649532710280373e-06, - "loss": 4.0806, - "step": 1589 - }, - { - "epoch": 0.89, - "learning_rate": 4.626168224299065e-06, - "loss": 3.962, - "step": 1590 - }, - { - "epoch": 0.89, - "learning_rate": 4.602803738317757e-06, - "loss": 4.0991, - "step": 1591 - }, - { - "epoch": 0.89, - "learning_rate": 4.5794392523364485e-06, - "loss": 3.9898, - "step": 1592 - }, - { - "epoch": 0.89, - "learning_rate": 4.55607476635514e-06, - "loss": 3.9846, - "step": 1593 - }, - { - "epoch": 0.89, - "learning_rate": 4.532710280373832e-06, - "loss": 4.1954, - "step": 1594 - }, - { - "epoch": 0.89, - "learning_rate": 4.509345794392523e-06, - "loss": 4.0624, - "step": 1595 - }, - { - "epoch": 0.89, - "learning_rate": 4.4859813084112145e-06, - "loss": 4.3319, - "step": 1596 - }, - { - "epoch": 0.9, - "learning_rate": 4.462616822429906e-06, - "loss": 4.2242, - "step": 1597 - }, - { - "epoch": 0.9, - "learning_rate": 4.439252336448598e-06, - "loss": 4.4162, - "step": 1598 - }, - { - "epoch": 0.9, - "learning_rate": 4.4158878504672896e-06, - "loss": 5.0513, - "step": 1599 - }, - { - "epoch": 0.9, - "learning_rate": 4.392523364485981e-06, - "loss": 4.6567, - "step": 1600 - }, - { - "epoch": 0.9, - "learning_rate": 4.369158878504673e-06, - "loss": 4.0805, - "step": 1601 - }, - { - "epoch": 0.9, - "learning_rate": 4.345794392523365e-06, - "loss": 3.9221, - "step": 1602 - }, - { - "epoch": 0.9, - "learning_rate": 4.3224299065420555e-06, - "loss": 4.0248, - "step": 1603 - }, - { - "epoch": 0.9, - "learning_rate": 4.299065420560747e-06, - "loss": 4.1355, - "step": 1604 - }, - { - "epoch": 0.9, - "learning_rate": 4.275700934579439e-06, - "loss": 3.9084, - "step": 1605 - }, - { - "epoch": 0.9, - "learning_rate": 4.252336448598131e-06, - "loss": 4.1655, - "step": 1606 - }, - { - "epoch": 0.9, - "learning_rate": 4.228971962616822e-06, - "loss": 4.0852, - "step": 1607 - }, - { - "epoch": 0.9, - "learning_rate": 4.205607476635514e-06, - "loss": 3.9907, - "step": 1608 - }, - { - "epoch": 0.9, - "learning_rate": 4.182242990654206e-06, - "loss": 4.2936, - "step": 1609 - }, - { - "epoch": 0.9, - "learning_rate": 4.158878504672897e-06, - "loss": 3.945, - "step": 1610 - }, - { - "epoch": 0.9, - "learning_rate": 4.135514018691588e-06, - "loss": 4.2846, - "step": 1611 - }, - { - "epoch": 0.9, - "learning_rate": 4.11214953271028e-06, - "loss": 3.9944, - "step": 1612 - }, - { - "epoch": 0.9, - "learning_rate": 4.088785046728972e-06, - "loss": 4.1348, - "step": 1613 - }, - { - "epoch": 0.9, - "learning_rate": 4.065420560747663e-06, - "loss": 3.8733, - "step": 1614 - }, - { - "epoch": 0.91, - "learning_rate": 4.042056074766355e-06, - "loss": 4.1246, - "step": 1615 - }, - { - "epoch": 0.91, - "learning_rate": 4.018691588785047e-06, - "loss": 4.0774, - "step": 1616 - }, - { - "epoch": 0.91, - "learning_rate": 3.995327102803738e-06, - "loss": 4.0991, - "step": 1617 - }, - { - "epoch": 0.91, - "learning_rate": 3.971962616822429e-06, - "loss": 4.0718, - "step": 1618 - }, - { - "epoch": 0.91, - "learning_rate": 3.948598130841121e-06, - "loss": 4.1337, - "step": 1619 - }, - { - "epoch": 0.91, - "learning_rate": 3.925233644859813e-06, - "loss": 4.1297, - "step": 1620 - }, - { - "epoch": 0.91, - "learning_rate": 3.9018691588785045e-06, - "loss": 4.2224, - "step": 1621 - }, - { - "epoch": 0.91, - "learning_rate": 3.878504672897196e-06, - "loss": 3.8608, - "step": 1622 - }, - { - "epoch": 0.91, - "learning_rate": 3.855140186915888e-06, - "loss": 3.9987, - "step": 1623 - }, - { - "epoch": 0.91, - "learning_rate": 3.831775700934579e-06, - "loss": 3.9592, - "step": 1624 - }, - { - "epoch": 0.91, - "learning_rate": 3.808411214953271e-06, - "loss": 4.0687, - "step": 1625 - }, - { - "epoch": 0.91, - "learning_rate": 3.785046728971962e-06, - "loss": 4.2544, - "step": 1626 - }, - { - "epoch": 0.91, - "learning_rate": 3.761682242990654e-06, - "loss": 4.1496, - "step": 1627 - }, - { - "epoch": 0.91, - "learning_rate": 3.738317757009346e-06, - "loss": 3.999, - "step": 1628 - }, - { - "epoch": 0.91, - "learning_rate": 3.7149532710280376e-06, - "loss": 4.2071, - "step": 1629 - }, - { - "epoch": 0.91, - "learning_rate": 3.691588785046729e-06, - "loss": 3.9809, - "step": 1630 - }, - { - "epoch": 0.91, - "learning_rate": 3.6682242990654206e-06, - "loss": 4.1445, - "step": 1631 - }, - { - "epoch": 0.91, - "learning_rate": 3.6448598130841123e-06, - "loss": 4.1166, - "step": 1632 - }, - { - "epoch": 0.92, - "learning_rate": 3.6214953271028036e-06, - "loss": 4.1841, - "step": 1633 - }, - { - "epoch": 0.92, - "learning_rate": 3.5981308411214953e-06, - "loss": 4.0224, - "step": 1634 - }, - { - "epoch": 0.92, - "learning_rate": 3.574766355140187e-06, - "loss": 4.1591, - "step": 1635 - }, - { - "epoch": 0.92, - "learning_rate": 3.5514018691588787e-06, - "loss": 3.991, - "step": 1636 - }, - { - "epoch": 0.92, - "learning_rate": 3.52803738317757e-06, - "loss": 4.0546, - "step": 1637 - }, - { - "epoch": 0.92, - "learning_rate": 3.5046728971962617e-06, - "loss": 4.2639, - "step": 1638 - }, - { - "epoch": 0.92, - "learning_rate": 3.4813084112149534e-06, - "loss": 3.9807, - "step": 1639 - }, - { - "epoch": 0.92, - "learning_rate": 3.457943925233645e-06, - "loss": 4.1094, - "step": 1640 - }, - { - "epoch": 0.92, - "learning_rate": 3.4345794392523364e-06, - "loss": 4.2343, - "step": 1641 - }, - { - "epoch": 0.92, - "learning_rate": 3.411214953271028e-06, - "loss": 4.1518, - "step": 1642 - }, - { - "epoch": 0.92, - "learning_rate": 3.3878504672897198e-06, - "loss": 4.064, - "step": 1643 - }, - { - "epoch": 0.92, - "learning_rate": 3.364485981308411e-06, - "loss": 4.162, - "step": 1644 - }, - { - "epoch": 0.92, - "learning_rate": 3.3411214953271028e-06, - "loss": 4.3617, - "step": 1645 - }, - { - "epoch": 0.92, - "learning_rate": 3.3177570093457945e-06, - "loss": 4.5435, - "step": 1646 - }, - { - "epoch": 0.92, - "learning_rate": 3.294392523364486e-06, - "loss": 4.0725, - "step": 1647 - }, - { - "epoch": 0.92, - "learning_rate": 3.2710280373831774e-06, - "loss": 4.7151, - "step": 1648 - }, { "epoch": 0.92, - "learning_rate": 3.247663551401869e-06, - "loss": 4.4425, - "step": 1649 + "learning_rate": 5.433673469387755e-06, + "loss": 4.0661, + "step": 824 }, { "epoch": 0.92, - "learning_rate": 3.224299065420561e-06, - "loss": 4.0702, - "step": 1650 - }, - { - "epoch": 0.93, - "learning_rate": 3.2009345794392525e-06, - "loss": 4.2053, - "step": 1651 - }, - { - "epoch": 0.93, - "learning_rate": 3.177570093457944e-06, - "loss": 4.0842, - "step": 1652 - }, - { - "epoch": 0.93, - "learning_rate": 3.1542056074766355e-06, - "loss": 4.1126, - "step": 1653 - }, - { - "epoch": 0.93, - "learning_rate": 3.1308411214953272e-06, - "loss": 4.0547, - "step": 1654 - }, - { - "epoch": 0.93, - "learning_rate": 3.1074766355140185e-06, - "loss": 3.9678, - "step": 1655 - }, - { - "epoch": 0.93, - "learning_rate": 3.08411214953271e-06, - "loss": 4.0489, - "step": 1656 - }, - { - "epoch": 0.93, - "learning_rate": 3.060747663551402e-06, - "loss": 4.1916, - "step": 1657 - }, - { - "epoch": 0.93, - "learning_rate": 3.0373831775700936e-06, - "loss": 4.1362, - "step": 1658 - }, - { - "epoch": 0.93, - "learning_rate": 3.014018691588785e-06, - "loss": 3.9667, - "step": 1659 - }, - { - "epoch": 0.93, - "learning_rate": 2.9906542056074766e-06, - "loss": 4.108, - "step": 1660 - }, - { - "epoch": 0.93, - "learning_rate": 2.9672897196261683e-06, - "loss": 4.2037, - "step": 1661 - }, - { - "epoch": 0.93, - "learning_rate": 2.94392523364486e-06, - "loss": 4.1952, - "step": 1662 - }, - { - "epoch": 0.93, - "learning_rate": 2.9205607476635513e-06, - "loss": 3.8044, - "step": 1663 - }, - { - "epoch": 0.93, - "learning_rate": 2.897196261682243e-06, - "loss": 4.1709, - "step": 1664 - }, - { - "epoch": 0.93, - "learning_rate": 2.8738317757009347e-06, - "loss": 3.9257, - "step": 1665 - }, - { - "epoch": 0.93, - "learning_rate": 2.850467289719626e-06, - "loss": 4.0337, - "step": 1666 - }, - { - "epoch": 0.93, - "learning_rate": 2.8271028037383177e-06, - "loss": 4.0067, - "step": 1667 + "learning_rate": 5.357142857142857e-06, + "loss": 4.3232, + "step": 825 }, { "epoch": 0.93, - "learning_rate": 2.8037383177570094e-06, - "loss": 4.0033, - "step": 1668 - }, - { - "epoch": 0.94, - "learning_rate": 2.780373831775701e-06, - "loss": 4.0537, - "step": 1669 + "learning_rate": 5.280612244897959e-06, + "loss": 4.2064, + "step": 826 }, { - "epoch": 0.94, - "learning_rate": 2.7570093457943923e-06, - "loss": 4.0012, - "step": 1670 + "epoch": 0.93, + "learning_rate": 5.204081632653062e-06, + "loss": 4.1131, + "step": 827 }, { - "epoch": 0.94, - "learning_rate": 2.733644859813084e-06, - "loss": 4.0172, - "step": 1671 + "epoch": 0.93, + "learning_rate": 5.127551020408164e-06, + "loss": 4.0079, + "step": 828 }, { - "epoch": 0.94, - "learning_rate": 2.7102803738317757e-06, - "loss": 4.2428, - "step": 1672 + "epoch": 0.93, + "learning_rate": 5.0510204081632655e-06, + "loss": 3.988, + "step": 829 }, { - "epoch": 0.94, - "learning_rate": 2.6869158878504674e-06, - "loss": 3.9845, - "step": 1673 + "epoch": 0.93, + "learning_rate": 4.9744897959183674e-06, + "loss": 4.1855, + "step": 830 }, { - "epoch": 0.94, - "learning_rate": 2.6635514018691587e-06, - "loss": 4.1325, - "step": 1674 + "epoch": 0.93, + "learning_rate": 4.897959183673469e-06, + "loss": 3.9875, + "step": 831 }, { - "epoch": 0.94, - "learning_rate": 2.6401869158878504e-06, - "loss": 4.3764, - "step": 1675 + "epoch": 0.93, + "learning_rate": 4.821428571428572e-06, + "loss": 3.9255, + "step": 832 }, { - "epoch": 0.94, - "learning_rate": 2.616822429906542e-06, - "loss": 4.0186, - "step": 1676 + "epoch": 0.93, + "learning_rate": 4.744897959183674e-06, + "loss": 4.2085, + "step": 833 }, { - "epoch": 0.94, - "learning_rate": 2.5934579439252334e-06, - "loss": 3.848, - "step": 1677 + "epoch": 0.93, + "learning_rate": 4.668367346938776e-06, + "loss": 4.0791, + "step": 834 }, { "epoch": 0.94, - "learning_rate": 2.570093457943925e-06, - "loss": 4.091, - "step": 1678 + "learning_rate": 4.591836734693878e-06, + "loss": 4.1169, + "step": 835 }, { "epoch": 0.94, - "learning_rate": 2.546728971962617e-06, - "loss": 4.246, - "step": 1679 + "learning_rate": 4.51530612244898e-06, + "loss": 4.096, + "step": 836 }, { "epoch": 0.94, - "learning_rate": 2.5233644859813085e-06, - "loss": 4.2399, - "step": 1680 + "learning_rate": 4.438775510204082e-06, + "loss": 4.1276, + "step": 837 }, { "epoch": 0.94, - "learning_rate": 2.4999999999999998e-06, - "loss": 3.8142, - "step": 1681 + "learning_rate": 4.3622448979591836e-06, + "loss": 4.1881, + "step": 838 }, { "epoch": 0.94, - "learning_rate": 2.4766355140186915e-06, - "loss": 3.8551, - "step": 1682 + "learning_rate": 4.2857142857142855e-06, + "loss": 3.9964, + "step": 839 }, { "epoch": 0.94, - "learning_rate": 2.453271028037383e-06, - "loss": 4.3228, - "step": 1683 + "learning_rate": 4.209183673469387e-06, + "loss": 4.1282, + "step": 840 }, { "epoch": 0.94, - "learning_rate": 2.429906542056075e-06, - "loss": 3.833, - "step": 1684 + "learning_rate": 4.132653061224489e-06, + "loss": 4.2978, + "step": 841 }, { "epoch": 0.94, - "learning_rate": 2.406542056074766e-06, - "loss": 4.1549, - "step": 1685 - }, - { - "epoch": 0.95, - "learning_rate": 2.383177570093458e-06, - "loss": 4.1394, - "step": 1686 - }, - { - "epoch": 0.95, - "learning_rate": 2.3598130841121496e-06, - "loss": 3.9934, - "step": 1687 - }, - { - "epoch": 0.95, - "learning_rate": 2.336448598130841e-06, - "loss": 4.2384, - "step": 1688 - }, - { - "epoch": 0.95, - "learning_rate": 2.3130841121495325e-06, - "loss": 4.1392, - "step": 1689 - }, - { - "epoch": 0.95, - "learning_rate": 2.2897196261682242e-06, - "loss": 3.9704, - "step": 1690 - }, - { - "epoch": 0.95, - "learning_rate": 2.266355140186916e-06, - "loss": 4.3617, - "step": 1691 - }, - { - "epoch": 0.95, - "learning_rate": 2.2429906542056072e-06, - "loss": 4.1711, - "step": 1692 - }, - { - "epoch": 0.95, - "learning_rate": 2.219626168224299e-06, - "loss": 4.417, - "step": 1693 - }, - { - "epoch": 0.95, - "learning_rate": 2.1962616822429906e-06, - "loss": 4.4343, - "step": 1694 + "learning_rate": 4.056122448979592e-06, + "loss": 4.1185, + "step": 842 }, { "epoch": 0.95, - "learning_rate": 2.1728971962616823e-06, - "loss": 4.4066, - "step": 1695 + "learning_rate": 3.979591836734694e-06, + "loss": 4.298, + "step": 843 }, { "epoch": 0.95, - "learning_rate": 2.1495327102803736e-06, - "loss": 4.8174, - "step": 1696 + "learning_rate": 3.903061224489796e-06, + "loss": 4.2988, + "step": 844 }, { "epoch": 0.95, - "learning_rate": 2.1261682242990653e-06, - "loss": 4.2197, - "step": 1697 + "learning_rate": 3.826530612244898e-06, + "loss": 4.3789, + "step": 845 }, { "epoch": 0.95, - "learning_rate": 2.102803738317757e-06, - "loss": 4.7544, - "step": 1698 + "learning_rate": 3.75e-06, + "loss": 4.5702, + "step": 846 }, { "epoch": 0.95, - "learning_rate": 2.0794392523364483e-06, - "loss": 4.5876, - "step": 1699 + "learning_rate": 3.673469387755102e-06, + "loss": 4.1955, + "step": 847 }, { "epoch": 0.95, - "learning_rate": 2.05607476635514e-06, - "loss": 4.5133, - "step": 1700 + "learning_rate": 3.596938775510204e-06, + "loss": 4.5673, + "step": 848 }, { "epoch": 0.95, - "learning_rate": 2.0327102803738317e-06, - "loss": 4.3535, - "step": 1701 + "learning_rate": 3.5204081632653062e-06, + "loss": 4.5096, + "step": 849 }, { "epoch": 0.95, - "learning_rate": 2.0093457943925234e-06, - "loss": 3.9685, - "step": 1702 + "learning_rate": 3.443877551020408e-06, + "loss": 4.2798, + "step": 850 }, { "epoch": 0.95, - "learning_rate": 1.9859813084112147e-06, - "loss": 3.9364, - "step": 1703 - }, - { - "epoch": 0.96, - "learning_rate": 1.9626168224299064e-06, - "loss": 3.9482, - "step": 1704 - }, - { - "epoch": 0.96, - "learning_rate": 1.939252336448598e-06, - "loss": 4.0321, - "step": 1705 - }, - { - "epoch": 0.96, - "learning_rate": 1.9158878504672894e-06, - "loss": 3.8892, - "step": 1706 - }, - { - "epoch": 0.96, - "learning_rate": 1.892523364485981e-06, - "loss": 3.9527, - "step": 1707 - }, - { - "epoch": 0.96, - "learning_rate": 1.869158878504673e-06, - "loss": 4.1113, - "step": 1708 - }, - { - "epoch": 0.96, - "learning_rate": 1.8457943925233645e-06, - "loss": 4.2602, - "step": 1709 - }, - { - "epoch": 0.96, - "learning_rate": 1.8224299065420562e-06, - "loss": 4.2309, - "step": 1710 - }, - { - "epoch": 0.96, - "learning_rate": 1.7990654205607477e-06, - "loss": 3.9337, - "step": 1711 - }, - { - "epoch": 0.96, - "learning_rate": 1.7757009345794394e-06, - "loss": 4.1777, - "step": 1712 + "learning_rate": 3.3673469387755105e-06, + "loss": 4.0646, + "step": 851 }, { "epoch": 0.96, - "learning_rate": 1.7523364485981308e-06, - "loss": 4.386, - "step": 1713 + "learning_rate": 3.2908163265306124e-06, + "loss": 3.978, + "step": 852 }, { "epoch": 0.96, - "learning_rate": 1.7289719626168225e-06, - "loss": 4.0444, - "step": 1714 + "learning_rate": 3.2142857142857143e-06, + "loss": 4.0582, + "step": 853 }, { "epoch": 0.96, - "learning_rate": 1.705607476635514e-06, - "loss": 4.0592, - "step": 1715 + "learning_rate": 3.1377551020408166e-06, + "loss": 4.0127, + "step": 854 }, { "epoch": 0.96, - "learning_rate": 1.6822429906542055e-06, - "loss": 4.1411, - "step": 1716 + "learning_rate": 3.0612244897959185e-06, + "loss": 3.9874, + "step": 855 }, { "epoch": 0.96, - "learning_rate": 1.6588785046728972e-06, - "loss": 4.5118, - "step": 1717 + "learning_rate": 2.9846938775510205e-06, + "loss": 4.054, + "step": 856 }, { "epoch": 0.96, - "learning_rate": 1.6355140186915887e-06, - "loss": 4.3736, - "step": 1718 + "learning_rate": 2.9081632653061224e-06, + "loss": 4.1327, + "step": 857 }, { "epoch": 0.96, - "learning_rate": 1.6121495327102804e-06, - "loss": 3.9026, - "step": 1719 + "learning_rate": 2.8316326530612243e-06, + "loss": 4.1236, + "step": 858 }, { "epoch": 0.96, - "learning_rate": 1.588785046728972e-06, - "loss": 3.9982, - "step": 1720 + "learning_rate": 2.7551020408163266e-06, + "loss": 3.9831, + "step": 859 }, { "epoch": 0.96, - "learning_rate": 1.5654205607476636e-06, - "loss": 4.1393, - "step": 1721 - }, - { - "epoch": 0.97, - "learning_rate": 1.542056074766355e-06, - "loss": 4.1242, - "step": 1722 - }, - { - "epoch": 0.97, - "learning_rate": 1.5186915887850468e-06, - "loss": 4.261, - "step": 1723 - }, - { - "epoch": 0.97, - "learning_rate": 1.4953271028037383e-06, - "loss": 4.1525, - "step": 1724 - }, - { - "epoch": 0.97, - "learning_rate": 1.47196261682243e-06, - "loss": 4.0871, - "step": 1725 - }, - { - "epoch": 0.97, - "learning_rate": 1.4485981308411215e-06, - "loss": 4.0289, - "step": 1726 - }, - { - "epoch": 0.97, - "learning_rate": 1.425233644859813e-06, - "loss": 4.2446, - "step": 1727 - }, - { - "epoch": 0.97, - "learning_rate": 1.4018691588785047e-06, - "loss": 4.2803, - "step": 1728 - }, - { - "epoch": 0.97, - "learning_rate": 1.3785046728971962e-06, - "loss": 4.3613, - "step": 1729 - }, - { - "epoch": 0.97, - "learning_rate": 1.3551401869158879e-06, - "loss": 4.0993, - "step": 1730 + "learning_rate": 2.6785714285714285e-06, + "loss": 4.0765, + "step": 860 }, { "epoch": 0.97, - "learning_rate": 1.3317757009345794e-06, - "loss": 3.8335, - "step": 1731 + "learning_rate": 2.602040816326531e-06, + "loss": 4.2494, + "step": 861 }, { "epoch": 0.97, - "learning_rate": 1.308411214953271e-06, - "loss": 4.0755, - "step": 1732 + "learning_rate": 2.5255102040816328e-06, + "loss": 3.9901, + "step": 862 }, { "epoch": 0.97, - "learning_rate": 1.2850467289719625e-06, - "loss": 4.0524, - "step": 1733 + "learning_rate": 2.4489795918367347e-06, + "loss": 4.1015, + "step": 863 }, { "epoch": 0.97, - "learning_rate": 1.2616822429906543e-06, - "loss": 4.2421, - "step": 1734 + "learning_rate": 2.372448979591837e-06, + "loss": 4.2779, + "step": 864 }, { "epoch": 0.97, - "learning_rate": 1.2383177570093457e-06, - "loss": 4.5127, - "step": 1735 + "learning_rate": 2.295918367346939e-06, + "loss": 4.1575, + "step": 865 }, { "epoch": 0.97, - "learning_rate": 1.2149532710280374e-06, - "loss": 4.5618, - "step": 1736 + "learning_rate": 2.219387755102041e-06, + "loss": 3.8855, + "step": 866 }, { "epoch": 0.97, - "learning_rate": 1.191588785046729e-06, - "loss": 3.8363, - "step": 1737 + "learning_rate": 2.1428571428571427e-06, + "loss": 4.0976, + "step": 867 }, { "epoch": 0.97, - "learning_rate": 1.1682242990654204e-06, - "loss": 4.1024, - "step": 1738 + "learning_rate": 2.0663265306122446e-06, + "loss": 3.9664, + "step": 868 }, { "epoch": 0.97, - "learning_rate": 1.1448598130841121e-06, - "loss": 4.069, - "step": 1739 - }, - { - "epoch": 0.98, - "learning_rate": 1.1214953271028036e-06, - "loss": 3.9845, - "step": 1740 - }, - { - "epoch": 0.98, - "learning_rate": 1.0981308411214953e-06, - "loss": 4.1161, - "step": 1741 - }, - { - "epoch": 0.98, - "learning_rate": 1.0747663551401868e-06, - "loss": 4.1233, - "step": 1742 - }, - { - "epoch": 0.98, - "learning_rate": 1.0514018691588785e-06, - "loss": 3.9332, - "step": 1743 - }, - { - "epoch": 0.98, - "learning_rate": 1.02803738317757e-06, - "loss": 3.9231, - "step": 1744 - }, - { - "epoch": 0.98, - "learning_rate": 1.0046728971962617e-06, - "loss": 4.1985, - "step": 1745 - }, - { - "epoch": 0.98, - "learning_rate": 9.813084112149532e-07, - "loss": 4.3305, - "step": 1746 - }, - { - "epoch": 0.98, - "learning_rate": 9.579439252336447e-07, - "loss": 4.2018, - "step": 1747 - }, - { - "epoch": 0.98, - "learning_rate": 9.345794392523365e-07, - "loss": 4.1221, - "step": 1748 + "learning_rate": 1.989795918367347e-06, + "loss": 4.0142, + "step": 869 }, { "epoch": 0.98, - "learning_rate": 9.112149532710281e-07, - "loss": 4.2828, - "step": 1749 + "learning_rate": 1.913265306122449e-06, + "loss": 4.1759, + "step": 870 }, { "epoch": 0.98, - "learning_rate": 8.878504672897197e-07, - "loss": 4.0967, - "step": 1750 + "learning_rate": 1.836734693877551e-06, + "loss": 3.9869, + "step": 871 }, { "epoch": 0.98, - "learning_rate": 8.644859813084113e-07, - "loss": 3.8762, - "step": 1751 + "learning_rate": 1.7602040816326531e-06, + "loss": 4.153, + "step": 872 }, { "epoch": 0.98, - "learning_rate": 8.411214953271028e-07, - "loss": 4.087, - "step": 1752 + "learning_rate": 1.6836734693877552e-06, + "loss": 4.1535, + "step": 873 }, { "epoch": 0.98, - "learning_rate": 8.177570093457944e-07, - "loss": 4.2492, - "step": 1753 + "learning_rate": 1.6071428571428572e-06, + "loss": 4.1515, + "step": 874 }, { "epoch": 0.98, - "learning_rate": 7.94392523364486e-07, - "loss": 4.0876, - "step": 1754 + "learning_rate": 1.5306122448979593e-06, + "loss": 4.2312, + "step": 875 }, { "epoch": 0.98, - "learning_rate": 7.710280373831776e-07, - "loss": 4.272, - "step": 1755 + "learning_rate": 1.4540816326530612e-06, + "loss": 3.9983, + "step": 876 }, { "epoch": 0.98, - "learning_rate": 7.476635514018691e-07, - "loss": 3.8828, - "step": 1756 + "learning_rate": 1.3775510204081633e-06, + "loss": 4.0585, + "step": 877 }, { "epoch": 0.98, - "learning_rate": 7.242990654205607e-07, - "loss": 3.9841, - "step": 1757 - }, - { - "epoch": 0.99, - "learning_rate": 7.009345794392523e-07, - "loss": 4.0423, - "step": 1758 - }, - { - "epoch": 0.99, - "learning_rate": 6.775700934579439e-07, - "loss": 4.0956, - "step": 1759 - }, - { - "epoch": 0.99, - "learning_rate": 6.542056074766355e-07, - "loss": 3.964, - "step": 1760 - }, - { - "epoch": 0.99, - "learning_rate": 6.308411214953271e-07, - "loss": 4.2357, - "step": 1761 - }, - { - "epoch": 0.99, - "learning_rate": 6.074766355140187e-07, - "loss": 3.8661, - "step": 1762 - }, - { - "epoch": 0.99, - "learning_rate": 5.841121495327102e-07, - "loss": 4.0183, - "step": 1763 - }, - { - "epoch": 0.99, - "learning_rate": 5.607476635514018e-07, - "loss": 4.233, - "step": 1764 - }, - { - "epoch": 0.99, - "learning_rate": 5.373831775700934e-07, - "loss": 3.8625, - "step": 1765 - }, - { - "epoch": 0.99, - "learning_rate": 5.14018691588785e-07, - "loss": 4.0392, - "step": 1766 + "learning_rate": 1.3010204081632654e-06, + "loss": 4.1122, + "step": 878 }, { "epoch": 0.99, - "learning_rate": 4.906542056074766e-07, - "loss": 3.9741, - "step": 1767 + "learning_rate": 1.2244897959183673e-06, + "loss": 4.2815, + "step": 879 }, { "epoch": 0.99, - "learning_rate": 4.6728971962616824e-07, - "loss": 4.0145, - "step": 1768 + "learning_rate": 1.1479591836734695e-06, + "loss": 4.3714, + "step": 880 }, { "epoch": 0.99, - "learning_rate": 4.4392523364485984e-07, - "loss": 4.0376, - "step": 1769 + "learning_rate": 1.0714285714285714e-06, + "loss": 3.9905, + "step": 881 }, { "epoch": 0.99, - "learning_rate": 4.205607476635514e-07, - "loss": 3.997, - "step": 1770 + "learning_rate": 9.948979591836735e-07, + "loss": 4.1339, + "step": 882 }, { "epoch": 0.99, - "learning_rate": 3.97196261682243e-07, - "loss": 4.1271, - "step": 1771 + "learning_rate": 9.183673469387755e-07, + "loss": 4.0394, + "step": 883 }, { "epoch": 0.99, - "learning_rate": 3.7383177570093457e-07, - "loss": 4.0518, - "step": 1772 + "learning_rate": 8.418367346938776e-07, + "loss": 4.0357, + "step": 884 }, { "epoch": 0.99, - "learning_rate": 3.5046728971962617e-07, - "loss": 4.0569, - "step": 1773 + "learning_rate": 7.653061224489796e-07, + "loss": 4.096, + "step": 885 }, { "epoch": 0.99, - "learning_rate": 3.2710280373831776e-07, - "loss": 4.1645, - "step": 1774 + "learning_rate": 6.887755102040817e-07, + "loss": 4.1527, + "step": 886 }, { "epoch": 0.99, - "learning_rate": 3.0373831775700936e-07, - "loss": 4.0194, - "step": 1775 - }, - { - "epoch": 1.0, - "learning_rate": 2.803738317757009e-07, - "loss": 4.1338, - "step": 1776 - }, - { - "epoch": 1.0, - "learning_rate": 2.570093457943925e-07, - "loss": 4.0394, - "step": 1777 - }, - { - "epoch": 1.0, - "learning_rate": 2.3364485981308412e-07, - "loss": 4.2285, - "step": 1778 - }, - { - "epoch": 1.0, - "learning_rate": 2.102803738317757e-07, - "loss": 4.3856, - "step": 1779 + "learning_rate": 6.122448979591837e-07, + "loss": 4.0574, + "step": 887 }, { "epoch": 1.0, - "learning_rate": 1.8691588785046729e-07, - "loss": 4.042, - "step": 1780 + "learning_rate": 5.357142857142857e-07, + "loss": 4.2464, + "step": 888 }, { "epoch": 1.0, - "learning_rate": 1.6355140186915888e-07, - "loss": 4.397, - "step": 1781 + "learning_rate": 4.5918367346938775e-07, + "loss": 4.2688, + "step": 889 }, { "epoch": 1.0, - "learning_rate": 1.4018691588785045e-07, - "loss": 4.0514, - "step": 1782 + "learning_rate": 3.826530612244898e-07, + "loss": 4.1339, + "step": 890 }, { "epoch": 1.0, - "learning_rate": 1.1682242990654206e-07, - "loss": 4.4125, - "step": 1783 + "learning_rate": 3.0612244897959183e-07, + "loss": 4.3171, + "step": 891 }, { "epoch": 1.0, - "learning_rate": 9.345794392523364e-08, - "loss": 3.5579, - "step": 1784 + "learning_rate": 2.2959183673469388e-07, + "loss": 3.8865, + "step": 892 }, { "epoch": 1.0, - "step": 1784, + "step": 892, "total_flos": 0.0, - "train_loss": 4.24715919291492, - "train_runtime": 8514.9113, - "train_samples_per_second": 3.352, - "train_steps_per_second": 0.21 + "train_loss": 4.249312744012328, + "train_runtime": 6250.8757, + "train_samples_per_second": 4.565, + "train_steps_per_second": 0.143 } ], - "max_steps": 1784, + "max_steps": 892, "num_train_epochs": 1, "total_flos": 0.0, "trial_name": null,