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": 8.9778, + "loss": 9.0999, "step": 1 }, { "epoch": 0.0, "learning_rate": 0.0, - "loss": 9.3204, + "loss": 8.7186, "step": 2 }, { "epoch": 0.0, "learning_rate": 0.0, - "loss": 9.2694, + "loss": 8.6553, "step": 3 }, { "epoch": 0.0, - "learning_rate": 0.0, - "loss": 9.2219, + "learning_rate": 6e-07, + "loss": 9.057, "step": 4 }, { - "epoch": 0.0, + "epoch": 0.01, "learning_rate": 6e-07, - "loss": 8.6268, + "loss": 8.9152, "step": 5 }, { - "epoch": 0.0, + "epoch": 0.01, "learning_rate": 1.2e-06, - "loss": 8.8603, + "loss": 8.6189, "step": 6 }, { - "epoch": 0.0, + "epoch": 0.01, "learning_rate": 1.8e-06, - "loss": 8.6022, + "loss": 8.9645, "step": 7 }, { - "epoch": 0.0, - "learning_rate": 2.4e-06, - "loss": 8.6248, + "epoch": 0.01, + "learning_rate": 1.8e-06, + "loss": 8.6078, "step": 8 }, { "epoch": 0.01, - "learning_rate": 2.9999999999999997e-06, - "loss": 9.1036, + "learning_rate": 2.4e-06, + "loss": 8.5715, "step": 9 }, { "epoch": 0.01, "learning_rate": 2.9999999999999997e-06, - "loss": 8.5279, + "loss": 8.4233, "step": 10 }, { "epoch": 0.01, - "learning_rate": 2.9999999999999997e-06, - "loss": 8.2693, + "learning_rate": 3.6e-06, + "loss": 8.2472, "step": 11 }, { "epoch": 0.01, - "learning_rate": 3.6e-06, - "loss": 8.9331, + "learning_rate": 4.2e-06, + "loss": 8.2332, "step": 12 }, { "epoch": 0.01, - "learning_rate": 4.2e-06, - "loss": 8.4494, + "learning_rate": 4.8e-06, + "loss": 7.9932, "step": 13 }, { - "epoch": 0.01, - "learning_rate": 4.8e-06, - "loss": 8.1951, + "epoch": 0.02, + "learning_rate": 5.399999999999999e-06, + "loss": 8.0419, "step": 14 }, { - "epoch": 0.01, - "learning_rate": 5.399999999999999e-06, - "loss": 8.361, + "epoch": 0.02, + "learning_rate": 5.999999999999999e-06, + "loss": 7.9345, "step": 15 }, { - "epoch": 0.01, - "learning_rate": 5.999999999999999e-06, - "loss": 8.1135, + "epoch": 0.02, + "learning_rate": 6.599999999999999e-06, + "loss": 7.7465, "step": 16 }, { - "epoch": 0.01, - "learning_rate": 6.599999999999999e-06, - "loss": 7.9529, + "epoch": 0.02, + "learning_rate": 7.2e-06, + "loss": 7.6716, "step": 17 }, { - "epoch": 0.01, - "learning_rate": 7.2e-06, - "loss": 7.756, + "epoch": 0.02, + "learning_rate": 7.799999999999998e-06, + "loss": 7.6993, "step": 18 }, { - "epoch": 0.01, - "learning_rate": 7.799999999999998e-06, - "loss": 7.7691, + "epoch": 0.02, + "learning_rate": 8.4e-06, + "loss": 7.5265, "step": 19 }, { - "epoch": 0.01, - "learning_rate": 8.4e-06, - "loss": 7.7333, + "epoch": 0.02, + "learning_rate": 8.999999999999999e-06, + "loss": 7.5412, "step": 20 }, { - "epoch": 0.01, - "learning_rate": 8.999999999999999e-06, - "loss": 7.7099, + "epoch": 0.02, + "learning_rate": 9.6e-06, + "loss": 7.5321, "step": 21 }, { - "epoch": 0.01, - "learning_rate": 9.6e-06, - "loss": 7.9213, + "epoch": 0.02, + "learning_rate": 1.02e-05, + "loss": 7.4543, "step": 22 }, { - "epoch": 0.01, - "learning_rate": 1.02e-05, - "loss": 7.9129, + "epoch": 0.03, + "learning_rate": 1.0799999999999998e-05, + "loss": 7.4675, "step": 23 }, { - "epoch": 0.01, - "learning_rate": 1.0799999999999998e-05, - "loss": 7.4724, + "epoch": 0.03, + "learning_rate": 1.14e-05, + "loss": 7.4239, "step": 24 }, { - "epoch": 0.01, - "learning_rate": 1.14e-05, - "loss": 7.3973, + "epoch": 0.03, + "learning_rate": 1.1999999999999999e-05, + "loss": 7.3541, "step": 25 }, { - "epoch": 0.01, - "learning_rate": 1.1999999999999999e-05, - "loss": 7.3973, + "epoch": 0.03, + "learning_rate": 1.26e-05, + "loss": 7.424, "step": 26 }, { - "epoch": 0.02, - "learning_rate": 1.26e-05, - "loss": 7.4904, + "epoch": 0.03, + "learning_rate": 1.3199999999999997e-05, + "loss": 7.4128, "step": 27 }, { - "epoch": 0.02, - "learning_rate": 1.3199999999999997e-05, - "loss": 7.2232, + "epoch": 0.03, + "learning_rate": 1.3799999999999998e-05, + "loss": 7.1907, "step": 28 }, { - "epoch": 0.02, - "learning_rate": 1.3799999999999998e-05, - "loss": 7.4805, + "epoch": 0.03, + "learning_rate": 1.44e-05, + "loss": 7.3099, "step": 29 }, { - "epoch": 0.02, - "learning_rate": 1.44e-05, - "loss": 7.425, + "epoch": 0.03, + "learning_rate": 1.4999999999999999e-05, + "loss": 7.28, "step": 30 }, { - "epoch": 0.02, - "learning_rate": 1.4999999999999999e-05, - "loss": 7.3732, + "epoch": 0.03, + "learning_rate": 1.5599999999999996e-05, + "loss": 7.3751, "step": 31 }, { - "epoch": 0.02, - "learning_rate": 1.5599999999999996e-05, - "loss": 7.3674, + "epoch": 0.04, + "learning_rate": 1.6199999999999997e-05, + "loss": 7.2808, "step": 32 }, { - "epoch": 0.02, - "learning_rate": 1.6199999999999997e-05, - "loss": 7.3653, + "epoch": 0.04, + "learning_rate": 1.68e-05, + "loss": 7.2691, "step": 33 }, { - "epoch": 0.02, - "learning_rate": 1.68e-05, - "loss": 7.1539, + "epoch": 0.04, + "learning_rate": 1.74e-05, + "loss": 7.173, "step": 34 }, { - "epoch": 0.02, - "learning_rate": 1.74e-05, - "loss": 7.2006, + "epoch": 0.04, + "learning_rate": 1.7999999999999997e-05, + "loss": 7.2658, "step": 35 }, { - "epoch": 0.02, - "learning_rate": 1.7999999999999997e-05, - "loss": 7.3857, + "epoch": 0.04, + "learning_rate": 1.8599999999999998e-05, + "loss": 7.234, "step": 36 }, { - "epoch": 0.02, - "learning_rate": 1.8599999999999998e-05, - "loss": 7.1764, + "epoch": 0.04, + "learning_rate": 1.92e-05, + "loss": 7.2649, "step": 37 }, { - "epoch": 0.02, - "learning_rate": 1.92e-05, - "loss": 7.3351, + "epoch": 0.04, + "learning_rate": 1.98e-05, + "loss": 7.2282, "step": 38 }, { - "epoch": 0.02, - "learning_rate": 1.98e-05, - "loss": 7.3475, + "epoch": 0.04, + "learning_rate": 2.04e-05, + "loss": 7.363, "step": 39 }, { - "epoch": 0.02, - "learning_rate": 2.04e-05, - "loss": 7.2391, + "epoch": 0.04, + "learning_rate": 2.1e-05, + "loss": 7.1886, "step": 40 }, { - "epoch": 0.02, - "learning_rate": 2.1e-05, - "loss": 7.2695, + "epoch": 0.05, + "learning_rate": 2.1599999999999996e-05, + "loss": 7.1061, "step": 41 }, { - "epoch": 0.02, - "learning_rate": 2.1599999999999996e-05, - "loss": 7.1683, + "epoch": 0.05, + "learning_rate": 2.2199999999999998e-05, + "loss": 7.2658, "step": 42 }, { - "epoch": 0.02, - "learning_rate": 2.2199999999999998e-05, - "loss": 7.3831, + "epoch": 0.05, + "learning_rate": 2.28e-05, + "loss": 7.0366, "step": 43 }, { - "epoch": 0.02, - "learning_rate": 2.28e-05, - "loss": 7.1583, + "epoch": 0.05, + "learning_rate": 2.34e-05, + "loss": 7.2335, "step": 44 }, { - "epoch": 0.03, - "learning_rate": 2.34e-05, - "loss": 7.4664, + "epoch": 0.05, + "learning_rate": 2.3999999999999997e-05, + "loss": 7.2617, "step": 45 }, { - "epoch": 0.03, - "learning_rate": 2.3999999999999997e-05, - "loss": 7.2593, + "epoch": 0.05, + "learning_rate": 2.4599999999999998e-05, + "loss": 7.2279, "step": 46 }, { - "epoch": 0.03, - "learning_rate": 2.4599999999999998e-05, - "loss": 7.063, + "epoch": 0.05, + "learning_rate": 2.52e-05, + "loss": 6.9998, "step": 47 }, { - "epoch": 0.03, - "learning_rate": 2.52e-05, - "loss": 7.4411, + "epoch": 0.05, + "learning_rate": 2.5799999999999997e-05, + "loss": 7.0601, "step": 48 }, { - "epoch": 0.03, - "learning_rate": 2.5799999999999997e-05, - "loss": 7.0079, + "epoch": 0.05, + "learning_rate": 2.6399999999999995e-05, + "loss": 6.891, "step": 49 }, { - "epoch": 0.03, - "learning_rate": 2.6399999999999995e-05, - "loss": 6.8929, + "epoch": 0.06, + "learning_rate": 2.6999999999999996e-05, + "loss": 6.7733, "step": 50 }, { - "epoch": 0.03, - "learning_rate": 2.6999999999999996e-05, - "loss": 7.2776, + "epoch": 0.06, + "learning_rate": 2.7599999999999997e-05, + "loss": 7.2612, "step": 51 }, { - "epoch": 0.03, - "learning_rate": 2.7599999999999997e-05, - "loss": 7.3885, + "epoch": 0.06, + "learning_rate": 2.8199999999999998e-05, + "loss": 7.2054, "step": 52 }, { - "epoch": 0.03, - "learning_rate": 2.8199999999999998e-05, - "loss": 7.1351, + "epoch": 0.06, + "learning_rate": 2.88e-05, + "loss": 7.0902, "step": 53 }, { - "epoch": 0.03, - "learning_rate": 2.88e-05, - "loss": 7.2158, + "epoch": 0.06, + "learning_rate": 2.94e-05, + "loss": 7.2092, "step": 54 }, { - "epoch": 0.03, - "learning_rate": 2.94e-05, - "loss": 7.4691, + "epoch": 0.06, + "learning_rate": 2.9999999999999997e-05, + "loss": 7.0408, "step": 55 }, { - "epoch": 0.03, - "learning_rate": 2.9999999999999997e-05, - "loss": 7.163, + "epoch": 0.06, + "learning_rate": 3.06e-05, + "loss": 7.1568, "step": 56 }, { - "epoch": 0.03, - "learning_rate": 3.06e-05, - "loss": 7.1084, + "epoch": 0.06, + "learning_rate": 3.119999999999999e-05, + "loss": 7.1719, "step": 57 }, { - "epoch": 0.03, - "learning_rate": 3.119999999999999e-05, - "loss": 7.02, + "epoch": 0.07, + "learning_rate": 3.1799999999999994e-05, + "loss": 7.2245, "step": 58 }, { - "epoch": 0.03, - "learning_rate": 3.1799999999999994e-05, - "loss": 7.0571, + "epoch": 0.07, + "learning_rate": 3.2399999999999995e-05, + "loss": 7.0008, "step": 59 }, { - "epoch": 0.03, - "learning_rate": 3.2399999999999995e-05, - "loss": 7.0927, + "epoch": 0.07, + "learning_rate": 3.2999999999999996e-05, + "loss": 7.1629, "step": 60 }, { - "epoch": 0.03, - "learning_rate": 3.2999999999999996e-05, - "loss": 7.1893, + "epoch": 0.07, + "learning_rate": 3.36e-05, + "loss": 7.154, "step": 61 }, { - "epoch": 0.03, - "learning_rate": 3.36e-05, - "loss": 7.2026, + "epoch": 0.07, + "learning_rate": 3.42e-05, + "loss": 7.103, "step": 62 }, { - "epoch": 0.04, - "learning_rate": 3.42e-05, - "loss": 7.2578, + "epoch": 0.07, + "learning_rate": 3.48e-05, + "loss": 7.0558, "step": 63 }, { - "epoch": 0.04, - "learning_rate": 3.48e-05, - "loss": 7.1085, + "epoch": 0.07, + "learning_rate": 3.539999999999999e-05, + "loss": 7.1201, "step": 64 }, { - "epoch": 0.04, - "learning_rate": 3.539999999999999e-05, - "loss": 7.0803, + "epoch": 0.07, + "learning_rate": 3.5999999999999994e-05, + "loss": 6.9715, "step": 65 }, { - "epoch": 0.04, - "learning_rate": 3.5999999999999994e-05, - "loss": 7.0584, + "epoch": 0.07, + "learning_rate": 3.6599999999999995e-05, + "loss": 7.2314, "step": 66 }, { - "epoch": 0.04, - "learning_rate": 3.6599999999999995e-05, - "loss": 7.0741, + "epoch": 0.08, + "learning_rate": 3.7199999999999996e-05, + "loss": 7.0264, "step": 67 }, { - "epoch": 0.04, - "learning_rate": 3.7199999999999996e-05, - "loss": 7.0758, + "epoch": 0.08, + "learning_rate": 3.78e-05, + "loss": 6.9733, "step": 68 }, { - "epoch": 0.04, - "learning_rate": 3.78e-05, - "loss": 7.1667, + "epoch": 0.08, + "learning_rate": 3.84e-05, + "loss": 7.0547, "step": 69 }, { - "epoch": 0.04, - "learning_rate": 3.84e-05, - "loss": 6.9501, + "epoch": 0.08, + "learning_rate": 3.9e-05, + "loss": 7.0174, "step": 70 }, { - "epoch": 0.04, - "learning_rate": 3.9e-05, - "loss": 7.0904, + "epoch": 0.08, + "learning_rate": 3.96e-05, + "loss": 7.1657, "step": 71 }, { - "epoch": 0.04, - "learning_rate": 3.96e-05, - "loss": 7.03, + "epoch": 0.08, + "learning_rate": 4.02e-05, + "loss": 7.2093, "step": 72 }, { - "epoch": 0.04, - "learning_rate": 4.02e-05, - "loss": 7.1219, + "epoch": 0.08, + "learning_rate": 4.08e-05, + "loss": 6.9803, "step": 73 }, { - "epoch": 0.04, - "learning_rate": 4.08e-05, - "loss": 7.0043, + "epoch": 0.08, + "learning_rate": 4.14e-05, + "loss": 7.1377, "step": 74 }, { - "epoch": 0.04, - "learning_rate": 4.14e-05, - "loss": 7.257, + "epoch": 0.08, + "learning_rate": 4.2e-05, + "loss": 6.9877, "step": 75 }, { - "epoch": 0.04, - "learning_rate": 4.2e-05, - "loss": 7.0523, + "epoch": 0.09, + "learning_rate": 4.259999999999999e-05, + "loss": 7.0808, "step": 76 }, { - "epoch": 0.04, - "learning_rate": 4.259999999999999e-05, - "loss": 7.1927, + "epoch": 0.09, + "learning_rate": 4.319999999999999e-05, + "loss": 7.0058, "step": 77 }, { - "epoch": 0.04, - "learning_rate": 4.319999999999999e-05, - "loss": 7.0165, + "epoch": 0.09, + "learning_rate": 4.3799999999999994e-05, + "loss": 7.0018, "step": 78 }, { - "epoch": 0.04, - "learning_rate": 4.3799999999999994e-05, - "loss": 6.9658, + "epoch": 0.09, + "learning_rate": 4.4399999999999995e-05, + "loss": 6.953, "step": 79 }, { - "epoch": 0.04, - "learning_rate": 4.4399999999999995e-05, - "loss": 6.9433, + "epoch": 0.09, + "learning_rate": 4.4999999999999996e-05, + "loss": 7.0265, "step": 80 }, { - "epoch": 0.05, - "learning_rate": 4.4999999999999996e-05, - "loss": 7.1195, + "epoch": 0.09, + "learning_rate": 4.56e-05, + "loss": 7.0557, "step": 81 }, { - "epoch": 0.05, - "learning_rate": 4.56e-05, - "loss": 7.2124, + "epoch": 0.09, + "learning_rate": 4.62e-05, + "loss": 6.9811, "step": 82 }, { - "epoch": 0.05, - "learning_rate": 4.62e-05, - "loss": 6.9126, + "epoch": 0.09, + "learning_rate": 4.68e-05, + "loss": 7.1628, "step": 83 }, { - "epoch": 0.05, - "learning_rate": 4.68e-05, - "loss": 7.0298, + "epoch": 0.09, + "learning_rate": 4.7399999999999993e-05, + "loss": 6.9113, "step": 84 }, { - "epoch": 0.05, - "learning_rate": 4.7399999999999993e-05, - "loss": 7.0752, + "epoch": 0.1, + "learning_rate": 4.7999999999999994e-05, + "loss": 6.9547, "step": 85 }, { - "epoch": 0.05, - "learning_rate": 4.7999999999999994e-05, - "loss": 7.0812, + "epoch": 0.1, + "learning_rate": 4.8599999999999995e-05, + "loss": 7.0466, "step": 86 }, { - "epoch": 0.05, - "learning_rate": 4.8599999999999995e-05, - "loss": 7.183, + "epoch": 0.1, + "learning_rate": 4.9199999999999997e-05, + "loss": 6.9777, "step": 87 }, { - "epoch": 0.05, - "learning_rate": 4.9199999999999997e-05, - "loss": 7.0191, + "epoch": 0.1, + "learning_rate": 4.98e-05, + "loss": 6.9685, "step": 88 }, { - "epoch": 0.05, - "learning_rate": 4.98e-05, - "loss": 7.0935, + "epoch": 0.1, + "learning_rate": 5.04e-05, + "loss": 7.0273, "step": 89 }, { - "epoch": 0.05, - "learning_rate": 5.04e-05, - "loss": 6.8812, + "epoch": 0.1, + "learning_rate": 5.1e-05, + "loss": 7.0866, "step": 90 }, { - "epoch": 0.05, - "learning_rate": 5.1e-05, - "loss": 7.1619, + "epoch": 0.1, + "learning_rate": 5.1599999999999994e-05, + "loss": 7.1824, "step": 91 }, { - "epoch": 0.05, - "learning_rate": 5.1599999999999994e-05, - "loss": 6.8071, + "epoch": 0.1, + "learning_rate": 5.2199999999999995e-05, + "loss": 7.0954, "step": 92 }, { - "epoch": 0.05, - "learning_rate": 5.2199999999999995e-05, - "loss": 6.9042, + "epoch": 0.1, + "learning_rate": 5.279999999999999e-05, + "loss": 6.8591, "step": 93 }, { - "epoch": 0.05, - "learning_rate": 5.279999999999999e-05, - "loss": 7.1522, + "epoch": 0.11, + "learning_rate": 5.339999999999999e-05, + "loss": 6.9837, "step": 94 }, { - "epoch": 0.05, - "learning_rate": 5.339999999999999e-05, - "loss": 7.1483, + "epoch": 0.11, + "learning_rate": 5.399999999999999e-05, + "loss": 7.1024, "step": 95 }, { - "epoch": 0.05, - "learning_rate": 5.399999999999999e-05, - "loss": 6.9112, + "epoch": 0.11, + "learning_rate": 5.459999999999999e-05, + "loss": 7.1105, "step": 96 }, { - "epoch": 0.05, - "learning_rate": 5.459999999999999e-05, - "loss": 6.8584, + "epoch": 0.11, + "learning_rate": 5.519999999999999e-05, + "loss": 6.9283, "step": 97 }, { - "epoch": 0.05, - "learning_rate": 5.519999999999999e-05, - "loss": 6.5535, + "epoch": 0.11, + "learning_rate": 5.5799999999999994e-05, + "loss": 6.9465, "step": 98 }, { - "epoch": 0.06, - "learning_rate": 5.5799999999999994e-05, - "loss": 6.8542, + "epoch": 0.11, + "learning_rate": 5.6399999999999995e-05, + "loss": 6.6928, "step": 99 }, { - "epoch": 0.06, - "learning_rate": 5.6399999999999995e-05, - "loss": 6.6788, + "epoch": 0.11, + "learning_rate": 5.6999999999999996e-05, + "loss": 6.3529, "step": 100 }, { - "epoch": 0.06, - "learning_rate": 5.6999999999999996e-05, - "loss": 7.2568, + "epoch": 0.11, + "learning_rate": 5.76e-05, + "loss": 7.2781, "step": 101 }, { - "epoch": 0.06, - "learning_rate": 5.76e-05, - "loss": 7.203, + "epoch": 0.11, + "learning_rate": 5.82e-05, + "loss": 7.2954, "step": 102 }, { - "epoch": 0.06, - "learning_rate": 5.82e-05, - "loss": 7.1932, + "epoch": 0.12, + "learning_rate": 5.88e-05, + "loss": 7.1813, "step": 103 }, { - "epoch": 0.06, - "learning_rate": 5.88e-05, - "loss": 7.1015, + "epoch": 0.12, + "learning_rate": 5.94e-05, + "loss": 7.0713, "step": 104 }, { - "epoch": 0.06, - "learning_rate": 5.94e-05, - "loss": 7.0146, + "epoch": 0.12, + "learning_rate": 5.9999999999999995e-05, + "loss": 6.9738, "step": 105 }, { - "epoch": 0.06, - "learning_rate": 5.9999999999999995e-05, - "loss": 6.9727, + "epoch": 0.12, + "learning_rate": 6.0599999999999996e-05, + "loss": 7.0337, "step": 106 }, { - "epoch": 0.06, - "learning_rate": 6.0599999999999996e-05, - "loss": 7.4335, + "epoch": 0.12, + "learning_rate": 6.12e-05, + "loss": 7.1243, "step": 107 }, { - "epoch": 0.06, - "learning_rate": 6.12e-05, - "loss": 7.074, + "epoch": 0.12, + "learning_rate": 6.18e-05, + "loss": 7.1842, "step": 108 }, { - "epoch": 0.06, - "learning_rate": 6.18e-05, - "loss": 6.9655, + "epoch": 0.12, + "learning_rate": 6.239999999999999e-05, + "loss": 6.8733, "step": 109 }, { - "epoch": 0.06, - "learning_rate": 6.239999999999999e-05, - "loss": 7.133, + "epoch": 0.12, + "learning_rate": 6.299999999999999e-05, + "loss": 7.0055, "step": 110 }, { - "epoch": 0.06, - "learning_rate": 6.299999999999999e-05, - "loss": 6.9319, + "epoch": 0.12, + "learning_rate": 6.359999999999999e-05, + "loss": 7.0452, "step": 111 }, { - "epoch": 0.06, - "learning_rate": 6.359999999999999e-05, - "loss": 6.8698, + "epoch": 0.13, + "learning_rate": 6.419999999999999e-05, + "loss": 7.167, "step": 112 }, { - "epoch": 0.06, - "learning_rate": 6.419999999999999e-05, - "loss": 7.0967, + "epoch": 0.13, + "learning_rate": 6.479999999999999e-05, + "loss": 7.0355, "step": 113 }, { - "epoch": 0.06, - "learning_rate": 6.479999999999999e-05, - "loss": 7.1664, + "epoch": 0.13, + "learning_rate": 6.539999999999999e-05, + "loss": 7.0055, "step": 114 }, { - "epoch": 0.06, - "learning_rate": 6.539999999999999e-05, - "loss": 7.0042, + "epoch": 0.13, + "learning_rate": 6.599999999999999e-05, + "loss": 7.0149, "step": 115 }, { - "epoch": 0.07, - "learning_rate": 6.599999999999999e-05, - "loss": 7.0904, + "epoch": 0.13, + "learning_rate": 6.659999999999999e-05, + "loss": 6.8778, "step": 116 }, { - "epoch": 0.07, - "learning_rate": 6.659999999999999e-05, - "loss": 7.1693, + "epoch": 0.13, + "learning_rate": 6.72e-05, + "loss": 7.0386, "step": 117 }, { - "epoch": 0.07, - "learning_rate": 6.72e-05, - "loss": 6.9144, + "epoch": 0.13, + "learning_rate": 6.78e-05, + "loss": 6.9789, "step": 118 }, { - "epoch": 0.07, - "learning_rate": 6.78e-05, - "loss": 7.033, + "epoch": 0.13, + "learning_rate": 6.84e-05, + "loss": 7.0005, "step": 119 }, { - "epoch": 0.07, - "learning_rate": 6.84e-05, - "loss": 7.0672, + "epoch": 0.13, + "learning_rate": 6.9e-05, + "loss": 7.0509, "step": 120 }, { - "epoch": 0.07, - "learning_rate": 6.9e-05, - "loss": 7.2383, + "epoch": 0.14, + "learning_rate": 6.96e-05, + "loss": 6.9427, "step": 121 }, { - "epoch": 0.07, - "learning_rate": 6.96e-05, - "loss": 7.1445, + "epoch": 0.14, + "learning_rate": 7.02e-05, + "loss": 7.098, "step": 122 }, { - "epoch": 0.07, - "learning_rate": 7.02e-05, - "loss": 7.1164, + "epoch": 0.14, + "learning_rate": 7.079999999999999e-05, + "loss": 6.8362, "step": 123 }, { - "epoch": 0.07, - "learning_rate": 7.079999999999999e-05, - "loss": 7.1511, + "epoch": 0.14, + "learning_rate": 7.139999999999999e-05, + "loss": 6.9948, "step": 124 }, { - "epoch": 0.07, - "learning_rate": 7.139999999999999e-05, - "loss": 6.8093, + "epoch": 0.14, + "learning_rate": 7.199999999999999e-05, + "loss": 7.0379, "step": 125 }, { - "epoch": 0.07, - "learning_rate": 7.199999999999999e-05, - "loss": 7.2817, + "epoch": 0.14, + "learning_rate": 7.259999999999999e-05, + "loss": 7.1541, "step": 126 }, { - "epoch": 0.07, - "learning_rate": 7.259999999999999e-05, - "loss": 7.0185, + "epoch": 0.14, + "learning_rate": 7.319999999999999e-05, + "loss": 7.1344, "step": 127 }, { - "epoch": 0.07, - "learning_rate": 7.319999999999999e-05, - "loss": 7.0554, + "epoch": 0.14, + "learning_rate": 7.379999999999999e-05, + "loss": 7.0745, "step": 128 }, { - "epoch": 0.07, - "learning_rate": 7.379999999999999e-05, - "loss": 6.9324, - "step": 129 + "epoch": 0.14, + "learning_rate": 7.439999999999999e-05, + "loss": 6.8535, + "step": 129 }, { - "epoch": 0.07, - "learning_rate": 7.439999999999999e-05, - "loss": 6.9673, + "epoch": 0.15, + "learning_rate": 7.5e-05, + "loss": 7.0349, "step": 130 }, { - "epoch": 0.07, - "learning_rate": 7.5e-05, - "loss": 7.2394, + "epoch": 0.15, + "learning_rate": 7.56e-05, + "loss": 6.9984, "step": 131 }, { - "epoch": 0.07, - "learning_rate": 7.56e-05, - "loss": 7.0446, + "epoch": 0.15, + "learning_rate": 7.62e-05, + "loss": 7.0613, "step": 132 }, { - "epoch": 0.07, - "learning_rate": 7.62e-05, - "loss": 7.1781, + "epoch": 0.15, + "learning_rate": 7.68e-05, + "loss": 6.9975, "step": 133 }, { - "epoch": 0.08, - "learning_rate": 7.68e-05, - "loss": 6.984, + "epoch": 0.15, + "learning_rate": 7.74e-05, + "loss": 7.0731, "step": 134 }, { - "epoch": 0.08, - "learning_rate": 7.74e-05, - "loss": 6.9574, + "epoch": 0.15, + "learning_rate": 7.8e-05, + "loss": 6.8142, "step": 135 }, { - "epoch": 0.08, - "learning_rate": 7.8e-05, - "loss": 7.2157, + "epoch": 0.15, + "learning_rate": 7.86e-05, + "loss": 6.9361, "step": 136 }, { - "epoch": 0.08, - "learning_rate": 7.86e-05, - "loss": 7.0709, + "epoch": 0.15, + "learning_rate": 7.92e-05, + "loss": 7.1242, "step": 137 }, { - "epoch": 0.08, - "learning_rate": 7.92e-05, - "loss": 7.0112, + "epoch": 0.15, + "learning_rate": 7.98e-05, + "loss": 6.9682, "step": 138 }, { - "epoch": 0.08, - "learning_rate": 7.98e-05, - "loss": 7.1832, + "epoch": 0.16, + "learning_rate": 8.04e-05, + "loss": 6.8986, "step": 139 }, { - "epoch": 0.08, - "learning_rate": 8.04e-05, - "loss": 7.1033, + "epoch": 0.16, + "learning_rate": 8.1e-05, + "loss": 6.8913, "step": 140 }, { - "epoch": 0.08, - "learning_rate": 8.1e-05, - "loss": 7.2643, + "epoch": 0.16, + "learning_rate": 8.16e-05, + "loss": 7.1061, "step": 141 }, { - "epoch": 0.08, - "learning_rate": 8.16e-05, - "loss": 7.2417, + "epoch": 0.16, + "learning_rate": 8.22e-05, + "loss": 6.9993, "step": 142 }, { - "epoch": 0.08, - "learning_rate": 8.22e-05, - "loss": 6.9406, + "epoch": 0.16, + "learning_rate": 8.28e-05, + "loss": 6.9815, "step": 143 }, { - "epoch": 0.08, - "learning_rate": 8.28e-05, - "loss": 7.0778, + "epoch": 0.16, + "learning_rate": 8.34e-05, + "loss": 6.7689, "step": 144 }, { - "epoch": 0.08, - "learning_rate": 8.34e-05, - "loss": 7.1858, + "epoch": 0.16, + "learning_rate": 8.4e-05, + "loss": 6.8309, "step": 145 }, { - "epoch": 0.08, - "learning_rate": 8.4e-05, - "loss": 7.0879, + "epoch": 0.16, + "learning_rate": 8.459999999999998e-05, + "loss": 6.7216, "step": 146 }, { - "epoch": 0.08, - "learning_rate": 8.459999999999998e-05, - "loss": 7.0759, + "epoch": 0.16, + "learning_rate": 8.519999999999998e-05, + "loss": 6.9865, "step": 147 }, { - "epoch": 0.08, - "learning_rate": 8.519999999999998e-05, - "loss": 6.9921, + "epoch": 0.17, + "learning_rate": 8.579999999999998e-05, + "loss": 6.5033, "step": 148 }, { - "epoch": 0.08, - "learning_rate": 8.579999999999998e-05, - "loss": 7.0277, + "epoch": 0.17, + "learning_rate": 8.639999999999999e-05, + "loss": 6.7463, "step": 149 }, { - "epoch": 0.08, - "learning_rate": 8.639999999999999e-05, - "loss": 6.2568, + "epoch": 0.17, + "learning_rate": 8.699999999999999e-05, + "loss": 6.2402, "step": 150 }, { - "epoch": 0.08, - "learning_rate": 8.699999999999999e-05, - "loss": 6.9986, + "epoch": 0.17, + "learning_rate": 8.759999999999999e-05, + "loss": 7.0729, "step": 151 }, { - "epoch": 0.09, - "learning_rate": 8.759999999999999e-05, - "loss": 7.0887, + "epoch": 0.17, + "learning_rate": 8.819999999999999e-05, + "loss": 7.0306, "step": 152 }, { - "epoch": 0.09, - "learning_rate": 8.819999999999999e-05, - "loss": 6.9428, + "epoch": 0.17, + "learning_rate": 8.879999999999999e-05, + "loss": 7.0497, "step": 153 }, { - "epoch": 0.09, - "learning_rate": 8.879999999999999e-05, - "loss": 7.2295, + "epoch": 0.17, + "learning_rate": 8.939999999999999e-05, + "loss": 6.8533, "step": 154 }, { - "epoch": 0.09, - "learning_rate": 8.939999999999999e-05, - "loss": 6.8682, + "epoch": 0.17, + "learning_rate": 8.999999999999999e-05, + "loss": 6.8796, "step": 155 }, { - "epoch": 0.09, - "learning_rate": 8.999999999999999e-05, - "loss": 7.2236, + "epoch": 0.17, + "learning_rate": 9.059999999999999e-05, + "loss": 6.9443, "step": 156 }, { - "epoch": 0.09, - "learning_rate": 9.059999999999999e-05, - "loss": 7.0307, + "epoch": 0.18, + "learning_rate": 9.12e-05, + "loss": 7.1109, "step": 157 }, { - "epoch": 0.09, - "learning_rate": 9.12e-05, - "loss": 7.0102, + "epoch": 0.18, + "learning_rate": 9.18e-05, + "loss": 7.0729, "step": 158 }, { - "epoch": 0.09, - "learning_rate": 9.18e-05, - "loss": 7.043, + "epoch": 0.18, + "learning_rate": 9.24e-05, + "loss": 7.0457, "step": 159 }, { - "epoch": 0.09, - "learning_rate": 9.24e-05, - "loss": 7.1861, + "epoch": 0.18, + "learning_rate": 9.3e-05, + "loss": 7.0171, "step": 160 }, { - "epoch": 0.09, - "learning_rate": 9.3e-05, - "loss": 7.1796, + "epoch": 0.18, + "learning_rate": 9.36e-05, + "loss": 6.8649, "step": 161 }, { - "epoch": 0.09, - "learning_rate": 9.36e-05, - "loss": 7.3017, + "epoch": 0.18, + "learning_rate": 9.419999999999999e-05, + "loss": 6.9131, "step": 162 }, { - "epoch": 0.09, - "learning_rate": 9.419999999999999e-05, - "loss": 7.1031, + "epoch": 0.18, + "learning_rate": 9.479999999999999e-05, + "loss": 6.949, "step": 163 }, { - "epoch": 0.09, - "learning_rate": 9.479999999999999e-05, - "loss": 7.0744, + "epoch": 0.18, + "learning_rate": 9.539999999999999e-05, + "loss": 6.9085, "step": 164 }, { - "epoch": 0.09, - "learning_rate": 9.539999999999999e-05, - "loss": 7.0914, + "epoch": 0.18, + "learning_rate": 9.599999999999999e-05, + "loss": 6.957, "step": 165 }, { - "epoch": 0.09, - "learning_rate": 9.599999999999999e-05, - "loss": 7.1491, + "epoch": 0.19, + "learning_rate": 9.659999999999999e-05, + "loss": 6.9318, "step": 166 }, { - "epoch": 0.09, - "learning_rate": 9.659999999999999e-05, - "loss": 7.118, + "epoch": 0.19, + "learning_rate": 9.719999999999999e-05, + "loss": 6.847, "step": 167 }, { - "epoch": 0.09, - "learning_rate": 9.719999999999999e-05, - "loss": 6.9208, + "epoch": 0.19, + "learning_rate": 9.779999999999999e-05, + "loss": 6.9671, "step": 168 }, { - "epoch": 0.09, - "learning_rate": 9.779999999999999e-05, - "loss": 6.8531, + "epoch": 0.19, + "learning_rate": 9.839999999999999e-05, + "loss": 6.9981, "step": 169 }, { - "epoch": 0.1, - "learning_rate": 9.839999999999999e-05, - "loss": 7.1779, + "epoch": 0.19, + "learning_rate": 9.9e-05, + "loss": 6.9109, "step": 170 }, { - "epoch": 0.1, - "learning_rate": 9.9e-05, - "loss": 7.1118, + "epoch": 0.19, + "learning_rate": 9.96e-05, + "loss": 7.1042, "step": 171 }, { - "epoch": 0.1, - "learning_rate": 9.96e-05, - "loss": 7.0365, + "epoch": 0.19, + "learning_rate": 0.0001002, + "loss": 7.0595, "step": 172 }, { - "epoch": 0.1, - "learning_rate": 0.0001002, - "loss": 7.0296, + "epoch": 0.19, + "learning_rate": 0.0001008, + "loss": 7.0398, "step": 173 }, { - "epoch": 0.1, - "learning_rate": 0.0001008, - "loss": 6.9835, + "epoch": 0.2, + "learning_rate": 0.0001014, + "loss": 6.934, "step": 174 }, { - "epoch": 0.1, - "learning_rate": 0.0001014, - "loss": 7.152, + "epoch": 0.2, + "learning_rate": 0.000102, + "loss": 6.7943, "step": 175 }, { - "epoch": 0.1, - "learning_rate": 0.000102, - "loss": 6.9076, + "epoch": 0.2, + "learning_rate": 0.0001026, + "loss": 6.9293, "step": 176 }, { - "epoch": 0.1, - "learning_rate": 0.0001026, - "loss": 6.9272, + "epoch": 0.2, + "learning_rate": 0.00010319999999999999, + "loss": 6.9352, "step": 177 }, { - "epoch": 0.1, - "learning_rate": 0.00010319999999999999, - "loss": 7.0148, + "epoch": 0.2, + "learning_rate": 0.00010379999999999999, + "loss": 6.82, "step": 178 }, { - "epoch": 0.1, - "learning_rate": 0.00010379999999999999, - "loss": 6.8947, + "epoch": 0.2, + "learning_rate": 0.00010439999999999999, + "loss": 7.03, "step": 179 }, { - "epoch": 0.1, - "learning_rate": 0.00010439999999999999, - "loss": 7.0724, + "epoch": 0.2, + "learning_rate": 0.00010499999999999999, + "loss": 7.015, "step": 180 }, { - "epoch": 0.1, - "learning_rate": 0.00010499999999999999, - "loss": 6.9329, + "epoch": 0.2, + "learning_rate": 0.00010559999999999998, + "loss": 7.0113, "step": 181 }, { - "epoch": 0.1, - "learning_rate": 0.00010559999999999998, - "loss": 6.894, + "epoch": 0.2, + "learning_rate": 0.00010619999999999998, + "loss": 6.8457, "step": 182 }, { - "epoch": 0.1, - "learning_rate": 0.00010619999999999998, - "loss": 7.1505, + "epoch": 0.21, + "learning_rate": 0.00010679999999999998, + "loss": 6.8977, "step": 183 }, { - "epoch": 0.1, - "learning_rate": 0.00010679999999999998, - "loss": 6.834, + "epoch": 0.21, + "learning_rate": 0.00010739999999999998, + "loss": 7.0413, "step": 184 }, { - "epoch": 0.1, - "learning_rate": 0.00010739999999999998, - "loss": 7.0352, + "epoch": 0.21, + "learning_rate": 0.00010799999999999998, + "loss": 6.9299, "step": 185 }, { - "epoch": 0.1, - "learning_rate": 0.00010799999999999998, - "loss": 6.9676, + "epoch": 0.21, + "learning_rate": 0.00010859999999999998, + "loss": 7.0626, "step": 186 }, { - "epoch": 0.1, - "learning_rate": 0.00010859999999999998, - "loss": 6.8961, + "epoch": 0.21, + "learning_rate": 0.00010919999999999998, + "loss": 7.0246, "step": 187 }, { - "epoch": 0.11, - "learning_rate": 0.00010919999999999998, - "loss": 6.9863, + "epoch": 0.21, + "learning_rate": 0.00010979999999999999, + "loss": 6.8238, "step": 188 }, { - "epoch": 0.11, - "learning_rate": 0.00010979999999999999, - "loss": 6.9593, + "epoch": 0.21, + "learning_rate": 0.00011039999999999999, + "loss": 6.8894, "step": 189 }, { - "epoch": 0.11, - "learning_rate": 0.00011039999999999999, - "loss": 6.9157, + "epoch": 0.21, + "learning_rate": 0.00011099999999999999, + "loss": 6.8789, "step": 190 }, { - "epoch": 0.11, - "learning_rate": 0.00011099999999999999, - "loss": 7.1256, + "epoch": 0.21, + "learning_rate": 0.00011159999999999999, + "loss": 6.8802, "step": 191 }, { - "epoch": 0.11, - "learning_rate": 0.00011159999999999999, - "loss": 6.8306, + "epoch": 0.22, + "learning_rate": 0.00011219999999999999, + "loss": 7.0403, "step": 192 }, { - "epoch": 0.11, - "learning_rate": 0.00011219999999999999, - "loss": 6.8756, + "epoch": 0.22, + "learning_rate": 0.00011279999999999999, + "loss": 7.0831, "step": 193 }, { - "epoch": 0.11, - "learning_rate": 0.00011279999999999999, - "loss": 6.9912, + "epoch": 0.22, + "learning_rate": 0.00011339999999999999, + "loss": 6.8587, "step": 194 }, { - "epoch": 0.11, - "learning_rate": 0.00011339999999999999, - "loss": 7.0043, + "epoch": 0.22, + "learning_rate": 0.00011399999999999999, + "loss": 6.6345, "step": 195 }, { - "epoch": 0.11, - "learning_rate": 0.00011399999999999999, - "loss": 6.9428, + "epoch": 0.22, + "learning_rate": 0.0001146, + "loss": 7.0325, "step": 196 }, { - "epoch": 0.11, - "learning_rate": 0.0001146, - "loss": 6.953, + "epoch": 0.22, + "learning_rate": 0.0001152, + "loss": 6.6957, "step": 197 }, { - "epoch": 0.11, - "learning_rate": 0.0001152, - "loss": 6.6743, + "epoch": 0.22, + "learning_rate": 0.0001158, + "loss": 6.626, "step": 198 }, { - "epoch": 0.11, - "learning_rate": 0.0001158, - "loss": 6.67, + "epoch": 0.22, + "learning_rate": 0.0001164, + "loss": 6.5982, "step": 199 }, { - "epoch": 0.11, - "learning_rate": 0.0001164, - "loss": 6.459, + "epoch": 0.22, + "learning_rate": 0.000117, + "loss": 6.2445, "step": 200 }, { - "epoch": 0.11, - "learning_rate": 0.000117, - "loss": 7.2535, + "epoch": 0.23, + "learning_rate": 0.0001176, + "loss": 6.9832, "step": 201 }, { - "epoch": 0.11, - "learning_rate": 0.0001176, - "loss": 6.9968, + "epoch": 0.23, + "learning_rate": 0.0001182, + "loss": 7.0513, "step": 202 }, { - "epoch": 0.11, - "learning_rate": 0.0001182, - "loss": 7.307, + "epoch": 0.23, + "learning_rate": 0.0001188, + "loss": 6.9627, "step": 203 }, { - "epoch": 0.11, - "learning_rate": 0.0001188, - "loss": 6.9003, + "epoch": 0.23, + "learning_rate": 0.0001194, + "loss": 6.9495, "step": 204 }, { - "epoch": 0.11, - "learning_rate": 0.0001194, - "loss": 6.9206, + "epoch": 0.23, + "learning_rate": 0.00011999999999999999, + "loss": 6.9463, "step": 205 }, { - "epoch": 0.12, - "learning_rate": 0.00011999999999999999, - "loss": 7.0771, + "epoch": 0.23, + "learning_rate": 0.00012059999999999999, + "loss": 6.8831, "step": 206 }, { - "epoch": 0.12, - "learning_rate": 0.00012059999999999999, - "loss": 7.097, + "epoch": 0.23, + "learning_rate": 0.00012119999999999999, + "loss": 6.9067, "step": 207 }, { - "epoch": 0.12, - "learning_rate": 0.00012119999999999999, - "loss": 6.9847, + "epoch": 0.23, + "learning_rate": 0.00012179999999999999, + "loss": 6.8835, "step": 208 }, { - "epoch": 0.12, - "learning_rate": 0.00012179999999999999, - "loss": 7.0654, + "epoch": 0.23, + "learning_rate": 0.0001224, + "loss": 6.8963, "step": 209 }, { - "epoch": 0.12, - "learning_rate": 0.0001224, - "loss": 7.1915, + "epoch": 0.24, + "learning_rate": 0.00012299999999999998, + "loss": 6.8928, "step": 210 }, { - "epoch": 0.12, - "learning_rate": 0.00012299999999999998, - "loss": 6.9976, + "epoch": 0.24, + "learning_rate": 0.0001236, + "loss": 6.7872, "step": 211 }, { - "epoch": 0.12, - "learning_rate": 0.0001236, - "loss": 7.1421, + "epoch": 0.24, + "learning_rate": 0.00012419999999999998, + "loss": 6.8213, "step": 212 }, { - "epoch": 0.12, - "learning_rate": 0.00012419999999999998, - "loss": 7.0266, + "epoch": 0.24, + "learning_rate": 0.00012479999999999997, + "loss": 6.8724, "step": 213 }, { - "epoch": 0.12, - "learning_rate": 0.00012479999999999997, - "loss": 6.8724, + "epoch": 0.24, + "learning_rate": 0.00012539999999999999, + "loss": 6.9498, "step": 214 }, { - "epoch": 0.12, - "learning_rate": 0.00012539999999999999, - "loss": 7.1315, + "epoch": 0.24, + "learning_rate": 0.00012599999999999997, + "loss": 6.8644, "step": 215 }, { - "epoch": 0.12, - "learning_rate": 0.00012599999999999997, - "loss": 6.8606, + "epoch": 0.24, + "learning_rate": 0.0001266, + "loss": 6.9756, "step": 216 }, { - "epoch": 0.12, - "learning_rate": 0.0001266, - "loss": 7.0208, + "epoch": 0.24, + "learning_rate": 0.00012719999999999997, + "loss": 6.894, "step": 217 }, { - "epoch": 0.12, - "learning_rate": 0.00012719999999999997, - "loss": 6.9267, + "epoch": 0.24, + "learning_rate": 0.0001278, + "loss": 6.9161, "step": 218 }, { - "epoch": 0.12, - "learning_rate": 0.0001278, - "loss": 6.9895, + "epoch": 0.25, + "learning_rate": 0.00012839999999999998, + "loss": 6.8155, "step": 219 }, { - "epoch": 0.12, - "learning_rate": 0.00012839999999999998, - "loss": 7.0626, + "epoch": 0.25, + "learning_rate": 0.000129, + "loss": 6.9999, "step": 220 }, { - "epoch": 0.12, - "learning_rate": 0.000129, - "loss": 7.1158, + "epoch": 0.25, + "learning_rate": 0.00012959999999999998, + "loss": 6.8851, "step": 221 }, { - "epoch": 0.12, - "learning_rate": 0.00012959999999999998, - "loss": 7.2082, + "epoch": 0.25, + "learning_rate": 0.0001302, + "loss": 6.8, "step": 222 }, { - "epoch": 0.12, - "learning_rate": 0.0001302, - "loss": 6.8897, + "epoch": 0.25, + "learning_rate": 0.00013079999999999998, + "loss": 6.9343, "step": 223 }, { - "epoch": 0.13, - "learning_rate": 0.00013079999999999998, - "loss": 7.0155, + "epoch": 0.25, + "learning_rate": 0.0001314, + "loss": 6.8666, "step": 224 }, { - "epoch": 0.13, - "learning_rate": 0.0001314, - "loss": 7.1008, + "epoch": 0.25, + "learning_rate": 0.00013199999999999998, + "loss": 6.898, "step": 225 }, { - "epoch": 0.13, - "learning_rate": 0.00013199999999999998, - "loss": 7.3372, + "epoch": 0.25, + "learning_rate": 0.0001326, + "loss": 7.059, "step": 226 }, { - "epoch": 0.13, - "learning_rate": 0.0001326, - "loss": 7.23, + "epoch": 0.25, + "learning_rate": 0.00013319999999999999, + "loss": 6.8504, "step": 227 }, { - "epoch": 0.13, - "learning_rate": 0.00013319999999999999, - "loss": 7.1338, + "epoch": 0.26, + "learning_rate": 0.0001338, + "loss": 6.8917, "step": 228 }, { - "epoch": 0.13, - "learning_rate": 0.0001338, - "loss": 6.9769, + "epoch": 0.26, + "learning_rate": 0.0001344, + "loss": 6.9666, "step": 229 }, { - "epoch": 0.13, - "learning_rate": 0.0001344, - "loss": 7.0106, + "epoch": 0.26, + "learning_rate": 0.000135, + "loss": 7.0132, "step": 230 }, { - "epoch": 0.13, - "learning_rate": 0.000135, - "loss": 7.0053, + "epoch": 0.26, + "learning_rate": 0.0001356, + "loss": 6.8798, "step": 231 }, { - "epoch": 0.13, - "learning_rate": 0.0001356, - "loss": 7.111, + "epoch": 0.26, + "learning_rate": 0.0001362, + "loss": 7.0676, "step": 232 }, { - "epoch": 0.13, - "learning_rate": 0.0001362, - "loss": 7.2376, + "epoch": 0.26, + "learning_rate": 0.0001368, + "loss": 6.9102, "step": 233 }, { - "epoch": 0.13, - "learning_rate": 0.0001368, - "loss": 7.0327, + "epoch": 0.26, + "learning_rate": 0.0001374, + "loss": 6.9132, "step": 234 }, { - "epoch": 0.13, - "learning_rate": 0.0001374, - "loss": 6.8749, + "epoch": 0.26, + "learning_rate": 0.000138, + "loss": 7.0219, "step": 235 }, { - "epoch": 0.13, - "learning_rate": 0.000138, - "loss": 7.2692, + "epoch": 0.26, + "learning_rate": 0.0001386, + "loss": 6.8001, "step": 236 }, { - "epoch": 0.13, - "learning_rate": 0.0001386, - "loss": 7.1072, + "epoch": 0.27, + "learning_rate": 0.0001392, + "loss": 6.8554, "step": 237 }, { - "epoch": 0.13, - "learning_rate": 0.0001392, - "loss": 7.0829, + "epoch": 0.27, + "learning_rate": 0.00013979999999999998, + "loss": 7.0862, "step": 238 }, { - "epoch": 0.13, - "learning_rate": 0.00013979999999999998, - "loss": 7.0216, + "epoch": 0.27, + "learning_rate": 0.0001404, + "loss": 6.7867, "step": 239 }, { - "epoch": 0.13, - "learning_rate": 0.0001404, - "loss": 7.0167, + "epoch": 0.27, + "learning_rate": 0.00014099999999999998, + "loss": 6.8978, "step": 240 }, { - "epoch": 0.14, - "learning_rate": 0.00014099999999999998, - "loss": 7.0276, + "epoch": 0.27, + "learning_rate": 0.00014159999999999997, + "loss": 6.8487, "step": 241 }, { - "epoch": 0.14, - "learning_rate": 0.00014159999999999997, - "loss": 6.9919, + "epoch": 0.27, + "learning_rate": 0.0001422, + "loss": 6.986, "step": 242 }, { - "epoch": 0.14, - "learning_rate": 0.0001422, - "loss": 6.884, + "epoch": 0.27, + "learning_rate": 0.00014279999999999997, + "loss": 6.8579, "step": 243 }, { - "epoch": 0.14, - "learning_rate": 0.00014279999999999997, - "loss": 7.0359, + "epoch": 0.27, + "learning_rate": 0.0001434, + "loss": 6.736, "step": 244 }, { - "epoch": 0.14, - "learning_rate": 0.0001434, - "loss": 6.9716, + "epoch": 0.27, + "learning_rate": 0.00014399999999999998, + "loss": 6.8802, "step": 245 }, { - "epoch": 0.14, - "learning_rate": 0.00014399999999999998, - "loss": 7.077, + "epoch": 0.28, + "learning_rate": 0.0001446, + "loss": 6.8379, "step": 246 }, { - "epoch": 0.14, - "learning_rate": 0.0001446, - "loss": 6.5391, + "epoch": 0.28, + "learning_rate": 0.00014519999999999998, + "loss": 6.9125, "step": 247 }, { - "epoch": 0.14, - "learning_rate": 0.00014519999999999998, - "loss": 6.9454, + "epoch": 0.28, + "learning_rate": 0.0001458, + "loss": 6.9465, "step": 248 }, { - "epoch": 0.14, - "learning_rate": 0.0001458, - "loss": 6.6715, + "epoch": 0.28, + "learning_rate": 0.00014639999999999998, + "loss": 6.7558, "step": 249 }, { - "epoch": 0.14, - "learning_rate": 0.00014639999999999998, - "loss": 6.5271, + "epoch": 0.28, + "learning_rate": 0.000147, + "loss": 6.5203, "step": 250 }, { - "epoch": 0.14, - "learning_rate": 0.000147, - "loss": 7.1498, + "epoch": 0.28, + "learning_rate": 0.00014759999999999998, + "loss": 7.1518, "step": 251 }, { - "epoch": 0.14, - "learning_rate": 0.00014759999999999998, - "loss": 7.4529, + "epoch": 0.28, + "learning_rate": 0.0001482, + "loss": 7.0714, "step": 252 }, { - "epoch": 0.14, - "learning_rate": 0.0001482, - "loss": 7.0867, + "epoch": 0.28, + "learning_rate": 0.00014879999999999998, + "loss": 6.9628, "step": 253 }, { - "epoch": 0.14, - "learning_rate": 0.00014879999999999998, - "loss": 7.2117, + "epoch": 0.28, + "learning_rate": 0.0001494, + "loss": 7.0631, "step": 254 }, { - "epoch": 0.14, - "learning_rate": 0.0001494, - "loss": 7.0941, + "epoch": 0.29, + "learning_rate": 0.00015, + "loss": 6.8613, "step": 255 }, { - "epoch": 0.14, - "learning_rate": 0.00015, - "loss": 7.0979, + "epoch": 0.29, + "learning_rate": 0.00015059999999999997, + "loss": 7.0024, "step": 256 }, { - "epoch": 0.14, - "learning_rate": 0.00015059999999999997, - "loss": 7.1284, + "epoch": 0.29, + "learning_rate": 0.0001512, + "loss": 6.9627, "step": 257 }, { - "epoch": 0.14, - "learning_rate": 0.0001512, - "loss": 7.0766, + "epoch": 0.29, + "learning_rate": 0.00015179999999999998, + "loss": 6.9489, "step": 258 }, { - "epoch": 0.15, - "learning_rate": 0.00015179999999999998, - "loss": 7.327, + "epoch": 0.29, + "learning_rate": 0.0001524, + "loss": 6.8986, "step": 259 }, { - "epoch": 0.15, - "learning_rate": 0.0001524, - "loss": 6.803, + "epoch": 0.29, + "learning_rate": 0.00015299999999999998, + "loss": 7.0233, "step": 260 }, { - "epoch": 0.15, - "learning_rate": 0.00015299999999999998, - "loss": 6.9529, + "epoch": 0.29, + "learning_rate": 0.0001536, + "loss": 6.9037, "step": 261 }, { - "epoch": 0.15, - "learning_rate": 0.0001536, - "loss": 6.9829, + "epoch": 0.29, + "learning_rate": 0.00015419999999999998, + "loss": 6.8283, "step": 262 }, { - "epoch": 0.15, - "learning_rate": 0.00015419999999999998, - "loss": 7.2449, + "epoch": 0.29, + "learning_rate": 0.0001548, + "loss": 6.9515, "step": 263 }, { - "epoch": 0.15, - "learning_rate": 0.0001548, - "loss": 7.2041, + "epoch": 0.3, + "learning_rate": 0.00015539999999999998, + "loss": 6.9285, "step": 264 }, { - "epoch": 0.15, - "learning_rate": 0.00015539999999999998, - "loss": 6.9986, + "epoch": 0.3, + "learning_rate": 0.000156, + "loss": 6.8905, "step": 265 }, { - "epoch": 0.15, - "learning_rate": 0.000156, - "loss": 6.8876, + "epoch": 0.3, + "learning_rate": 0.00015659999999999998, + "loss": 7.0416, "step": 266 }, { - "epoch": 0.15, - "learning_rate": 0.00015659999999999998, - "loss": 7.1312, + "epoch": 0.3, + "learning_rate": 0.0001572, + "loss": 7.0372, "step": 267 }, { - "epoch": 0.15, - "learning_rate": 0.0001572, - "loss": 7.0652, + "epoch": 0.3, + "learning_rate": 0.0001578, + "loss": 6.9921, "step": 268 }, { - "epoch": 0.15, - "learning_rate": 0.0001578, - "loss": 7.1611, + "epoch": 0.3, + "learning_rate": 0.0001584, + "loss": 6.888, "step": 269 }, { - "epoch": 0.15, - "learning_rate": 0.0001584, - "loss": 6.9164, + "epoch": 0.3, + "learning_rate": 0.000159, + "loss": 6.9847, "step": 270 }, { - "epoch": 0.15, - "learning_rate": 0.000159, - "loss": 6.9852, + "epoch": 0.3, + "learning_rate": 0.0001596, + "loss": 6.8239, "step": 271 }, { - "epoch": 0.15, - "learning_rate": 0.0001596, - "loss": 7.0395, + "epoch": 0.3, + "learning_rate": 0.0001602, + "loss": 6.8453, "step": 272 }, { - "epoch": 0.15, - "learning_rate": 0.0001602, - "loss": 6.9035, + "epoch": 0.31, + "learning_rate": 0.0001608, + "loss": 6.8599, "step": 273 }, { - "epoch": 0.15, - "learning_rate": 0.0001608, - "loss": 7.0151, + "epoch": 0.31, + "learning_rate": 0.0001614, + "loss": 6.7597, "step": 274 }, { - "epoch": 0.15, - "learning_rate": 0.0001614, - "loss": 7.1616, + "epoch": 0.31, + "learning_rate": 0.000162, + "loss": 7.0201, "step": 275 }, { - "epoch": 0.15, - "learning_rate": 0.000162, - "loss": 7.2044, + "epoch": 0.31, + "learning_rate": 0.0001626, + "loss": 6.873, "step": 276 }, { - "epoch": 0.16, - "learning_rate": 0.0001626, - "loss": 7.0652, + "epoch": 0.31, + "learning_rate": 0.0001632, + "loss": 6.9579, "step": 277 }, { - "epoch": 0.16, - "learning_rate": 0.0001632, - "loss": 7.2084, + "epoch": 0.31, + "learning_rate": 0.0001638, + "loss": 6.8859, "step": 278 }, { - "epoch": 0.16, - "learning_rate": 0.0001638, - "loss": 6.9941, + "epoch": 0.31, + "learning_rate": 0.0001644, + "loss": 6.9457, "step": 279 }, { - "epoch": 0.16, - "learning_rate": 0.0001644, - "loss": 7.0366, + "epoch": 0.31, + "learning_rate": 0.000165, + "loss": 7.0055, "step": 280 }, { - "epoch": 0.16, - "learning_rate": 0.000165, - "loss": 7.0654, + "epoch": 0.32, + "learning_rate": 0.0001656, + "loss": 6.7921, "step": 281 }, { - "epoch": 0.16, - "learning_rate": 0.0001656, - "loss": 7.2163, + "epoch": 0.32, + "learning_rate": 0.0001662, + "loss": 6.9926, "step": 282 }, { - "epoch": 0.16, - "learning_rate": 0.0001662, - "loss": 7.1591, + "epoch": 0.32, + "learning_rate": 0.0001668, + "loss": 6.6923, "step": 283 }, { - "epoch": 0.16, - "learning_rate": 0.0001668, - "loss": 6.7827, + "epoch": 0.32, + "learning_rate": 0.0001674, + "loss": 7.0047, "step": 284 }, { - "epoch": 0.16, - "learning_rate": 0.0001674, - "loss": 7.1517, + "epoch": 0.32, + "learning_rate": 0.000168, + "loss": 6.9741, "step": 285 }, { - "epoch": 0.16, - "learning_rate": 0.000168, - "loss": 6.89, + "epoch": 0.32, + "learning_rate": 0.0001686, + "loss": 6.7653, "step": 286 }, { - "epoch": 0.16, - "learning_rate": 0.0001686, - "loss": 6.9508, + "epoch": 0.32, + "learning_rate": 0.00016919999999999997, + "loss": 6.9668, "step": 287 }, { - "epoch": 0.16, - "learning_rate": 0.00016919999999999997, - "loss": 7.2212, + "epoch": 0.32, + "learning_rate": 0.00016979999999999998, + "loss": 6.8915, "step": 288 }, { - "epoch": 0.16, - "learning_rate": 0.00016979999999999998, - "loss": 6.9056, + "epoch": 0.32, + "learning_rate": 0.00017039999999999997, + "loss": 7.0164, "step": 289 }, { - "epoch": 0.16, - "learning_rate": 0.00017039999999999997, - "loss": 7.1479, + "epoch": 0.33, + "learning_rate": 0.00017099999999999998, + "loss": 7.0853, "step": 290 }, { - "epoch": 0.16, - "learning_rate": 0.00017099999999999998, - "loss": 7.1261, + "epoch": 0.33, + "learning_rate": 0.00017159999999999997, + "loss": 7.002, "step": 291 }, { - "epoch": 0.16, - "learning_rate": 0.00017159999999999997, - "loss": 7.2645, + "epoch": 0.33, + "learning_rate": 0.00017219999999999998, + "loss": 6.922, "step": 292 }, { - "epoch": 0.16, - "learning_rate": 0.00017219999999999998, - "loss": 7.0333, + "epoch": 0.33, + "learning_rate": 0.00017279999999999997, + "loss": 6.8444, "step": 293 }, { - "epoch": 0.16, - "learning_rate": 0.00017279999999999997, - "loss": 7.077, + "epoch": 0.33, + "learning_rate": 0.00017339999999999996, + "loss": 6.7917, "step": 294 }, { - "epoch": 0.17, - "learning_rate": 0.00017339999999999996, - "loss": 6.8987, + "epoch": 0.33, + "learning_rate": 0.00017399999999999997, + "loss": 6.6621, "step": 295 }, { - "epoch": 0.17, - "learning_rate": 0.00017399999999999997, - "loss": 6.8211, + "epoch": 0.33, + "learning_rate": 0.00017459999999999996, + "loss": 6.7716, "step": 296 }, { - "epoch": 0.17, - "learning_rate": 0.00017459999999999996, - "loss": 6.8077, + "epoch": 0.33, + "learning_rate": 0.00017519999999999998, + "loss": 6.7518, "step": 297 }, { - "epoch": 0.17, - "learning_rate": 0.00017519999999999998, - "loss": 7.1395, + "epoch": 0.33, + "learning_rate": 0.00017579999999999996, + "loss": 6.5265, "step": 298 }, { - "epoch": 0.17, - "learning_rate": 0.00017579999999999996, - "loss": 6.7953, + "epoch": 0.34, + "learning_rate": 0.00017639999999999998, + "loss": 6.5202, "step": 299 }, { - "epoch": 0.17, - "learning_rate": 0.00017639999999999998, - "loss": 6.3837, + "epoch": 0.34, + "learning_rate": 0.00017699999999999997, + "loss": 6.3254, "step": 300 }, { - "epoch": 0.17, - "learning_rate": 0.00017699999999999997, - "loss": 7.0741, + "epoch": 0.34, + "learning_rate": 0.00017759999999999998, + "loss": 7.0382, "step": 301 }, { - "epoch": 0.17, - "learning_rate": 0.00017759999999999998, - "loss": 6.9875, + "epoch": 0.34, + "learning_rate": 0.00017819999999999997, + "loss": 7.019, "step": 302 }, { - "epoch": 0.17, - "learning_rate": 0.00017819999999999997, - "loss": 7.0582, + "epoch": 0.34, + "learning_rate": 0.00017879999999999998, + "loss": 6.9302, "step": 303 }, { - "epoch": 0.17, - "learning_rate": 0.00017879999999999998, - "loss": 6.9265, + "epoch": 0.34, + "learning_rate": 0.00017939999999999997, + "loss": 7.0206, "step": 304 }, { - "epoch": 0.17, - "learning_rate": 0.00017939999999999997, - "loss": 7.0881, + "epoch": 0.34, + "learning_rate": 0.00017999999999999998, + "loss": 6.7929, "step": 305 }, { - "epoch": 0.17, - "learning_rate": 0.00017999999999999998, - "loss": 7.0218, - "step": 306 + "epoch": 0.34, + "learning_rate": 0.00018059999999999997, + "loss": 6.8614, + "step": 306 }, { - "epoch": 0.17, - "learning_rate": 0.00018059999999999997, - "loss": 7.0922, + "epoch": 0.34, + "learning_rate": 0.00018119999999999999, + "loss": 6.9282, "step": 307 }, { - "epoch": 0.17, - "learning_rate": 0.00018119999999999999, - "loss": 7.2078, + "epoch": 0.35, + "learning_rate": 0.00018179999999999997, + "loss": 6.9358, "step": 308 }, { - "epoch": 0.17, - "learning_rate": 0.00018179999999999997, - "loss": 7.0497, + "epoch": 0.35, + "learning_rate": 0.0001824, + "loss": 6.8621, "step": 309 }, { - "epoch": 0.17, - "learning_rate": 0.0001824, - "loss": 7.2077, + "epoch": 0.35, + "learning_rate": 0.00018299999999999998, + "loss": 6.9782, "step": 310 }, { - "epoch": 0.17, - "learning_rate": 0.00018299999999999998, - "loss": 6.9793, + "epoch": 0.35, + "learning_rate": 0.0001836, + "loss": 7.0606, "step": 311 }, { - "epoch": 0.17, - "learning_rate": 0.0001836, - "loss": 7.1166, + "epoch": 0.35, + "learning_rate": 0.00018419999999999998, + "loss": 6.9457, "step": 312 }, { - "epoch": 0.18, - "learning_rate": 0.00018419999999999998, - "loss": 7.0466, + "epoch": 0.35, + "learning_rate": 0.0001848, + "loss": 6.9702, "step": 313 }, { - "epoch": 0.18, - "learning_rate": 0.0001848, - "loss": 7.0704, + "epoch": 0.35, + "learning_rate": 0.00018539999999999998, + "loss": 6.9048, "step": 314 }, { - "epoch": 0.18, - "learning_rate": 0.00018539999999999998, - "loss": 7.1107, + "epoch": 0.35, + "learning_rate": 0.000186, + "loss": 6.796, "step": 315 }, { - "epoch": 0.18, - "learning_rate": 0.000186, - "loss": 6.925, + "epoch": 0.35, + "learning_rate": 0.00018659999999999998, + "loss": 7.0186, "step": 316 }, { - "epoch": 0.18, - "learning_rate": 0.00018659999999999998, - "loss": 6.9426, + "epoch": 0.36, + "learning_rate": 0.0001872, + "loss": 6.9154, "step": 317 }, { - "epoch": 0.18, - "learning_rate": 0.0001872, - "loss": 6.9727, + "epoch": 0.36, + "learning_rate": 0.00018779999999999998, + "loss": 6.8215, "step": 318 }, { - "epoch": 0.18, - "learning_rate": 0.00018779999999999998, - "loss": 7.1612, + "epoch": 0.36, + "learning_rate": 0.00018839999999999997, + "loss": 6.7335, "step": 319 }, { - "epoch": 0.18, - "learning_rate": 0.00018839999999999997, - "loss": 7.1232, + "epoch": 0.36, + "learning_rate": 0.00018899999999999999, + "loss": 6.9722, "step": 320 }, { - "epoch": 0.18, - "learning_rate": 0.00018899999999999999, - "loss": 7.3173, + "epoch": 0.36, + "learning_rate": 0.00018959999999999997, + "loss": 6.922, "step": 321 }, { - "epoch": 0.18, - "learning_rate": 0.00018959999999999997, - "loss": 7.225, + "epoch": 0.36, + "learning_rate": 0.0001902, + "loss": 6.7933, "step": 322 }, { - "epoch": 0.18, - "learning_rate": 0.0001902, - "loss": 7.263, + "epoch": 0.36, + "learning_rate": 0.00019079999999999998, + "loss": 6.9454, "step": 323 }, { - "epoch": 0.18, - "learning_rate": 0.00019079999999999998, - "loss": 7.0181, + "epoch": 0.36, + "learning_rate": 0.0001914, + "loss": 6.9225, "step": 324 }, { - "epoch": 0.18, - "learning_rate": 0.0001914, - "loss": 6.9233, + "epoch": 0.36, + "learning_rate": 0.00019199999999999998, + "loss": 6.9729, "step": 325 }, { - "epoch": 0.18, - "learning_rate": 0.00019199999999999998, - "loss": 7.1751, + "epoch": 0.37, + "learning_rate": 0.0001926, + "loss": 6.9193, "step": 326 }, { - "epoch": 0.18, - "learning_rate": 0.0001926, - "loss": 7.0304, + "epoch": 0.37, + "learning_rate": 0.00019319999999999998, + "loss": 6.8074, "step": 327 }, { - "epoch": 0.18, - "learning_rate": 0.00019319999999999998, - "loss": 7.0526, + "epoch": 0.37, + "learning_rate": 0.0001938, + "loss": 6.9116, "step": 328 }, { - "epoch": 0.18, - "learning_rate": 0.0001938, - "loss": 7.0126, + "epoch": 0.37, + "learning_rate": 0.00019439999999999998, + "loss": 6.9647, "step": 329 }, { - "epoch": 0.18, - "learning_rate": 0.00019439999999999998, - "loss": 7.1458, + "epoch": 0.37, + "learning_rate": 0.000195, + "loss": 6.9612, "step": 330 }, { - "epoch": 0.19, - "learning_rate": 0.000195, - "loss": 7.0605, + "epoch": 0.37, + "learning_rate": 0.00019559999999999998, + "loss": 6.9816, "step": 331 }, { - "epoch": 0.19, - "learning_rate": 0.00019559999999999998, - "loss": 7.059, + "epoch": 0.37, + "learning_rate": 0.0001962, + "loss": 6.7078, "step": 332 }, { - "epoch": 0.19, - "learning_rate": 0.0001962, - "loss": 7.2207, + "epoch": 0.37, + "learning_rate": 0.00019679999999999999, + "loss": 6.8303, "step": 333 }, { - "epoch": 0.19, - "learning_rate": 0.00019679999999999999, - "loss": 7.1819, + "epoch": 0.37, + "learning_rate": 0.0001974, + "loss": 6.9353, "step": 334 }, { - "epoch": 0.19, - "learning_rate": 0.0001974, - "loss": 7.0484, + "epoch": 0.38, + "learning_rate": 0.000198, + "loss": 6.9853, "step": 335 }, { - "epoch": 0.19, - "learning_rate": 0.000198, - "loss": 7.075, + "epoch": 0.38, + "learning_rate": 0.0001986, + "loss": 6.7819, "step": 336 }, { - "epoch": 0.19, - "learning_rate": 0.0001986, - "loss": 7.0644, + "epoch": 0.38, + "learning_rate": 0.0001992, + "loss": 6.7857, "step": 337 }, { - "epoch": 0.19, - "learning_rate": 0.0001992, - "loss": 6.8872, + "epoch": 0.38, + "learning_rate": 0.0001998, + "loss": 6.9085, "step": 338 }, { - "epoch": 0.19, - "learning_rate": 0.0001998, - "loss": 6.996, + "epoch": 0.38, + "learning_rate": 0.0002004, + "loss": 6.9153, "step": 339 }, { - "epoch": 0.19, - "learning_rate": 0.0002004, - "loss": 7.0572, + "epoch": 0.38, + "learning_rate": 0.000201, + "loss": 6.8557, "step": 340 }, { - "epoch": 0.19, - "learning_rate": 0.000201, - "loss": 6.8383, + "epoch": 0.38, + "learning_rate": 0.0002016, + "loss": 6.9274, "step": 341 }, { - "epoch": 0.19, - "learning_rate": 0.0002016, - "loss": 7.1998, + "epoch": 0.38, + "learning_rate": 0.0002022, + "loss": 6.8212, "step": 342 }, { - "epoch": 0.19, - "learning_rate": 0.0002022, - "loss": 7.1556, + "epoch": 0.38, + "learning_rate": 0.0002028, + "loss": 6.7922, "step": 343 }, { - "epoch": 0.19, - "learning_rate": 0.0002028, - "loss": 7.0771, + "epoch": 0.39, + "learning_rate": 0.00020339999999999998, + "loss": 6.7738, "step": 344 }, { - "epoch": 0.19, - "learning_rate": 0.00020339999999999998, - "loss": 6.8129, + "epoch": 0.39, + "learning_rate": 0.000204, + "loss": 6.6782, "step": 345 }, { - "epoch": 0.19, - "learning_rate": 0.000204, - "loss": 6.8826, + "epoch": 0.39, + "learning_rate": 0.00020459999999999999, + "loss": 6.765, "step": 346 }, { - "epoch": 0.19, - "learning_rate": 0.00020459999999999999, - "loss": 7.4772, + "epoch": 0.39, + "learning_rate": 0.0002052, + "loss": 6.7327, "step": 347 }, { - "epoch": 0.2, - "learning_rate": 0.0002052, - "loss": 6.8934, + "epoch": 0.39, + "learning_rate": 0.0002058, + "loss": 6.7604, "step": 348 }, { - "epoch": 0.2, - "learning_rate": 0.0002058, - "loss": 6.4536, + "epoch": 0.39, + "learning_rate": 0.00020639999999999998, + "loss": 6.6491, "step": 349 }, { - "epoch": 0.2, - "learning_rate": 0.00020639999999999998, - "loss": 6.2402, + "epoch": 0.39, + "learning_rate": 0.00020699999999999996, + "loss": 6.5297, "step": 350 }, { - "epoch": 0.2, - "learning_rate": 0.00020699999999999996, - "loss": 7.1799, + "epoch": 0.39, + "learning_rate": 0.00020759999999999998, + "loss": 6.9381, "step": 351 }, { - "epoch": 0.2, - "learning_rate": 0.00020759999999999998, - "loss": 7.2743, + "epoch": 0.39, + "learning_rate": 0.00020819999999999996, + "loss": 6.9526, "step": 352 }, { - "epoch": 0.2, - "learning_rate": 0.00020819999999999996, - "loss": 7.1228, + "epoch": 0.4, + "learning_rate": 0.00020879999999999998, + "loss": 6.9151, "step": 353 }, { - "epoch": 0.2, - "learning_rate": 0.00020879999999999998, - "loss": 7.0324, + "epoch": 0.4, + "learning_rate": 0.00020939999999999997, + "loss": 6.8661, "step": 354 }, { - "epoch": 0.2, - "learning_rate": 0.00020939999999999997, - "loss": 6.9689, + "epoch": 0.4, + "learning_rate": 0.00020999999999999998, + "loss": 6.9779, "step": 355 }, { - "epoch": 0.2, - "learning_rate": 0.00020999999999999998, - "loss": 7.1384, + "epoch": 0.4, + "learning_rate": 0.00021059999999999997, + "loss": 6.8233, "step": 356 }, { - "epoch": 0.2, - "learning_rate": 0.00021059999999999997, - "loss": 7.2514, + "epoch": 0.4, + "learning_rate": 0.00021119999999999996, + "loss": 7.0543, "step": 357 }, { - "epoch": 0.2, - "learning_rate": 0.00021119999999999996, - "loss": 7.0969, + "epoch": 0.4, + "learning_rate": 0.00021179999999999997, + "loss": 6.8327, "step": 358 }, { - "epoch": 0.2, - "learning_rate": 0.00021179999999999997, - "loss": 7.2459, + "epoch": 0.4, + "learning_rate": 0.00021239999999999996, + "loss": 6.937, "step": 359 }, { - "epoch": 0.2, - "learning_rate": 0.00021239999999999996, - "loss": 7.1452, + "epoch": 0.4, + "learning_rate": 0.00021299999999999997, + "loss": 6.7641, "step": 360 }, { - "epoch": 0.2, - "learning_rate": 0.00021299999999999997, - "loss": 6.9236, + "epoch": 0.4, + "learning_rate": 0.00021359999999999996, + "loss": 6.8524, "step": 361 }, { - "epoch": 0.2, - "learning_rate": 0.00021359999999999996, - "loss": 6.8407, + "epoch": 0.41, + "learning_rate": 0.00021419999999999998, + "loss": 6.9425, "step": 362 }, { - "epoch": 0.2, - "learning_rate": 0.00021419999999999998, - "loss": 7.0515, + "epoch": 0.41, + "learning_rate": 0.00021479999999999996, + "loss": 6.8399, "step": 363 }, { - "epoch": 0.2, - "learning_rate": 0.00021479999999999996, - "loss": 6.9801, + "epoch": 0.41, + "learning_rate": 0.00021539999999999998, + "loss": 6.8941, "step": 364 }, { - "epoch": 0.2, - "learning_rate": 0.00021539999999999998, - "loss": 7.0104, + "epoch": 0.41, + "learning_rate": 0.00021599999999999996, + "loss": 6.8468, "step": 365 }, { - "epoch": 0.21, - "learning_rate": 0.00021599999999999996, - "loss": 7.1709, + "epoch": 0.41, + "learning_rate": 0.00021659999999999998, + "loss": 6.7709, "step": 366 }, { - "epoch": 0.21, - "learning_rate": 0.00021659999999999998, - "loss": 6.9946, + "epoch": 0.41, + "learning_rate": 0.00021719999999999997, + "loss": 6.8377, "step": 367 }, { - "epoch": 0.21, - "learning_rate": 0.00021719999999999997, - "loss": 7.0377, + "epoch": 0.41, + "learning_rate": 0.00021779999999999998, + "loss": 6.8153, "step": 368 }, { - "epoch": 0.21, - "learning_rate": 0.00021779999999999998, - "loss": 7.1526, + "epoch": 0.41, + "learning_rate": 0.00021839999999999997, + "loss": 6.7531, "step": 369 }, { - "epoch": 0.21, - "learning_rate": 0.00021839999999999997, - "loss": 6.9612, + "epoch": 0.41, + "learning_rate": 0.00021899999999999998, + "loss": 6.9445, "step": 370 }, { - "epoch": 0.21, - "learning_rate": 0.00021899999999999998, - "loss": 6.9493, + "epoch": 0.42, + "learning_rate": 0.00021959999999999997, + "loss": 6.8509, "step": 371 }, { - "epoch": 0.21, - "learning_rate": 0.00021959999999999997, - "loss": 7.1863, + "epoch": 0.42, + "learning_rate": 0.00022019999999999999, + "loss": 7.0474, "step": 372 }, { - "epoch": 0.21, - "learning_rate": 0.00022019999999999999, - "loss": 7.0842, + "epoch": 0.42, + "learning_rate": 0.00022079999999999997, + "loss": 6.8976, "step": 373 }, { - "epoch": 0.21, - "learning_rate": 0.00022079999999999997, - "loss": 6.9515, + "epoch": 0.42, + "learning_rate": 0.0002214, + "loss": 6.956, "step": 374 }, { - "epoch": 0.21, - "learning_rate": 0.0002214, - "loss": 6.8493, + "epoch": 0.42, + "learning_rate": 0.00022199999999999998, + "loss": 6.7783, "step": 375 }, { - "epoch": 0.21, - "learning_rate": 0.00022199999999999998, - "loss": 6.9883, + "epoch": 0.42, + "learning_rate": 0.0002226, + "loss": 6.8574, "step": 376 }, { - "epoch": 0.21, - "learning_rate": 0.0002226, - "loss": 7.0052, + "epoch": 0.42, + "learning_rate": 0.00022319999999999998, + "loss": 6.7116, "step": 377 }, { - "epoch": 0.21, - "learning_rate": 0.00022319999999999998, - "loss": 6.96, + "epoch": 0.42, + "learning_rate": 0.0002238, + "loss": 6.8273, "step": 378 }, { - "epoch": 0.21, - "learning_rate": 0.0002238, - "loss": 7.0645, + "epoch": 0.42, + "learning_rate": 0.00022439999999999998, + "loss": 6.8252, "step": 379 }, { - "epoch": 0.21, - "learning_rate": 0.00022439999999999998, - "loss": 7.1831, + "epoch": 0.43, + "learning_rate": 0.000225, + "loss": 6.9099, "step": 380 }, { - "epoch": 0.21, - "learning_rate": 0.000225, - "loss": 6.897, + "epoch": 0.43, + "learning_rate": 0.00022559999999999998, + "loss": 6.9029, "step": 381 }, { - "epoch": 0.21, - "learning_rate": 0.00022559999999999998, - "loss": 6.9036, + "epoch": 0.43, + "learning_rate": 0.00022619999999999997, + "loss": 6.7356, "step": 382 }, { - "epoch": 0.21, - "learning_rate": 0.00022619999999999997, - "loss": 7.0022, + "epoch": 0.43, + "learning_rate": 0.00022679999999999998, + "loss": 6.935, "step": 383 }, { - "epoch": 0.22, - "learning_rate": 0.00022679999999999998, - "loss": 7.0357, + "epoch": 0.43, + "learning_rate": 0.00022739999999999997, + "loss": 6.9279, "step": 384 }, { - "epoch": 0.22, - "learning_rate": 0.00022739999999999997, - "loss": 7.1649, + "epoch": 0.43, + "learning_rate": 0.00022799999999999999, + "loss": 6.7585, "step": 385 }, { - "epoch": 0.22, - "learning_rate": 0.00022799999999999999, - "loss": 7.3387, + "epoch": 0.43, + "learning_rate": 0.00022859999999999997, + "loss": 6.8771, "step": 386 }, { - "epoch": 0.22, - "learning_rate": 0.00022859999999999997, - "loss": 7.0349, + "epoch": 0.43, + "learning_rate": 0.0002292, + "loss": 6.8192, "step": 387 }, { - "epoch": 0.22, - "learning_rate": 0.0002292, - "loss": 6.8661, + "epoch": 0.43, + "learning_rate": 0.00022979999999999997, + "loss": 6.8009, "step": 388 }, { - "epoch": 0.22, - "learning_rate": 0.00022979999999999997, - "loss": 6.9529, + "epoch": 0.44, + "learning_rate": 0.0002304, + "loss": 6.8385, "step": 389 }, { - "epoch": 0.22, - "learning_rate": 0.0002304, - "loss": 7.116, + "epoch": 0.44, + "learning_rate": 0.00023099999999999998, + "loss": 7.0575, "step": 390 }, { - "epoch": 0.22, - "learning_rate": 0.00023099999999999998, - "loss": 6.9367, + "epoch": 0.44, + "learning_rate": 0.0002316, + "loss": 6.8882, "step": 391 }, { - "epoch": 0.22, - "learning_rate": 0.0002316, - "loss": 7.5156, + "epoch": 0.44, + "learning_rate": 0.00023219999999999998, + "loss": 6.8121, "step": 392 }, { - "epoch": 0.22, - "learning_rate": 0.00023219999999999998, - "loss": 7.1011, + "epoch": 0.44, + "learning_rate": 0.0002328, + "loss": 6.8311, "step": 393 }, { - "epoch": 0.22, - "learning_rate": 0.0002328, - "loss": 6.9139, + "epoch": 0.44, + "learning_rate": 0.00023339999999999998, + "loss": 7.0132, "step": 394 }, { - "epoch": 0.22, - "learning_rate": 0.00023339999999999998, - "loss": 6.9251, + "epoch": 0.44, + "learning_rate": 0.000234, + "loss": 6.7881, "step": 395 }, { - "epoch": 0.22, - "learning_rate": 0.000234, - "loss": 6.8266, + "epoch": 0.44, + "learning_rate": 0.00023459999999999998, + "loss": 6.6595, "step": 396 }, { - "epoch": 0.22, - "learning_rate": 0.00023459999999999998, - "loss": 6.7855, + "epoch": 0.45, + "learning_rate": 0.0002352, + "loss": 6.5624, "step": 397 }, { - "epoch": 0.22, - "learning_rate": 0.0002352, - "loss": 6.9616, + "epoch": 0.45, + "learning_rate": 0.00023579999999999999, + "loss": 6.6603, "step": 398 }, { - "epoch": 0.22, - "learning_rate": 0.00023579999999999999, - "loss": 7.008, + "epoch": 0.45, + "learning_rate": 0.0002364, + "loss": 6.6189, "step": 399 }, { - "epoch": 0.22, - "learning_rate": 0.0002364, - "loss": 6.6132, + "epoch": 0.45, + "learning_rate": 0.000237, + "loss": 6.4334, "step": 400 }, { - "epoch": 0.22, - "learning_rate": 0.000237, - "loss": 7.2819, + "epoch": 0.45, + "learning_rate": 0.0002376, + "loss": 7.0859, "step": 401 }, { - "epoch": 0.23, - "learning_rate": 0.0002376, - "loss": 7.0191, + "epoch": 0.45, + "learning_rate": 0.0002382, + "loss": 7.0288, "step": 402 }, { - "epoch": 0.23, - "learning_rate": 0.0002382, - "loss": 7.0059, + "epoch": 0.45, + "learning_rate": 0.0002388, + "loss": 6.998, "step": 403 }, { - "epoch": 0.23, - "learning_rate": 0.0002388, - "loss": 7.1836, + "epoch": 0.45, + "learning_rate": 0.0002394, + "loss": 6.8879, "step": 404 }, { - "epoch": 0.23, - "learning_rate": 0.0002394, - "loss": 6.9707, + "epoch": 0.45, + "learning_rate": 0.00023999999999999998, + "loss": 7.0649, "step": 405 }, { - "epoch": 0.23, - "learning_rate": 0.00023999999999999998, - "loss": 7.0907, + "epoch": 0.46, + "learning_rate": 0.0002406, + "loss": 6.8729, "step": 406 }, { - "epoch": 0.23, - "learning_rate": 0.0002406, - "loss": 6.9978, + "epoch": 0.46, + "learning_rate": 0.00024119999999999998, + "loss": 6.9037, "step": 407 }, { - "epoch": 0.23, - "learning_rate": 0.00024119999999999998, - "loss": 7.0639, + "epoch": 0.46, + "learning_rate": 0.0002418, + "loss": 6.8431, "step": 408 }, { - "epoch": 0.23, - "learning_rate": 0.0002418, - "loss": 7.0238, + "epoch": 0.46, + "learning_rate": 0.00024239999999999998, + "loss": 7.0446, "step": 409 }, { - "epoch": 0.23, - "learning_rate": 0.00024239999999999998, - "loss": 6.9276, + "epoch": 0.46, + "learning_rate": 0.000243, + "loss": 6.9666, "step": 410 }, { - "epoch": 0.23, - "learning_rate": 0.000243, - "loss": 6.9775, + "epoch": 0.46, + "learning_rate": 0.00024359999999999999, + "loss": 6.8929, "step": 411 }, { - "epoch": 0.23, - "learning_rate": 0.00024359999999999999, - "loss": 6.9224, + "epoch": 0.46, + "learning_rate": 0.00024419999999999997, + "loss": 6.8846, "step": 412 }, { - "epoch": 0.23, - "learning_rate": 0.00024419999999999997, - "loss": 7.1259, + "epoch": 0.46, + "learning_rate": 0.0002448, + "loss": 6.8704, "step": 413 }, { - "epoch": 0.23, - "learning_rate": 0.0002448, - "loss": 6.9277, + "epoch": 0.46, + "learning_rate": 0.00024539999999999995, + "loss": 6.7452, "step": 414 }, { - "epoch": 0.23, - "learning_rate": 0.00024539999999999995, - "loss": 7.0661, + "epoch": 0.47, + "learning_rate": 0.00024599999999999996, + "loss": 6.7961, "step": 415 }, { - "epoch": 0.23, - "learning_rate": 0.00024599999999999996, - "loss": 7.079, + "epoch": 0.47, + "learning_rate": 0.0002466, + "loss": 6.9163, "step": 416 }, { - "epoch": 0.23, - "learning_rate": 0.0002466, - "loss": 6.9792, + "epoch": 0.47, + "learning_rate": 0.0002472, + "loss": 6.8202, "step": 417 }, { - "epoch": 0.23, - "learning_rate": 0.0002472, - "loss": 6.9939, + "epoch": 0.47, + "learning_rate": 0.00024779999999999995, + "loss": 6.8721, "step": 418 }, { - "epoch": 0.23, - "learning_rate": 0.00024779999999999995, - "loss": 6.9837, + "epoch": 0.47, + "learning_rate": 0.00024839999999999997, + "loss": 6.8826, "step": 419 }, { - "epoch": 0.24, - "learning_rate": 0.00024839999999999997, - "loss": 7.2184, + "epoch": 0.47, + "learning_rate": 0.000249, + "loss": 6.884, "step": 420 }, { - "epoch": 0.24, - "learning_rate": 0.000249, - "loss": 7.1127, + "epoch": 0.47, + "learning_rate": 0.00024959999999999994, + "loss": 6.8824, "step": 421 }, { - "epoch": 0.24, - "learning_rate": 0.00024959999999999994, - "loss": 7.0602, + "epoch": 0.47, + "learning_rate": 0.00025019999999999996, + "loss": 6.9743, "step": 422 }, { - "epoch": 0.24, - "learning_rate": 0.00025019999999999996, - "loss": 6.9942, + "epoch": 0.47, + "learning_rate": 0.00025079999999999997, + "loss": 6.8894, "step": 423 }, { - "epoch": 0.24, - "learning_rate": 0.00025079999999999997, - "loss": 7.1876, + "epoch": 0.48, + "learning_rate": 0.0002514, + "loss": 6.8829, "step": 424 }, { - "epoch": 0.24, - "learning_rate": 0.0002514, - "loss": 7.1962, + "epoch": 0.48, + "learning_rate": 0.00025199999999999995, + "loss": 6.9305, "step": 425 }, { - "epoch": 0.24, - "learning_rate": 0.00025199999999999995, - "loss": 7.1023, + "epoch": 0.48, + "learning_rate": 0.00025259999999999996, + "loss": 7.0412, "step": 426 }, { - "epoch": 0.24, - "learning_rate": 0.00025259999999999996, - "loss": 6.9652, + "epoch": 0.48, + "learning_rate": 0.0002532, + "loss": 6.788, "step": 427 }, { - "epoch": 0.24, - "learning_rate": 0.0002532, - "loss": 7.0197, + "epoch": 0.48, + "learning_rate": 0.0002538, + "loss": 6.9715, "step": 428 }, { - "epoch": 0.24, - "learning_rate": 0.0002538, - "loss": 7.02, + "epoch": 0.48, + "learning_rate": 0.00025439999999999995, + "loss": 6.8377, "step": 429 }, { - "epoch": 0.24, - "learning_rate": 0.00025439999999999995, - "loss": 7.0163, + "epoch": 0.48, + "learning_rate": 0.00025499999999999996, + "loss": 6.8427, "step": 430 }, { - "epoch": 0.24, - "learning_rate": 0.00025499999999999996, - "loss": 6.9658, + "epoch": 0.48, + "learning_rate": 0.0002556, + "loss": 6.996, "step": 431 }, { - "epoch": 0.24, - "learning_rate": 0.0002556, - "loss": 7.0774, + "epoch": 0.48, + "learning_rate": 0.0002562, + "loss": 6.9316, "step": 432 }, { - "epoch": 0.24, - "learning_rate": 0.0002562, - "loss": 7.0779, + "epoch": 0.49, + "learning_rate": 0.00025679999999999995, + "loss": 6.5887, "step": 433 }, { - "epoch": 0.24, - "learning_rate": 0.00025679999999999995, - "loss": 7.1834, + "epoch": 0.49, + "learning_rate": 0.00025739999999999997, + "loss": 6.8635, "step": 434 }, { - "epoch": 0.24, - "learning_rate": 0.00025739999999999997, - "loss": 7.0392, + "epoch": 0.49, + "learning_rate": 0.000258, + "loss": 6.767, "step": 435 }, { - "epoch": 0.24, - "learning_rate": 0.000258, - "loss": 6.9683, + "epoch": 0.49, + "learning_rate": 0.0002586, + "loss": 6.9778, "step": 436 }, { - "epoch": 0.24, - "learning_rate": 0.0002586, - "loss": 6.9718, + "epoch": 0.49, + "learning_rate": 0.00025919999999999996, + "loss": 6.8633, "step": 437 }, { - "epoch": 0.25, - "learning_rate": 0.00025919999999999996, - "loss": 7.0578, + "epoch": 0.49, + "learning_rate": 0.00025979999999999997, + "loss": 6.9399, "step": 438 }, { - "epoch": 0.25, - "learning_rate": 0.00025979999999999997, - "loss": 6.9207, + "epoch": 0.49, + "learning_rate": 0.0002604, + "loss": 6.7401, "step": 439 }, { - "epoch": 0.25, - "learning_rate": 0.0002604, - "loss": 7.0706, + "epoch": 0.49, + "learning_rate": 0.000261, + "loss": 6.8552, "step": 440 }, { - "epoch": 0.25, - "learning_rate": 0.000261, - "loss": 6.9728, + "epoch": 0.49, + "learning_rate": 0.00026159999999999996, + "loss": 6.9278, "step": 441 }, { - "epoch": 0.25, - "learning_rate": 0.00026159999999999996, - "loss": 7.0021, + "epoch": 0.5, + "learning_rate": 0.0002622, + "loss": 6.8031, "step": 442 }, { - "epoch": 0.25, - "learning_rate": 0.0002622, - "loss": 6.8181, + "epoch": 0.5, + "learning_rate": 0.0002628, + "loss": 6.7692, "step": 443 }, { - "epoch": 0.25, - "learning_rate": 0.0002628, - "loss": 6.8681, + "epoch": 0.5, + "learning_rate": 0.00026339999999999995, + "loss": 6.9128, "step": 444 }, { - "epoch": 0.25, - "learning_rate": 0.00026339999999999995, - "loss": 6.8901, + "epoch": 0.5, + "learning_rate": 0.00026399999999999997, + "loss": 6.6816, "step": 445 }, { - "epoch": 0.25, - "learning_rate": 0.00026399999999999997, - "loss": 7.2108, + "epoch": 0.5, + "learning_rate": 0.0002646, + "loss": 6.6995, "step": 446 }, { - "epoch": 0.25, - "learning_rate": 0.0002646, - "loss": 6.9756, + "epoch": 0.5, + "learning_rate": 0.0002652, + "loss": 6.6771, "step": 447 }, { - "epoch": 0.25, - "learning_rate": 0.0002652, - "loss": 6.9328, + "epoch": 0.5, + "learning_rate": 0.00026579999999999996, + "loss": 6.6629, "step": 448 }, { - "epoch": 0.25, - "learning_rate": 0.00026579999999999996, - "loss": 7.1892, + "epoch": 0.5, + "learning_rate": 0.00026639999999999997, + "loss": 6.3252, "step": 449 }, { - "epoch": 0.25, - "learning_rate": 0.00026639999999999997, - "loss": 6.4576, + "epoch": 0.5, + "learning_rate": 0.000267, + "loss": 6.3625, "step": 450 }, { - "epoch": 0.25, - "learning_rate": 0.000267, - "loss": 7.0339, + "epoch": 0.51, + "learning_rate": 0.0002676, + "loss": 6.9431, "step": 451 }, { - "epoch": 0.25, - "learning_rate": 0.0002676, - "loss": 7.1057, + "epoch": 0.51, + "learning_rate": 0.00026819999999999996, + "loss": 7.0696, "step": 452 }, { - "epoch": 0.25, - "learning_rate": 0.00026819999999999996, - "loss": 7.1649, + "epoch": 0.51, + "learning_rate": 0.0002688, + "loss": 7.0835, "step": 453 }, { - "epoch": 0.25, - "learning_rate": 0.0002688, - "loss": 7.0715, + "epoch": 0.51, + "learning_rate": 0.0002694, + "loss": 6.88, "step": 454 }, { - "epoch": 0.26, - "learning_rate": 0.0002694, - "loss": 6.9592, + "epoch": 0.51, + "learning_rate": 0.00027, + "loss": 6.98, "step": 455 }, { - "epoch": 0.26, - "learning_rate": 0.00027, - "loss": 7.1295, + "epoch": 0.51, + "learning_rate": 0.00027059999999999996, + "loss": 7.04, "step": 456 }, { - "epoch": 0.26, - "learning_rate": 0.00027059999999999996, - "loss": 6.9516, + "epoch": 0.51, + "learning_rate": 0.0002712, + "loss": 6.8282, "step": 457 }, { - "epoch": 0.26, - "learning_rate": 0.0002712, - "loss": 7.0106, + "epoch": 0.51, + "learning_rate": 0.0002718, + "loss": 6.9422, "step": 458 }, { - "epoch": 0.26, - "learning_rate": 0.0002718, - "loss": 6.8993, + "epoch": 0.51, + "learning_rate": 0.0002724, + "loss": 6.9104, "step": 459 }, { - "epoch": 0.26, - "learning_rate": 0.0002724, - "loss": 7.1038, + "epoch": 0.52, + "learning_rate": 0.00027299999999999997, + "loss": 7.0045, "step": 460 }, { - "epoch": 0.26, - "learning_rate": 0.00027299999999999997, - "loss": 6.8717, + "epoch": 0.52, + "learning_rate": 0.0002736, + "loss": 6.8041, "step": 461 }, { - "epoch": 0.26, - "learning_rate": 0.0002736, - "loss": 6.8407, + "epoch": 0.52, + "learning_rate": 0.0002742, + "loss": 6.9422, "step": 462 }, { - "epoch": 0.26, - "learning_rate": 0.0002742, - "loss": 7.0185, + "epoch": 0.52, + "learning_rate": 0.0002748, + "loss": 6.8194, "step": 463 }, { - "epoch": 0.26, - "learning_rate": 0.0002748, - "loss": 7.112, + "epoch": 0.52, + "learning_rate": 0.00027539999999999997, + "loss": 6.7778, "step": 464 }, { - "epoch": 0.26, - "learning_rate": 0.00027539999999999997, - "loss": 6.8938, + "epoch": 0.52, + "learning_rate": 0.000276, + "loss": 6.9162, "step": 465 }, { - "epoch": 0.26, - "learning_rate": 0.000276, - "loss": 7.1548, + "epoch": 0.52, + "learning_rate": 0.0002766, + "loss": 6.8971, "step": 466 }, { - "epoch": 0.26, - "learning_rate": 0.0002766, - "loss": 6.984, + "epoch": 0.52, + "learning_rate": 0.0002772, + "loss": 6.8153, "step": 467 }, { - "epoch": 0.26, - "learning_rate": 0.0002772, - "loss": 7.145, + "epoch": 0.52, + "learning_rate": 0.0002778, + "loss": 6.9018, "step": 468 }, { - "epoch": 0.26, - "learning_rate": 0.0002778, - "loss": 6.8399, + "epoch": 0.53, + "learning_rate": 0.0002784, + "loss": 6.8423, "step": 469 }, { - "epoch": 0.26, - "learning_rate": 0.0002784, - "loss": 7.0988, + "epoch": 0.53, + "learning_rate": 0.000279, + "loss": 6.9256, "step": 470 }, { - "epoch": 0.26, - "learning_rate": 0.000279, - "loss": 7.0583, + "epoch": 0.53, + "learning_rate": 0.00027959999999999997, + "loss": 6.8025, "step": 471 }, { - "epoch": 0.26, - "learning_rate": 0.00027959999999999997, - "loss": 6.8744, + "epoch": 0.53, + "learning_rate": 0.0002802, + "loss": 6.7987, "step": 472 }, { - "epoch": 0.27, - "learning_rate": 0.0002802, - "loss": 6.895, + "epoch": 0.53, + "learning_rate": 0.0002808, + "loss": 6.8809, "step": 473 }, { - "epoch": 0.27, - "learning_rate": 0.0002808, - "loss": 6.9742, + "epoch": 0.53, + "learning_rate": 0.00028139999999999996, + "loss": 6.7699, "step": 474 }, { - "epoch": 0.27, - "learning_rate": 0.00028139999999999996, - "loss": 6.9724, + "epoch": 0.53, + "learning_rate": 0.00028199999999999997, + "loss": 6.9145, "step": 475 }, { - "epoch": 0.27, - "learning_rate": 0.00028199999999999997, - "loss": 7.0225, + "epoch": 0.53, + "learning_rate": 0.0002826, + "loss": 6.9334, "step": 476 }, { - "epoch": 0.27, - "learning_rate": 0.0002826, - "loss": 7.0496, + "epoch": 0.53, + "learning_rate": 0.00028319999999999994, + "loss": 7.0333, "step": 477 }, { - "epoch": 0.27, - "learning_rate": 0.00028319999999999994, - "loss": 6.9553, + "epoch": 0.54, + "learning_rate": 0.00028379999999999996, + "loss": 6.8036, "step": 478 }, { - "epoch": 0.27, - "learning_rate": 0.00028379999999999996, - "loss": 7.0546, + "epoch": 0.54, + "learning_rate": 0.0002844, + "loss": 6.7853, "step": 479 }, { - "epoch": 0.27, - "learning_rate": 0.0002844, - "loss": 7.1255, + "epoch": 0.54, + "learning_rate": 0.000285, + "loss": 6.8205, "step": 480 }, { - "epoch": 0.27, - "learning_rate": 0.000285, - "loss": 7.1964, - "step": 481 + "epoch": 0.54, + "learning_rate": 0.00028559999999999995, + "loss": 6.7719, + "step": 481 }, { - "epoch": 0.27, - "learning_rate": 0.00028559999999999995, - "loss": 6.9996, + "epoch": 0.54, + "learning_rate": 0.00028619999999999996, + "loss": 6.8131, "step": 482 }, { - "epoch": 0.27, - "learning_rate": 0.00028619999999999996, - "loss": 7.3929, + "epoch": 0.54, + "learning_rate": 0.0002868, + "loss": 6.827, "step": 483 }, { - "epoch": 0.27, - "learning_rate": 0.0002868, - "loss": 6.9579, + "epoch": 0.54, + "learning_rate": 0.00028739999999999994, + "loss": 6.8999, "step": 484 }, { - "epoch": 0.27, - "learning_rate": 0.00028739999999999994, - "loss": 7.1277, + "epoch": 0.54, + "learning_rate": 0.00028799999999999995, + "loss": 6.8928, "step": 485 }, { - "epoch": 0.27, - "learning_rate": 0.00028799999999999995, - "loss": 7.0739, + "epoch": 0.54, + "learning_rate": 0.00028859999999999997, + "loss": 6.8204, "step": 486 }, { - "epoch": 0.27, - "learning_rate": 0.00028859999999999997, - "loss": 6.905, + "epoch": 0.55, + "learning_rate": 0.0002892, + "loss": 6.8927, "step": 487 }, { - "epoch": 0.27, - "learning_rate": 0.0002892, - "loss": 7.2388, + "epoch": 0.55, + "learning_rate": 0.00028979999999999994, + "loss": 6.7876, "step": 488 }, { - "epoch": 0.27, - "learning_rate": 0.00028979999999999994, - "loss": 7.0177, + "epoch": 0.55, + "learning_rate": 0.00029039999999999996, + "loss": 7.011, "step": 489 }, { - "epoch": 0.27, - "learning_rate": 0.00029039999999999996, - "loss": 6.9566, + "epoch": 0.55, + "learning_rate": 0.00029099999999999997, + "loss": 6.9904, "step": 490 }, { - "epoch": 0.28, - "learning_rate": 0.00029099999999999997, - "loss": 6.9018, + "epoch": 0.55, + "learning_rate": 0.0002916, + "loss": 6.7925, "step": 491 }, { - "epoch": 0.28, - "learning_rate": 0.0002916, - "loss": 7.1509, + "epoch": 0.55, + "learning_rate": 0.00029219999999999995, + "loss": 6.8486, "step": 492 }, { - "epoch": 0.28, - "learning_rate": 0.00029219999999999995, - "loss": 7.0388, + "epoch": 0.55, + "learning_rate": 0.00029279999999999996, + "loss": 6.8364, "step": 493 }, { - "epoch": 0.28, - "learning_rate": 0.00029279999999999996, - "loss": 7.15, + "epoch": 0.55, + "learning_rate": 0.0002934, + "loss": 6.7825, "step": 494 }, { - "epoch": 0.28, - "learning_rate": 0.0002934, - "loss": 6.9175, + "epoch": 0.55, + "learning_rate": 0.000294, + "loss": 6.9096, "step": 495 }, { - "epoch": 0.28, - "learning_rate": 0.000294, - "loss": 7.025, + "epoch": 0.56, + "learning_rate": 0.00029459999999999995, + "loss": 6.6241, "step": 496 }, { - "epoch": 0.28, - "learning_rate": 0.00029459999999999995, - "loss": 6.7853, + "epoch": 0.56, + "learning_rate": 0.00029519999999999997, + "loss": 6.3106, "step": 497 }, { - "epoch": 0.28, - "learning_rate": 0.00029519999999999997, - "loss": 7.1744, + "epoch": 0.56, + "learning_rate": 0.0002958, + "loss": 7.0315, "step": 498 }, { - "epoch": 0.28, - "learning_rate": 0.0002958, - "loss": 6.7261, + "epoch": 0.56, + "learning_rate": 0.0002964, + "loss": 6.7356, "step": 499 }, { - "epoch": 0.28, - "learning_rate": 0.0002964, - "loss": 6.8938, + "epoch": 0.56, + "learning_rate": 0.00029699999999999996, + "loss": 6.2901, "step": 500 }, { - "epoch": 0.28, - "eval_loss": 7.0276780128479, - "eval_runtime": 341.1124, - "eval_samples_per_second": 7.745, - "eval_steps_per_second": 0.97, + "epoch": 0.56, + "eval_loss": 6.832920551300049, + "eval_runtime": 362.0157, + "eval_samples_per_second": 7.298, + "eval_steps_per_second": 0.914, "eval_wer": 1.0, "step": 500 }, { - "epoch": 0.28, - "learning_rate": 0.00029699999999999996, - "loss": 7.0966, + "epoch": 0.56, + "learning_rate": 0.00029759999999999997, + "loss": 7.2547, "step": 501 }, { - "epoch": 0.28, - "learning_rate": 0.00029759999999999997, - "loss": 7.1217, + "epoch": 0.56, + "learning_rate": 0.0002982, + "loss": 6.8181, "step": 502 }, { - "epoch": 0.28, - "learning_rate": 0.0002982, - "loss": 7.023, + "epoch": 0.56, + "learning_rate": 0.0002988, + "loss": 6.9446, "step": 503 }, { - "epoch": 0.28, - "learning_rate": 0.0002988, - "loss": 7.1377, + "epoch": 0.57, + "learning_rate": 0.00029939999999999996, + "loss": 7.0057, "step": 504 }, { - "epoch": 0.28, - "learning_rate": 0.00029939999999999996, - "loss": 6.9172, + "epoch": 0.57, + "learning_rate": 0.0003, + "loss": 6.976, "step": 505 }, { - "epoch": 0.28, - "learning_rate": 0.0003, - "loss": 7.1241, + "epoch": 0.57, + "learning_rate": 0.000299234693877551, + "loss": 6.9817, "step": 506 }, { - "epoch": 0.28, - "learning_rate": 0.0002997663551401869, - "loss": 6.8878, + "epoch": 0.57, + "learning_rate": 0.00029846938775510205, + "loss": 6.9377, "step": 507 }, { - "epoch": 0.28, - "learning_rate": 0.00029953271028037383, - "loss": 7.1644, + "epoch": 0.57, + "learning_rate": 0.00029770408163265304, + "loss": 6.8353, "step": 508 }, { - "epoch": 0.29, - "learning_rate": 0.00029929906542056073, - "loss": 7.2148, + "epoch": 0.57, + "learning_rate": 0.0002969387755102041, + "loss": 6.8738, "step": 509 }, { - "epoch": 0.29, - "learning_rate": 0.00029906542056074763, - "loss": 7.0696, + "epoch": 0.57, + "learning_rate": 0.00029617346938775506, + "loss": 6.9288, "step": 510 }, { - "epoch": 0.29, - "learning_rate": 0.0002988317757009346, - "loss": 7.0247, + "epoch": 0.57, + "learning_rate": 0.0002954081632653061, + "loss": 6.8304, "step": 511 }, { - "epoch": 0.29, - "learning_rate": 0.0002985981308411215, - "loss": 6.9583, + "epoch": 0.57, + "learning_rate": 0.0002946428571428571, + "loss": 6.9171, "step": 512 }, { - "epoch": 0.29, - "learning_rate": 0.0002983644859813084, - "loss": 6.9546, + "epoch": 0.58, + "learning_rate": 0.0002938775510204081, + "loss": 7.022, "step": 513 }, { - "epoch": 0.29, - "learning_rate": 0.0002981308411214953, - "loss": 6.8928, + "epoch": 0.58, + "learning_rate": 0.00029311224489795917, + "loss": 6.9531, "step": 514 }, { - "epoch": 0.29, - "learning_rate": 0.00029789719626168224, - "loss": 7.094, + "epoch": 0.58, + "learning_rate": 0.0002923469387755102, + "loss": 6.8854, "step": 515 }, { - "epoch": 0.29, - "learning_rate": 0.00029766355140186914, - "loss": 7.089, + "epoch": 0.58, + "learning_rate": 0.0002915816326530612, + "loss": 6.7042, "step": 516 }, { - "epoch": 0.29, - "learning_rate": 0.00029742990654205604, - "loss": 7.0865, + "epoch": 0.58, + "learning_rate": 0.00029081632653061223, + "loss": 6.9831, "step": 517 }, { - "epoch": 0.29, - "learning_rate": 0.00029719626168224294, - "loss": 7.0651, + "epoch": 0.58, + "learning_rate": 0.0002900510204081632, + "loss": 6.7991, "step": 518 }, { - "epoch": 0.29, - "learning_rate": 0.0002969626168224299, - "loss": 7.0141, + "epoch": 0.58, + "learning_rate": 0.00028928571428571425, + "loss": 6.9969, "step": 519 }, { - "epoch": 0.29, - "learning_rate": 0.0002967289719626168, - "loss": 6.9969, + "epoch": 0.58, + "learning_rate": 0.0002885204081632653, + "loss": 6.9657, "step": 520 }, { - "epoch": 0.29, - "learning_rate": 0.0002964953271028037, - "loss": 6.9849, + "epoch": 0.58, + "learning_rate": 0.0002877551020408163, + "loss": 6.7878, "step": 521 }, { - "epoch": 0.29, - "learning_rate": 0.00029626168224299065, - "loss": 7.0939, + "epoch": 0.59, + "learning_rate": 0.0002869897959183673, + "loss": 6.8026, "step": 522 }, { - "epoch": 0.29, - "learning_rate": 0.00029602803738317755, - "loss": 6.8687, + "epoch": 0.59, + "learning_rate": 0.00028622448979591836, + "loss": 6.7209, "step": 523 }, { - "epoch": 0.29, - "learning_rate": 0.0002957943925233645, - "loss": 7.0734, + "epoch": 0.59, + "learning_rate": 0.0002854591836734694, + "loss": 6.8029, "step": 524 }, { - "epoch": 0.29, - "learning_rate": 0.00029556074766355135, - "loss": 7.0944, + "epoch": 0.59, + "learning_rate": 0.0002846938775510204, + "loss": 6.9162, "step": 525 }, { - "epoch": 0.29, - "learning_rate": 0.0002953271028037383, - "loss": 6.9508, + "epoch": 0.59, + "learning_rate": 0.00028392857142857137, + "loss": 6.7592, "step": 526 }, { - "epoch": 0.3, - "learning_rate": 0.0002950934579439252, - "loss": 7.2332, + "epoch": 0.59, + "learning_rate": 0.0002831632653061224, + "loss": 6.8772, "step": 527 }, { - "epoch": 0.3, - "learning_rate": 0.00029485981308411215, - "loss": 6.8933, + "epoch": 0.59, + "learning_rate": 0.00028239795918367345, + "loss": 6.8832, "step": 528 }, { - "epoch": 0.3, - "learning_rate": 0.00029462616822429905, - "loss": 7.0654, + "epoch": 0.59, + "learning_rate": 0.0002816326530612245, + "loss": 6.9216, "step": 529 }, { - "epoch": 0.3, - "learning_rate": 0.00029439252336448596, - "loss": 7.1729, + "epoch": 0.59, + "learning_rate": 0.00028086734693877547, + "loss": 6.9836, "step": 530 }, { - "epoch": 0.3, - "learning_rate": 0.00029415887850467286, - "loss": 6.8106, + "epoch": 0.6, + "learning_rate": 0.0002801020408163265, + "loss": 6.8133, "step": 531 }, { - "epoch": 0.3, - "learning_rate": 0.0002939252336448598, - "loss": 6.8655, + "epoch": 0.6, + "learning_rate": 0.00027933673469387755, + "loss": 6.781, "step": 532 }, { - "epoch": 0.3, - "learning_rate": 0.0002936915887850467, - "loss": 7.0379, + "epoch": 0.6, + "learning_rate": 0.00027857142857142854, + "loss": 6.7889, "step": 533 }, { - "epoch": 0.3, - "learning_rate": 0.0002934579439252336, - "loss": 6.8988, + "epoch": 0.6, + "learning_rate": 0.0002778061224489796, + "loss": 6.7747, "step": 534 }, { - "epoch": 0.3, - "learning_rate": 0.00029322429906542056, - "loss": 7.2063, + "epoch": 0.6, + "learning_rate": 0.00027704081632653056, + "loss": 6.8035, "step": 535 }, { - "epoch": 0.3, - "learning_rate": 0.00029299065420560746, - "loss": 7.0098, + "epoch": 0.6, + "learning_rate": 0.0002762755102040816, + "loss": 6.9291, "step": 536 }, { - "epoch": 0.3, - "learning_rate": 0.00029275700934579436, - "loss": 6.8802, + "epoch": 0.6, + "learning_rate": 0.00027551020408163264, + "loss": 6.8163, "step": 537 }, { - "epoch": 0.3, - "learning_rate": 0.00029252336448598126, - "loss": 6.9024, + "epoch": 0.6, + "learning_rate": 0.0002747448979591837, + "loss": 6.9158, "step": 538 }, { - "epoch": 0.3, - "learning_rate": 0.0002922897196261682, - "loss": 7.1021, + "epoch": 0.6, + "learning_rate": 0.00027397959183673466, + "loss": 6.9038, "step": 539 }, { - "epoch": 0.3, - "learning_rate": 0.0002920560747663551, - "loss": 7.2898, + "epoch": 0.61, + "learning_rate": 0.0002732142857142857, + "loss": 6.8366, "step": 540 }, { - "epoch": 0.3, - "learning_rate": 0.000291822429906542, - "loss": 6.8831, + "epoch": 0.61, + "learning_rate": 0.0002724489795918367, + "loss": 6.9376, "step": 541 }, { - "epoch": 0.3, - "learning_rate": 0.0002915887850467289, - "loss": 7.1892, + "epoch": 0.61, + "learning_rate": 0.00027168367346938773, + "loss": 6.8482, "step": 542 }, { - "epoch": 0.3, - "learning_rate": 0.00029135514018691587, - "loss": 6.9105, + "epoch": 0.61, + "learning_rate": 0.0002709183673469387, + "loss": 6.7671, "step": 543 }, { - "epoch": 0.3, - "learning_rate": 0.0002911214953271028, - "loss": 6.9292, + "epoch": 0.61, + "learning_rate": 0.00027015306122448975, + "loss": 6.6458, "step": 544 }, { - "epoch": 0.31, - "learning_rate": 0.0002908878504672897, - "loss": 6.8223, + "epoch": 0.61, + "learning_rate": 0.0002693877551020408, + "loss": 6.7696, "step": 545 }, { - "epoch": 0.31, - "learning_rate": 0.00029065420560747663, - "loss": 6.6742, + "epoch": 0.61, + "learning_rate": 0.00026862244897959183, + "loss": 6.6372, "step": 546 }, { - "epoch": 0.31, - "learning_rate": 0.00029042056074766353, - "loss": 7.1423, + "epoch": 0.61, + "learning_rate": 0.00026785714285714287, + "loss": 6.9219, "step": 547 }, { - "epoch": 0.31, - "learning_rate": 0.0002901869158878505, - "loss": 6.7272, + "epoch": 0.61, + "learning_rate": 0.00026709183673469386, + "loss": 6.6973, "step": 548 }, { - "epoch": 0.31, - "learning_rate": 0.00028995327102803733, - "loss": 6.6769, + "epoch": 0.62, + "learning_rate": 0.0002663265306122449, + "loss": 6.3204, "step": 549 }, { - "epoch": 0.31, - "learning_rate": 0.0002897196261682243, - "loss": 6.421, + "epoch": 0.62, + "learning_rate": 0.0002655612244897959, + "loss": 6.5614, "step": 550 }, { - "epoch": 0.31, - "learning_rate": 0.0002894859813084112, - "loss": 7.3565, + "epoch": 0.62, + "learning_rate": 0.0002647959183673469, + "loss": 6.9267, "step": 551 }, { - "epoch": 0.31, - "learning_rate": 0.00028925233644859814, - "loss": 7.4568, + "epoch": 0.62, + "learning_rate": 0.0002640306122448979, + "loss": 7.09, "step": 552 }, { - "epoch": 0.31, - "learning_rate": 0.00028901869158878504, - "loss": 7.0468, + "epoch": 0.62, + "learning_rate": 0.00026326530612244894, + "loss": 6.9054, "step": 553 }, { - "epoch": 0.31, - "learning_rate": 0.00028878504672897194, - "loss": 7.0041, + "epoch": 0.62, + "learning_rate": 0.0002625, + "loss": 6.8814, "step": 554 }, { - "epoch": 0.31, - "learning_rate": 0.00028855140186915884, - "loss": 7.0425, + "epoch": 0.62, + "learning_rate": 0.000261734693877551, + "loss": 6.8698, "step": 555 }, { - "epoch": 0.31, - "learning_rate": 0.0002883177570093458, - "loss": 7.1408, + "epoch": 0.62, + "learning_rate": 0.000260969387755102, + "loss": 6.8921, "step": 556 }, { - "epoch": 0.31, - "learning_rate": 0.0002880841121495327, - "loss": 7.1654, + "epoch": 0.62, + "learning_rate": 0.00026020408163265305, + "loss": 6.8406, "step": 557 }, { - "epoch": 0.31, - "learning_rate": 0.0002878504672897196, - "loss": 7.1483, + "epoch": 0.63, + "learning_rate": 0.00025943877551020403, + "loss": 6.9311, "step": 558 }, { - "epoch": 0.31, - "learning_rate": 0.00028761682242990655, - "loss": 6.8182, + "epoch": 0.63, + "learning_rate": 0.00025867346938775507, + "loss": 6.8586, "step": 559 }, { - "epoch": 0.31, - "learning_rate": 0.00028738317757009345, - "loss": 7.0514, + "epoch": 0.63, + "learning_rate": 0.0002579081632653061, + "loss": 6.7891, "step": 560 }, { - "epoch": 0.31, - "learning_rate": 0.00028714953271028035, - "loss": 7.0641, + "epoch": 0.63, + "learning_rate": 0.0002571428571428571, + "loss": 6.8937, "step": 561 }, { - "epoch": 0.32, - "learning_rate": 0.00028691588785046725, - "loss": 6.9528, + "epoch": 0.63, + "learning_rate": 0.00025637755102040814, + "loss": 6.7656, "step": 562 }, { - "epoch": 0.32, - "learning_rate": 0.0002866822429906542, - "loss": 7.0379, + "epoch": 0.63, + "learning_rate": 0.0002556122448979592, + "loss": 6.8204, "step": 563 }, { - "epoch": 0.32, - "learning_rate": 0.0002864485981308411, - "loss": 7.3118, + "epoch": 0.63, + "learning_rate": 0.0002548469387755102, + "loss": 6.8283, "step": 564 }, { - "epoch": 0.32, - "learning_rate": 0.000286214953271028, - "loss": 7.0804, + "epoch": 0.63, + "learning_rate": 0.0002540816326530612, + "loss": 7.0003, "step": 565 }, { - "epoch": 0.32, - "learning_rate": 0.0002859813084112149, - "loss": 6.946, + "epoch": 0.63, + "learning_rate": 0.0002533163265306122, + "loss": 6.6763, "step": 566 }, { - "epoch": 0.32, - "learning_rate": 0.00028574766355140185, - "loss": 7.1889, + "epoch": 0.64, + "learning_rate": 0.0002525510204081632, + "loss": 6.845, "step": 567 }, { - "epoch": 0.32, - "learning_rate": 0.00028551401869158875, - "loss": 7.1027, + "epoch": 0.64, + "learning_rate": 0.00025178571428571426, + "loss": 6.7038, "step": 568 }, { - "epoch": 0.32, - "learning_rate": 0.00028528037383177565, - "loss": 7.1397, + "epoch": 0.64, + "learning_rate": 0.0002510204081632653, + "loss": 6.786, "step": 569 }, { - "epoch": 0.32, - "learning_rate": 0.0002850467289719626, - "loss": 7.0473, + "epoch": 0.64, + "learning_rate": 0.0002502551020408163, + "loss": 6.8934, "step": 570 }, { - "epoch": 0.32, - "learning_rate": 0.0002848130841121495, - "loss": 6.9912, + "epoch": 0.64, + "learning_rate": 0.00024948979591836733, + "loss": 6.768, "step": 571 }, { - "epoch": 0.32, - "learning_rate": 0.00028457943925233646, - "loss": 6.9362, + "epoch": 0.64, + "learning_rate": 0.00024872448979591837, + "loss": 6.8869, "step": 572 }, { - "epoch": 0.32, - "learning_rate": 0.0002843457943925233, - "loss": 6.9146, + "epoch": 0.64, + "learning_rate": 0.00024795918367346935, + "loss": 6.7311, "step": 573 }, { - "epoch": 0.32, - "learning_rate": 0.00028411214953271026, - "loss": 6.8332, + "epoch": 0.64, + "learning_rate": 0.0002471938775510204, + "loss": 6.6567, "step": 574 }, { - "epoch": 0.32, - "learning_rate": 0.00028387850467289716, - "loss": 6.9503, + "epoch": 0.64, + "learning_rate": 0.0002464285714285714, + "loss": 6.8348, "step": 575 }, { - "epoch": 0.32, - "learning_rate": 0.0002836448598130841, - "loss": 7.003, + "epoch": 0.65, + "learning_rate": 0.0002456632653061224, + "loss": 6.6822, "step": 576 }, { - "epoch": 0.32, - "learning_rate": 0.000283411214953271, - "loss": 7.0075, + "epoch": 0.65, + "learning_rate": 0.00024489795918367346, + "loss": 6.8552, "step": 577 }, { - "epoch": 0.32, - "learning_rate": 0.0002831775700934579, - "loss": 6.8156, + "epoch": 0.65, + "learning_rate": 0.00024413265306122447, + "loss": 6.9091, "step": 578 }, { - "epoch": 0.32, - "learning_rate": 0.0002829439252336448, - "loss": 7.1451, + "epoch": 0.65, + "learning_rate": 0.00024336734693877548, + "loss": 6.7568, "step": 579 }, { - "epoch": 0.33, - "learning_rate": 0.00028271028037383177, - "loss": 7.1876, + "epoch": 0.65, + "learning_rate": 0.0002426020408163265, + "loss": 6.9483, "step": 580 }, { - "epoch": 0.33, - "learning_rate": 0.00028247663551401867, - "loss": 7.0159, + "epoch": 0.65, + "learning_rate": 0.00024183673469387753, + "loss": 6.9424, "step": 581 }, { - "epoch": 0.33, - "learning_rate": 0.00028224299065420557, - "loss": 6.9662, + "epoch": 0.65, + "learning_rate": 0.00024107142857142857, + "loss": 6.7909, "step": 582 }, { - "epoch": 0.33, - "learning_rate": 0.0002820093457943925, - "loss": 6.8869, + "epoch": 0.65, + "learning_rate": 0.00024030612244897956, + "loss": 6.7228, "step": 583 }, { - "epoch": 0.33, - "learning_rate": 0.0002817757009345794, - "loss": 7.1735, + "epoch": 0.65, + "learning_rate": 0.00023954081632653057, + "loss": 6.8291, "step": 584 }, { - "epoch": 0.33, - "learning_rate": 0.0002815420560747663, - "loss": 7.0412, + "epoch": 0.66, + "learning_rate": 0.0002387755102040816, + "loss": 6.8442, "step": 585 }, { - "epoch": 0.33, - "learning_rate": 0.0002813084112149532, - "loss": 6.8576, + "epoch": 0.66, + "learning_rate": 0.00023801020408163265, + "loss": 6.8812, "step": 586 }, { - "epoch": 0.33, - "learning_rate": 0.0002810747663551402, - "loss": 7.1152, + "epoch": 0.66, + "learning_rate": 0.00023724489795918366, + "loss": 6.8635, "step": 587 }, { - "epoch": 0.33, - "learning_rate": 0.0002808411214953271, - "loss": 7.0188, + "epoch": 0.66, + "learning_rate": 0.00023647959183673467, + "loss": 6.8035, "step": 588 }, { - "epoch": 0.33, - "learning_rate": 0.000280607476635514, - "loss": 7.2766, + "epoch": 0.66, + "learning_rate": 0.00023571428571428569, + "loss": 6.7253, "step": 589 }, { - "epoch": 0.33, - "learning_rate": 0.0002803738317757009, - "loss": 7.3146, + "epoch": 0.66, + "learning_rate": 0.00023494897959183673, + "loss": 6.7687, "step": 590 }, { - "epoch": 0.33, - "learning_rate": 0.00028014018691588784, - "loss": 7.0303, + "epoch": 0.66, + "learning_rate": 0.00023418367346938774, + "loss": 6.8081, "step": 591 }, { - "epoch": 0.33, - "learning_rate": 0.00027990654205607474, - "loss": 7.0401, + "epoch": 0.66, + "learning_rate": 0.00023341836734693875, + "loss": 6.8181, "step": 592 }, { - "epoch": 0.33, - "learning_rate": 0.00027967289719626164, - "loss": 6.7745, + "epoch": 0.66, + "learning_rate": 0.00023265306122448976, + "loss": 6.9183, "step": 593 }, { - "epoch": 0.33, - "learning_rate": 0.0002794392523364486, - "loss": 7.0324, + "epoch": 0.67, + "learning_rate": 0.0002318877551020408, + "loss": 6.605, "step": 594 }, { - "epoch": 0.33, - "learning_rate": 0.0002792056074766355, - "loss": 6.9225, + "epoch": 0.67, + "learning_rate": 0.00023112244897959181, + "loss": 6.7087, "step": 595 }, { - "epoch": 0.33, - "learning_rate": 0.00027897196261682244, - "loss": 6.8352, + "epoch": 0.67, + "learning_rate": 0.00023035714285714285, + "loss": 6.5905, "step": 596 }, { - "epoch": 0.33, - "learning_rate": 0.0002787383177570093, - "loss": 6.8133, + "epoch": 0.67, + "learning_rate": 0.00022959183673469384, + "loss": 6.5806, "step": 597 }, { - "epoch": 0.34, - "learning_rate": 0.00027850467289719624, - "loss": 6.7967, + "epoch": 0.67, + "learning_rate": 0.00022882653061224488, + "loss": 6.497, "step": 598 }, { - "epoch": 0.34, - "learning_rate": 0.00027827102803738314, - "loss": 6.9175, + "epoch": 0.67, + "learning_rate": 0.0002280612244897959, + "loss": 6.3569, "step": 599 }, { - "epoch": 0.34, - "learning_rate": 0.0002780373831775701, - "loss": 6.6765, + "epoch": 0.67, + "learning_rate": 0.00022729591836734693, + "loss": 6.036, "step": 600 }, { - "epoch": 0.34, - "learning_rate": 0.000277803738317757, - "loss": 7.0339, + "epoch": 0.67, + "learning_rate": 0.00022653061224489791, + "loss": 6.993, "step": 601 }, { - "epoch": 0.34, - "learning_rate": 0.0002775700934579439, - "loss": 7.2023, + "epoch": 0.67, + "learning_rate": 0.00022576530612244895, + "loss": 6.7804, "step": 602 }, { - "epoch": 0.34, - "learning_rate": 0.0002773364485981308, - "loss": 6.9408, + "epoch": 0.68, + "learning_rate": 0.000225, + "loss": 6.9282, "step": 603 }, { - "epoch": 0.34, - "learning_rate": 0.00027710280373831775, - "loss": 7.0758, + "epoch": 0.68, + "learning_rate": 0.000224234693877551, + "loss": 6.8609, "step": 604 }, { - "epoch": 0.34, - "learning_rate": 0.00027686915887850465, - "loss": 6.8359, + "epoch": 0.68, + "learning_rate": 0.00022346938775510205, + "loss": 6.9853, "step": 605 }, { - "epoch": 0.34, - "learning_rate": 0.00027663551401869155, - "loss": 6.9642, + "epoch": 0.68, + "learning_rate": 0.00022270408163265303, + "loss": 6.6673, "step": 606 }, { - "epoch": 0.34, - "learning_rate": 0.0002764018691588785, - "loss": 7.2525, + "epoch": 0.68, + "learning_rate": 0.00022193877551020407, + "loss": 6.8504, "step": 607 }, { - "epoch": 0.34, - "learning_rate": 0.0002761682242990654, - "loss": 7.0042, + "epoch": 0.68, + "learning_rate": 0.00022117346938775508, + "loss": 6.8609, "step": 608 }, { - "epoch": 0.34, - "learning_rate": 0.0002759345794392523, - "loss": 7.0956, + "epoch": 0.68, + "learning_rate": 0.00022040816326530612, + "loss": 6.8363, "step": 609 }, { - "epoch": 0.34, - "learning_rate": 0.0002757009345794392, - "loss": 6.914, + "epoch": 0.68, + "learning_rate": 0.0002196428571428571, + "loss": 6.8206, "step": 610 }, { - "epoch": 0.34, - "learning_rate": 0.00027546728971962616, - "loss": 7.2904, + "epoch": 0.68, + "learning_rate": 0.00021887755102040815, + "loss": 6.8077, "step": 611 }, { - "epoch": 0.34, - "learning_rate": 0.00027523364485981306, - "loss": 7.0227, + "epoch": 0.69, + "learning_rate": 0.00021811224489795916, + "loss": 6.8301, "step": 612 }, { - "epoch": 0.34, - "learning_rate": 0.00027499999999999996, - "loss": 7.0562, + "epoch": 0.69, + "learning_rate": 0.0002173469387755102, + "loss": 6.7382, "step": 613 }, { - "epoch": 0.34, - "learning_rate": 0.00027476635514018686, - "loss": 7.1043, + "epoch": 0.69, + "learning_rate": 0.0002165816326530612, + "loss": 6.839, "step": 614 }, { - "epoch": 0.34, - "learning_rate": 0.0002745327102803738, - "loss": 7.044, + "epoch": 0.69, + "learning_rate": 0.00021581632653061222, + "loss": 6.9974, "step": 615 }, { - "epoch": 0.35, - "learning_rate": 0.0002742990654205607, - "loss": 7.0187, + "epoch": 0.69, + "learning_rate": 0.00021505102040816324, + "loss": 6.6923, "step": 616 }, { - "epoch": 0.35, - "learning_rate": 0.0002740654205607476, - "loss": 7.0194, + "epoch": 0.69, + "learning_rate": 0.00021428571428571427, + "loss": 6.8107, "step": 617 }, { - "epoch": 0.35, - "learning_rate": 0.00027383177570093457, - "loss": 6.8312, + "epoch": 0.69, + "learning_rate": 0.0002135204081632653, + "loss": 6.658, "step": 618 }, { - "epoch": 0.35, - "learning_rate": 0.00027359813084112147, - "loss": 7.1007, + "epoch": 0.69, + "learning_rate": 0.0002127551020408163, + "loss": 6.7308, "step": 619 }, { - "epoch": 0.35, - "learning_rate": 0.0002733644859813084, - "loss": 7.0223, + "epoch": 0.7, + "learning_rate": 0.0002119897959183673, + "loss": 6.8192, "step": 620 }, { - "epoch": 0.35, - "learning_rate": 0.00027313084112149527, - "loss": 7.0529, + "epoch": 0.7, + "learning_rate": 0.00021122448979591835, + "loss": 6.8473, "step": 621 }, { - "epoch": 0.35, - "learning_rate": 0.0002728971962616822, - "loss": 7.0219, + "epoch": 0.7, + "learning_rate": 0.0002104591836734694, + "loss": 6.7264, "step": 622 }, { - "epoch": 0.35, - "learning_rate": 0.0002726635514018691, - "loss": 7.0449, + "epoch": 0.7, + "learning_rate": 0.0002096938775510204, + "loss": 6.9049, "step": 623 }, { - "epoch": 0.35, - "learning_rate": 0.0002724299065420561, - "loss": 7.2403, + "epoch": 0.7, + "learning_rate": 0.0002089285714285714, + "loss": 6.7086, "step": 624 }, { - "epoch": 0.35, - "learning_rate": 0.000272196261682243, - "loss": 7.0346, + "epoch": 0.7, + "learning_rate": 0.00020816326530612243, + "loss": 6.8953, "step": 625 }, { - "epoch": 0.35, - "learning_rate": 0.0002719626168224299, - "loss": 7.0924, + "epoch": 0.7, + "learning_rate": 0.00020739795918367347, + "loss": 6.7709, "step": 626 }, { - "epoch": 0.35, - "learning_rate": 0.0002717289719626168, - "loss": 6.9327, + "epoch": 0.7, + "learning_rate": 0.00020663265306122448, + "loss": 6.8614, "step": 627 }, { - "epoch": 0.35, - "learning_rate": 0.00027149532710280373, - "loss": 7.179, + "epoch": 0.7, + "learning_rate": 0.0002058673469387755, + "loss": 6.8522, "step": 628 }, { - "epoch": 0.35, - "learning_rate": 0.00027126168224299063, - "loss": 7.1993, + "epoch": 0.71, + "learning_rate": 0.0002051020408163265, + "loss": 6.8211, "step": 629 }, { - "epoch": 0.35, - "learning_rate": 0.00027102803738317753, - "loss": 6.9298, + "epoch": 0.71, + "learning_rate": 0.00020433673469387754, + "loss": 6.9599, "step": 630 }, { - "epoch": 0.35, - "learning_rate": 0.0002707943925233645, - "loss": 7.026, + "epoch": 0.71, + "learning_rate": 0.00020357142857142856, + "loss": 6.7826, "step": 631 }, { - "epoch": 0.35, - "learning_rate": 0.0002705607476635514, - "loss": 6.8306, + "epoch": 0.71, + "learning_rate": 0.00020280612244897957, + "loss": 6.6684, "step": 632 }, { - "epoch": 0.35, - "learning_rate": 0.0002703271028037383, - "loss": 6.9689, + "epoch": 0.71, + "learning_rate": 0.00020204081632653058, + "loss": 6.8832, "step": 633 }, { - "epoch": 0.36, - "learning_rate": 0.0002700934579439252, - "loss": 7.088, + "epoch": 0.71, + "learning_rate": 0.00020127551020408162, + "loss": 6.7759, "step": 634 }, { - "epoch": 0.36, - "learning_rate": 0.00026985981308411214, - "loss": 6.9823, + "epoch": 0.71, + "learning_rate": 0.00020051020408163263, + "loss": 6.8656, "step": 635 }, { - "epoch": 0.36, - "learning_rate": 0.00026962616822429904, - "loss": 6.9423, + "epoch": 0.71, + "learning_rate": 0.00019974489795918367, + "loss": 6.8239, "step": 636 }, { - "epoch": 0.36, - "learning_rate": 0.00026939252336448594, - "loss": 6.9123, + "epoch": 0.71, + "learning_rate": 0.00019897959183673466, + "loss": 6.8342, "step": 637 }, { - "epoch": 0.36, - "learning_rate": 0.00026915887850467284, - "loss": 6.7636, + "epoch": 0.72, + "learning_rate": 0.0001982142857142857, + "loss": 6.6465, "step": 638 }, { - "epoch": 0.36, - "learning_rate": 0.0002689252336448598, - "loss": 7.1612, + "epoch": 0.72, + "learning_rate": 0.0001974489795918367, + "loss": 6.9261, "step": 639 }, { - "epoch": 0.36, - "learning_rate": 0.0002686915887850467, - "loss": 7.0901, + "epoch": 0.72, + "learning_rate": 0.00019668367346938775, + "loss": 6.671, "step": 640 }, { - "epoch": 0.36, - "learning_rate": 0.0002684579439252336, - "loss": 6.9211, + "epoch": 0.72, + "learning_rate": 0.00019591836734693873, + "loss": 6.7317, "step": 641 }, { - "epoch": 0.36, - "learning_rate": 0.00026822429906542055, - "loss": 6.9711, + "epoch": 0.72, + "learning_rate": 0.00019515306122448977, + "loss": 6.922, "step": 642 }, { - "epoch": 0.36, - "learning_rate": 0.00026799065420560745, - "loss": 6.9598, + "epoch": 0.72, + "learning_rate": 0.0001943877551020408, + "loss": 6.8189, "step": 643 }, { - "epoch": 0.36, - "learning_rate": 0.0002677570093457944, - "loss": 6.9816, + "epoch": 0.72, + "learning_rate": 0.00019362244897959182, + "loss": 6.5989, "step": 644 }, { - "epoch": 0.36, - "learning_rate": 0.00026752336448598125, - "loss": 6.5757, + "epoch": 0.72, + "learning_rate": 0.00019285714285714286, + "loss": 6.5903, "step": 645 }, { - "epoch": 0.36, - "learning_rate": 0.0002672897196261682, - "loss": 6.9134, + "epoch": 0.72, + "learning_rate": 0.00019209183673469385, + "loss": 6.7557, "step": 646 }, { - "epoch": 0.36, - "learning_rate": 0.0002670560747663551, - "loss": 6.8526, + "epoch": 0.73, + "learning_rate": 0.0001913265306122449, + "loss": 6.5452, "step": 647 }, { - "epoch": 0.36, - "learning_rate": 0.00026682242990654206, - "loss": 7.0134, + "epoch": 0.73, + "learning_rate": 0.0001905612244897959, + "loss": 6.4467, "step": 648 }, { - "epoch": 0.36, - "learning_rate": 0.00026658878504672896, - "loss": 7.0689, + "epoch": 0.73, + "learning_rate": 0.00018979591836734694, + "loss": 6.6444, "step": 649 }, { - "epoch": 0.36, - "learning_rate": 0.00026635514018691586, - "loss": 6.5794, + "epoch": 0.73, + "learning_rate": 0.00018903061224489793, + "loss": 6.3974, "step": 650 }, { - "epoch": 0.36, - "learning_rate": 0.00026612149532710276, - "loss": 7.1871, + "epoch": 0.73, + "learning_rate": 0.00018826530612244896, + "loss": 7.0178, "step": 651 }, { - "epoch": 0.37, - "learning_rate": 0.0002658878504672897, - "loss": 6.9806, + "epoch": 0.73, + "learning_rate": 0.00018749999999999998, + "loss": 6.9998, "step": 652 }, { - "epoch": 0.37, - "learning_rate": 0.0002656542056074766, - "loss": 7.0562, + "epoch": 0.73, + "learning_rate": 0.00018673469387755102, + "loss": 6.8985, "step": 653 }, { - "epoch": 0.37, - "learning_rate": 0.0002654205607476635, - "loss": 7.0863, + "epoch": 0.73, + "learning_rate": 0.00018596938775510203, + "loss": 6.8224, "step": 654 }, { - "epoch": 0.37, - "learning_rate": 0.00026518691588785047, - "loss": 6.9552, + "epoch": 0.73, + "learning_rate": 0.00018520408163265304, + "loss": 6.8712, "step": 655 }, { - "epoch": 0.37, - "learning_rate": 0.00026495327102803737, - "loss": 6.9493, + "epoch": 0.74, + "learning_rate": 0.00018443877551020405, + "loss": 6.8948, "step": 656 }, { - "epoch": 0.37, - "learning_rate": 0.00026471962616822427, - "loss": 6.8604, + "epoch": 0.74, + "learning_rate": 0.0001836734693877551, + "loss": 6.7172, "step": 657 }, { - "epoch": 0.37, - "learning_rate": 0.00026448598130841117, - "loss": 7.1802, + "epoch": 0.74, + "learning_rate": 0.0001829081632653061, + "loss": 6.9833, "step": 658 }, { - "epoch": 0.37, - "learning_rate": 0.0002642523364485981, - "loss": 7.042, + "epoch": 0.74, + "learning_rate": 0.00018214285714285712, + "loss": 6.7223, "step": 659 }, { - "epoch": 0.37, - "learning_rate": 0.000264018691588785, - "loss": 7.0367, + "epoch": 0.74, + "learning_rate": 0.00018137755102040813, + "loss": 6.722, "step": 660 }, { - "epoch": 0.37, - "learning_rate": 0.0002637850467289719, - "loss": 7.1622, + "epoch": 0.74, + "learning_rate": 0.00018061224489795917, + "loss": 6.7625, "step": 661 }, { - "epoch": 0.37, - "learning_rate": 0.0002635514018691588, - "loss": 6.9996, + "epoch": 0.74, + "learning_rate": 0.0001798469387755102, + "loss": 6.8163, "step": 662 }, { - "epoch": 0.37, - "learning_rate": 0.0002633177570093458, - "loss": 7.0352, + "epoch": 0.74, + "learning_rate": 0.00017908163265306122, + "loss": 6.9104, "step": 663 }, { - "epoch": 0.37, - "learning_rate": 0.0002630841121495327, - "loss": 7.1075, + "epoch": 0.74, + "learning_rate": 0.0001783163265306122, + "loss": 6.6764, "step": 664 }, { - "epoch": 0.37, - "learning_rate": 0.0002628504672897196, - "loss": 6.9435, + "epoch": 0.75, + "learning_rate": 0.00017755102040816325, + "loss": 6.7642, "step": 665 }, { - "epoch": 0.37, - "learning_rate": 0.00026261682242990653, - "loss": 6.9959, + "epoch": 0.75, + "learning_rate": 0.00017678571428571428, + "loss": 6.8771, "step": 666 }, { - "epoch": 0.37, - "learning_rate": 0.00026238317757009343, - "loss": 7.0819, + "epoch": 0.75, + "learning_rate": 0.0001760204081632653, + "loss": 6.7731, "step": 667 }, { - "epoch": 0.37, - "learning_rate": 0.0002621495327102804, - "loss": 6.9272, + "epoch": 0.75, + "learning_rate": 0.0001752551020408163, + "loss": 6.7133, "step": 668 }, { - "epoch": 0.38, - "learning_rate": 0.00026191588785046723, - "loss": 6.9172, + "epoch": 0.75, + "learning_rate": 0.00017448979591836732, + "loss": 6.8526, "step": 669 }, { - "epoch": 0.38, - "learning_rate": 0.0002616822429906542, - "loss": 6.9313, + "epoch": 0.75, + "learning_rate": 0.00017372448979591836, + "loss": 6.7391, "step": 670 }, { - "epoch": 0.38, - "learning_rate": 0.0002614485981308411, - "loss": 7.2484, + "epoch": 0.75, + "learning_rate": 0.00017295918367346937, + "loss": 6.6748, "step": 671 }, { - "epoch": 0.38, - "learning_rate": 0.00026121495327102804, - "loss": 7.021, + "epoch": 0.75, + "learning_rate": 0.0001721938775510204, + "loss": 6.8101, "step": 672 }, { - "epoch": 0.38, - "learning_rate": 0.00026098130841121494, - "loss": 6.9219, + "epoch": 0.75, + "learning_rate": 0.0001714285714285714, + "loss": 6.7354, "step": 673 }, { - "epoch": 0.38, - "learning_rate": 0.00026074766355140184, - "loss": 6.8895, + "epoch": 0.76, + "learning_rate": 0.00017066326530612244, + "loss": 6.7228, "step": 674 }, { - "epoch": 0.38, - "learning_rate": 0.00026051401869158874, - "loss": 7.0309, + "epoch": 0.76, + "learning_rate": 0.00016989795918367345, + "loss": 6.8262, "step": 675 }, { - "epoch": 0.38, - "learning_rate": 0.0002602803738317757, - "loss": 6.9955, + "epoch": 0.76, + "learning_rate": 0.0001691326530612245, + "loss": 6.8924, "step": 676 }, { - "epoch": 0.38, - "learning_rate": 0.0002600467289719626, - "loss": 6.8417, + "epoch": 0.76, + "learning_rate": 0.00016836734693877547, + "loss": 6.765, "step": 677 }, { - "epoch": 0.38, - "learning_rate": 0.0002598130841121495, - "loss": 7.0601, + "epoch": 0.76, + "learning_rate": 0.00016760204081632651, + "loss": 6.5957, "step": 678 }, { - "epoch": 0.38, - "learning_rate": 0.00025957943925233645, - "loss": 6.9596, + "epoch": 0.76, + "learning_rate": 0.00016683673469387753, + "loss": 6.7606, "step": 679 }, { - "epoch": 0.38, - "learning_rate": 0.00025934579439252335, - "loss": 7.0703, + "epoch": 0.76, + "learning_rate": 0.00016607142857142857, + "loss": 6.7172, "step": 680 }, { - "epoch": 0.38, - "learning_rate": 0.00025911214953271025, - "loss": 7.1729, + "epoch": 0.76, + "learning_rate": 0.00016530612244897955, + "loss": 6.8478, "step": 681 }, { - "epoch": 0.38, - "learning_rate": 0.00025887850467289715, - "loss": 6.9599, + "epoch": 0.76, + "learning_rate": 0.0001645408163265306, + "loss": 6.7654, "step": 682 }, { - "epoch": 0.38, - "learning_rate": 0.0002586448598130841, - "loss": 6.9773, + "epoch": 0.77, + "learning_rate": 0.00016377551020408163, + "loss": 6.7287, "step": 683 }, { - "epoch": 0.38, - "learning_rate": 0.000258411214953271, - "loss": 6.9911, + "epoch": 0.77, + "learning_rate": 0.00016301020408163264, + "loss": 6.8379, "step": 684 }, { - "epoch": 0.38, - "learning_rate": 0.0002581775700934579, - "loss": 7.1956, + "epoch": 0.77, + "learning_rate": 0.00016224489795918368, + "loss": 6.8713, "step": 685 }, { - "epoch": 0.38, - "learning_rate": 0.0002579439252336448, - "loss": 7.0124, + "epoch": 0.77, + "learning_rate": 0.00016147959183673467, + "loss": 6.6862, "step": 686 }, { - "epoch": 0.39, - "learning_rate": 0.00025771028037383176, - "loss": 6.8336, + "epoch": 0.77, + "learning_rate": 0.0001607142857142857, + "loss": 6.6621, "step": 687 }, { - "epoch": 0.39, - "learning_rate": 0.00025747663551401866, - "loss": 7.0179, + "epoch": 0.77, + "learning_rate": 0.00015994897959183672, + "loss": 6.9119, "step": 688 }, { - "epoch": 0.39, - "learning_rate": 0.00025724299065420556, - "loss": 7.2718, + "epoch": 0.77, + "learning_rate": 0.00015918367346938776, + "loss": 6.7827, "step": 689 }, { - "epoch": 0.39, - "learning_rate": 0.0002570093457943925, - "loss": 7.0003, + "epoch": 0.77, + "learning_rate": 0.00015841836734693874, + "loss": 6.8166, "step": 690 }, { - "epoch": 0.39, - "learning_rate": 0.0002567757009345794, - "loss": 6.9629, + "epoch": 0.77, + "learning_rate": 0.00015765306122448978, + "loss": 6.6425, "step": 691 }, { - "epoch": 0.39, - "learning_rate": 0.00025654205607476637, - "loss": 7.1279, + "epoch": 0.78, + "learning_rate": 0.0001568877551020408, + "loss": 6.7403, "step": 692 }, { - "epoch": 0.39, - "learning_rate": 0.0002563084112149532, - "loss": 6.9728, + "epoch": 0.78, + "learning_rate": 0.00015612244897959183, + "loss": 6.5451, "step": 693 }, { - "epoch": 0.39, - "learning_rate": 0.00025607476635514017, - "loss": 7.0004, + "epoch": 0.78, + "learning_rate": 0.00015535714285714285, + "loss": 6.7156, "step": 694 }, { - "epoch": 0.39, - "learning_rate": 0.00025584112149532707, - "loss": 6.9956, + "epoch": 0.78, + "learning_rate": 0.00015459183673469386, + "loss": 6.7341, "step": 695 }, { - "epoch": 0.39, - "learning_rate": 0.000255607476635514, - "loss": 7.0694, + "epoch": 0.78, + "learning_rate": 0.00015382653061224487, + "loss": 6.7433, "step": 696 }, { - "epoch": 0.39, - "learning_rate": 0.0002553738317757009, - "loss": 6.9353, + "epoch": 0.78, + "learning_rate": 0.0001530612244897959, + "loss": 6.5209, "step": 697 }, { - "epoch": 0.39, - "learning_rate": 0.0002551401869158878, - "loss": 6.7548, + "epoch": 0.78, + "learning_rate": 0.00015229591836734692, + "loss": 6.3372, "step": 698 }, { - "epoch": 0.39, - "learning_rate": 0.0002549065420560747, - "loss": 6.6991, + "epoch": 0.78, + "learning_rate": 0.00015153061224489794, + "loss": 6.4115, "step": 699 }, { - "epoch": 0.39, - "learning_rate": 0.0002546728971962617, - "loss": 6.7345, + "epoch": 0.78, + "learning_rate": 0.00015076530612244895, + "loss": 6.0466, "step": 700 }, { - "epoch": 0.39, - "learning_rate": 0.0002544392523364486, - "loss": 6.9465, + "epoch": 0.79, + "learning_rate": 0.00015, + "loss": 6.8879, "step": 701 }, { - "epoch": 0.39, - "learning_rate": 0.0002542056074766355, - "loss": 7.0243, + "epoch": 0.79, + "learning_rate": 0.00014923469387755103, + "loss": 6.8209, "step": 702 }, { - "epoch": 0.39, - "learning_rate": 0.00025397196261682243, - "loss": 6.9317, + "epoch": 0.79, + "learning_rate": 0.00014846938775510204, + "loss": 6.767, "step": 703 }, { - "epoch": 0.39, - "learning_rate": 0.00025373831775700933, - "loss": 7.2064, + "epoch": 0.79, + "learning_rate": 0.00014770408163265305, + "loss": 6.6467, "step": 704 }, { - "epoch": 0.4, - "learning_rate": 0.00025350467289719623, - "loss": 6.9294, + "epoch": 0.79, + "learning_rate": 0.00014693877551020406, + "loss": 6.771, "step": 705 }, { - "epoch": 0.4, - "learning_rate": 0.00025327102803738313, - "loss": 7.1521, + "epoch": 0.79, + "learning_rate": 0.0001461734693877551, + "loss": 6.8762, "step": 706 }, { - "epoch": 0.4, - "learning_rate": 0.0002530373831775701, - "loss": 7.0257, + "epoch": 0.79, + "learning_rate": 0.00014540816326530611, + "loss": 6.8515, "step": 707 }, { - "epoch": 0.4, - "learning_rate": 0.000252803738317757, - "loss": 7.2038, + "epoch": 0.79, + "learning_rate": 0.00014464285714285713, + "loss": 6.9306, "step": 708 }, { - "epoch": 0.4, - "learning_rate": 0.0002525700934579439, - "loss": 6.8963, + "epoch": 0.79, + "learning_rate": 0.00014387755102040814, + "loss": 6.8259, "step": 709 }, { - "epoch": 0.4, - "learning_rate": 0.0002523364485981308, - "loss": 7.0408, + "epoch": 0.8, + "learning_rate": 0.00014311224489795918, + "loss": 6.7757, "step": 710 }, { - "epoch": 0.4, - "learning_rate": 0.00025210280373831774, - "loss": 6.9914, + "epoch": 0.8, + "learning_rate": 0.0001423469387755102, + "loss": 6.753, "step": 711 }, { - "epoch": 0.4, - "learning_rate": 0.00025186915887850464, - "loss": 7.0, + "epoch": 0.8, + "learning_rate": 0.0001415816326530612, + "loss": 6.6252, "step": 712 }, { - "epoch": 0.4, - "learning_rate": 0.00025163551401869154, - "loss": 6.8706, + "epoch": 0.8, + "learning_rate": 0.00014081632653061224, + "loss": 6.6745, "step": 713 }, { - "epoch": 0.4, - "learning_rate": 0.0002514018691588785, - "loss": 7.2142, + "epoch": 0.8, + "learning_rate": 0.00014005102040816326, + "loss": 6.7449, "step": 714 }, { - "epoch": 0.4, - "learning_rate": 0.0002511682242990654, - "loss": 7.0591, + "epoch": 0.8, + "learning_rate": 0.00013928571428571427, + "loss": 6.8163, "step": 715 }, { - "epoch": 0.4, - "learning_rate": 0.00025093457943925235, - "loss": 6.8853, + "epoch": 0.8, + "learning_rate": 0.00013852040816326528, + "loss": 6.8405, "step": 716 }, { - "epoch": 0.4, - "learning_rate": 0.0002507009345794392, - "loss": 6.9267, + "epoch": 0.8, + "learning_rate": 0.00013775510204081632, + "loss": 7.0215, "step": 717 }, { - "epoch": 0.4, - "learning_rate": 0.00025046728971962615, - "loss": 7.0836, + "epoch": 0.8, + "learning_rate": 0.00013698979591836733, + "loss": 6.7862, "step": 718 }, { - "epoch": 0.4, - "learning_rate": 0.00025023364485981305, - "loss": 6.8214, + "epoch": 0.81, + "learning_rate": 0.00013622448979591834, + "loss": 6.6331, "step": 719 }, { - "epoch": 0.4, - "learning_rate": 0.00025, - "loss": 7.028, + "epoch": 0.81, + "learning_rate": 0.00013545918367346936, + "loss": 6.717, "step": 720 }, { - "epoch": 0.4, - "learning_rate": 0.0002497663551401869, - "loss": 7.1094, + "epoch": 0.81, + "learning_rate": 0.0001346938775510204, + "loss": 6.8046, "step": 721 }, { - "epoch": 0.4, - "learning_rate": 0.0002495327102803738, - "loss": 7.1828, + "epoch": 0.81, + "learning_rate": 0.00013392857142857144, + "loss": 6.685, "step": 722 }, { - "epoch": 0.41, - "learning_rate": 0.0002492990654205607, - "loss": 7.1832, + "epoch": 0.81, + "learning_rate": 0.00013316326530612245, + "loss": 6.6772, "step": 723 }, { - "epoch": 0.41, - "learning_rate": 0.00024906542056074766, - "loss": 7.0369, + "epoch": 0.81, + "learning_rate": 0.00013239795918367346, + "loss": 6.8847, "step": 724 }, { - "epoch": 0.41, - "learning_rate": 0.00024883177570093456, - "loss": 6.9629, + "epoch": 0.81, + "learning_rate": 0.00013163265306122447, + "loss": 6.7749, "step": 725 }, { - "epoch": 0.41, - "learning_rate": 0.00024859813084112146, - "loss": 6.8586, + "epoch": 0.81, + "learning_rate": 0.0001308673469387755, + "loss": 6.8872, "step": 726 }, { - "epoch": 0.41, - "learning_rate": 0.0002483644859813084, - "loss": 6.6914, + "epoch": 0.82, + "learning_rate": 0.00013010204081632652, + "loss": 6.8119, "step": 727 }, { - "epoch": 0.41, - "learning_rate": 0.0002481308411214953, - "loss": 6.9927, + "epoch": 0.82, + "learning_rate": 0.00012933673469387754, + "loss": 6.676, "step": 728 }, { - "epoch": 0.41, - "learning_rate": 0.0002478971962616822, - "loss": 6.938, + "epoch": 0.82, + "learning_rate": 0.00012857142857142855, + "loss": 6.7485, "step": 729 }, { - "epoch": 0.41, - "learning_rate": 0.0002476635514018691, - "loss": 7.104, + "epoch": 0.82, + "learning_rate": 0.0001278061224489796, + "loss": 6.8757, "step": 730 }, { - "epoch": 0.41, - "learning_rate": 0.00024742990654205607, - "loss": 7.0555, + "epoch": 0.82, + "learning_rate": 0.0001270408163265306, + "loss": 7.0093, "step": 731 }, { - "epoch": 0.41, - "learning_rate": 0.00024719626168224297, - "loss": 6.9451, + "epoch": 0.82, + "learning_rate": 0.0001262755102040816, + "loss": 6.8234, "step": 732 }, { - "epoch": 0.41, - "learning_rate": 0.00024696261682242987, - "loss": 7.1299, + "epoch": 0.82, + "learning_rate": 0.00012551020408163265, + "loss": 6.7583, "step": 733 }, { - "epoch": 0.41, - "learning_rate": 0.00024672897196261677, - "loss": 7.0251, + "epoch": 0.82, + "learning_rate": 0.00012474489795918366, + "loss": 6.9749, "step": 734 }, { - "epoch": 0.41, - "learning_rate": 0.0002464953271028037, - "loss": 6.9898, + "epoch": 0.82, + "learning_rate": 0.00012397959183673468, + "loss": 6.7102, "step": 735 }, { - "epoch": 0.41, - "learning_rate": 0.0002462616822429906, - "loss": 6.9505, + "epoch": 0.83, + "learning_rate": 0.0001232142857142857, + "loss": 6.7685, "step": 736 }, { - "epoch": 0.41, - "learning_rate": 0.0002460280373831775, - "loss": 6.9652, + "epoch": 0.83, + "learning_rate": 0.00012244897959183673, + "loss": 6.7942, "step": 737 }, { - "epoch": 0.41, - "learning_rate": 0.0002457943925233645, - "loss": 6.9666, + "epoch": 0.83, + "learning_rate": 0.00012168367346938774, + "loss": 6.8079, "step": 738 }, { - "epoch": 0.41, - "learning_rate": 0.0002455607476635514, - "loss": 6.9593, + "epoch": 0.83, + "learning_rate": 0.00012091836734693877, + "loss": 6.695, "step": 739 }, { - "epoch": 0.41, - "learning_rate": 0.00024532710280373833, - "loss": 6.9893, + "epoch": 0.83, + "learning_rate": 0.00012015306122448978, + "loss": 6.8162, "step": 740 }, { - "epoch": 0.42, - "learning_rate": 0.0002450934579439252, - "loss": 7.2057, + "epoch": 0.83, + "learning_rate": 0.0001193877551020408, + "loss": 6.7592, "step": 741 }, { - "epoch": 0.42, - "learning_rate": 0.00024485981308411213, - "loss": 7.1942, + "epoch": 0.83, + "learning_rate": 0.00011862244897959183, + "loss": 6.79, "step": 742 }, { - "epoch": 0.42, - "learning_rate": 0.00024462616822429903, - "loss": 7.0353, + "epoch": 0.83, + "learning_rate": 0.00011785714285714284, + "loss": 6.6227, "step": 743 }, { - "epoch": 0.42, - "learning_rate": 0.000244392523364486, - "loss": 7.0544, + "epoch": 0.83, + "learning_rate": 0.00011709183673469387, + "loss": 6.7437, "step": 744 }, { - "epoch": 0.42, - "learning_rate": 0.0002441588785046729, - "loss": 7.0494, + "epoch": 0.84, + "learning_rate": 0.00011632653061224488, + "loss": 6.6668, "step": 745 }, { - "epoch": 0.42, - "learning_rate": 0.0002439252336448598, - "loss": 6.7, + "epoch": 0.84, + "learning_rate": 0.00011556122448979591, + "loss": 6.8181, "step": 746 }, { - "epoch": 0.42, - "learning_rate": 0.00024369158878504671, - "loss": 6.886, + "epoch": 0.84, + "learning_rate": 0.00011479591836734692, + "loss": 6.4942, "step": 747 }, { - "epoch": 0.42, - "learning_rate": 0.00024345794392523364, - "loss": 6.8199, + "epoch": 0.84, + "learning_rate": 0.00011403061224489795, + "loss": 6.4618, "step": 748 }, { - "epoch": 0.42, - "learning_rate": 0.00024322429906542051, - "loss": 6.6438, + "epoch": 0.84, + "learning_rate": 0.00011326530612244896, + "loss": 6.4448, "step": 749 }, { - "epoch": 0.42, - "learning_rate": 0.00024299065420560744, - "loss": 6.4659, + "epoch": 0.84, + "learning_rate": 0.0001125, + "loss": 6.3197, "step": 750 }, { - "epoch": 0.42, - "learning_rate": 0.00024275700934579437, - "loss": 7.1466, + "epoch": 0.84, + "learning_rate": 0.00011173469387755102, + "loss": 6.8133, "step": 751 }, { - "epoch": 0.42, - "learning_rate": 0.0002425233644859813, - "loss": 7.0757, + "epoch": 0.84, + "learning_rate": 0.00011096938775510204, + "loss": 6.8236, "step": 752 }, { - "epoch": 0.42, - "learning_rate": 0.0002422897196261682, - "loss": 7.0266, + "epoch": 0.84, + "learning_rate": 0.00011020408163265306, + "loss": 6.8616, "step": 753 }, { - "epoch": 0.42, - "learning_rate": 0.00024205607476635512, - "loss": 7.0512, + "epoch": 0.85, + "learning_rate": 0.00010943877551020407, + "loss": 6.7366, "step": 754 }, { - "epoch": 0.42, - "learning_rate": 0.00024182242990654205, - "loss": 6.914, + "epoch": 0.85, + "learning_rate": 0.0001086734693877551, + "loss": 6.7812, "step": 755 }, { - "epoch": 0.42, - "learning_rate": 0.00024158878504672895, - "loss": 7.1503, + "epoch": 0.85, + "learning_rate": 0.00010790816326530611, + "loss": 6.8787, "step": 756 }, { - "epoch": 0.42, - "learning_rate": 0.00024135514018691585, - "loss": 7.0942, + "epoch": 0.85, + "learning_rate": 0.00010714285714285714, + "loss": 6.7598, "step": 757 }, { - "epoch": 0.42, - "learning_rate": 0.00024112149532710278, - "loss": 7.151, + "epoch": 0.85, + "learning_rate": 0.00010637755102040815, + "loss": 6.6895, "step": 758 }, { - "epoch": 0.43, - "learning_rate": 0.0002408878504672897, - "loss": 6.8339, + "epoch": 0.85, + "learning_rate": 0.00010561224489795918, + "loss": 6.7956, "step": 759 }, { - "epoch": 0.43, - "learning_rate": 0.00024065420560747663, - "loss": 6.8857, + "epoch": 0.85, + "learning_rate": 0.0001048469387755102, + "loss": 6.724, "step": 760 }, { - "epoch": 0.43, - "learning_rate": 0.0002404205607476635, - "loss": 6.894, + "epoch": 0.85, + "learning_rate": 0.00010408163265306121, + "loss": 6.7332, "step": 761 }, { - "epoch": 0.43, - "learning_rate": 0.00024018691588785043, - "loss": 6.89, + "epoch": 0.85, + "learning_rate": 0.00010331632653061224, + "loss": 6.7386, "step": 762 }, { - "epoch": 0.43, - "learning_rate": 0.00023995327102803736, - "loss": 7.076, + "epoch": 0.86, + "learning_rate": 0.00010255102040816325, + "loss": 6.5965, "step": 763 }, { - "epoch": 0.43, - "learning_rate": 0.00023971962616822429, - "loss": 6.967, + "epoch": 0.86, + "learning_rate": 0.00010178571428571428, + "loss": 6.5961, "step": 764 }, { - "epoch": 0.43, - "learning_rate": 0.00023948598130841119, - "loss": 6.9879, + "epoch": 0.86, + "learning_rate": 0.00010102040816326529, + "loss": 6.7318, "step": 765 }, { - "epoch": 0.43, - "learning_rate": 0.0002392523364485981, - "loss": 6.8462, + "epoch": 0.86, + "learning_rate": 0.00010025510204081632, + "loss": 6.6829, "step": 766 }, { - "epoch": 0.43, - "learning_rate": 0.00023901869158878504, - "loss": 7.0929, + "epoch": 0.86, + "learning_rate": 9.948979591836733e-05, + "loss": 6.6803, "step": 767 }, { - "epoch": 0.43, - "learning_rate": 0.00023878504672897194, - "loss": 6.8328, + "epoch": 0.86, + "learning_rate": 9.872448979591835e-05, + "loss": 6.7533, "step": 768 }, { - "epoch": 0.43, - "learning_rate": 0.00023855140186915884, - "loss": 7.1, + "epoch": 0.86, + "learning_rate": 9.795918367346937e-05, + "loss": 6.8555, "step": 769 }, { - "epoch": 0.43, - "learning_rate": 0.00023831775700934577, - "loss": 6.9396, + "epoch": 0.86, + "learning_rate": 9.71938775510204e-05, + "loss": 6.652, "step": 770 }, { - "epoch": 0.43, - "learning_rate": 0.0002380841121495327, - "loss": 7.0375, + "epoch": 0.86, + "learning_rate": 9.642857142857143e-05, + "loss": 6.8121, "step": 771 }, { - "epoch": 0.43, - "learning_rate": 0.00023785046728971962, - "loss": 6.9884, + "epoch": 0.87, + "learning_rate": 9.566326530612244e-05, + "loss": 6.8903, "step": 772 }, { - "epoch": 0.43, - "learning_rate": 0.0002376168224299065, - "loss": 6.8863, + "epoch": 0.87, + "learning_rate": 9.489795918367347e-05, + "loss": 6.8009, "step": 773 }, { - "epoch": 0.43, - "learning_rate": 0.00023738317757009342, - "loss": 6.9729, + "epoch": 0.87, + "learning_rate": 9.413265306122448e-05, + "loss": 6.6108, "step": 774 }, { - "epoch": 0.43, - "learning_rate": 0.00023714953271028035, - "loss": 6.9121, + "epoch": 0.87, + "learning_rate": 9.336734693877551e-05, + "loss": 6.5482, "step": 775 }, { - "epoch": 0.43, - "learning_rate": 0.00023691588785046728, - "loss": 6.9866, + "epoch": 0.87, + "learning_rate": 9.260204081632652e-05, + "loss": 6.6848, "step": 776 }, { - "epoch": 0.44, - "learning_rate": 0.00023668224299065418, - "loss": 6.9531, + "epoch": 0.87, + "learning_rate": 9.183673469387755e-05, + "loss": 6.6021, "step": 777 }, { - "epoch": 0.44, - "learning_rate": 0.0002364485981308411, - "loss": 7.0502, + "epoch": 0.87, + "learning_rate": 9.107142857142856e-05, + "loss": 6.7729, "step": 778 }, { - "epoch": 0.44, - "learning_rate": 0.00023621495327102803, - "loss": 7.0478, + "epoch": 0.87, + "learning_rate": 9.030612244897958e-05, + "loss": 6.761, "step": 779 }, { - "epoch": 0.44, - "learning_rate": 0.00023598130841121493, - "loss": 6.9441, + "epoch": 0.87, + "learning_rate": 8.954081632653061e-05, + "loss": 6.704, "step": 780 }, { - "epoch": 0.44, - "learning_rate": 0.00023574766355140183, - "loss": 7.0433, + "epoch": 0.88, + "learning_rate": 8.877551020408162e-05, + "loss": 6.7131, "step": 781 }, { - "epoch": 0.44, - "learning_rate": 0.00023551401869158876, - "loss": 6.9838, + "epoch": 0.88, + "learning_rate": 8.801020408163265e-05, + "loss": 6.6933, "step": 782 }, { - "epoch": 0.44, - "learning_rate": 0.00023528037383177569, - "loss": 7.0477, + "epoch": 0.88, + "learning_rate": 8.724489795918366e-05, + "loss": 6.7786, "step": 783 }, { - "epoch": 0.44, - "learning_rate": 0.0002350467289719626, - "loss": 7.154, + "epoch": 0.88, + "learning_rate": 8.647959183673469e-05, + "loss": 6.6551, "step": 784 }, { - "epoch": 0.44, - "learning_rate": 0.00023481308411214949, - "loss": 6.8953, + "epoch": 0.88, + "learning_rate": 8.57142857142857e-05, + "loss": 6.7676, "step": 785 }, { - "epoch": 0.44, - "learning_rate": 0.0002345794392523364, - "loss": 6.9698, + "epoch": 0.88, + "learning_rate": 8.494897959183672e-05, + "loss": 6.7696, "step": 786 }, { - "epoch": 0.44, - "learning_rate": 0.00023434579439252334, - "loss": 6.8022, + "epoch": 0.88, + "learning_rate": 8.418367346938774e-05, + "loss": 6.7007, "step": 787 }, { - "epoch": 0.44, - "learning_rate": 0.00023411214953271027, - "loss": 7.1341, + "epoch": 0.88, + "learning_rate": 8.341836734693876e-05, + "loss": 6.7635, "step": 788 }, { - "epoch": 0.44, - "learning_rate": 0.00023387850467289717, - "loss": 6.948, + "epoch": 0.88, + "learning_rate": 8.265306122448978e-05, + "loss": 6.714, "step": 789 }, - { - "epoch": 0.44, - "learning_rate": 0.0002336448598130841, - "loss": 7.1699, - "step": 790 - }, - { - "epoch": 0.44, - "learning_rate": 0.00023341121495327102, - "loss": 6.9991, - "step": 791 - }, - { - "epoch": 0.44, - "learning_rate": 0.00023317757009345792, - "loss": 6.7534, - "step": 792 - }, - { - "epoch": 0.44, - "learning_rate": 0.00023294392523364482, - "loss": 7.0425, - "step": 793 - }, - { - "epoch": 0.45, - "learning_rate": 0.00023271028037383175, - "loss": 7.0147, - "step": 794 - }, - { - "epoch": 0.45, - "learning_rate": 0.00023247663551401868, - "loss": 6.908, - "step": 795 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002322429906542056, - "loss": 6.8493, - "step": 796 - }, - { - "epoch": 0.45, - "learning_rate": 0.00023200934579439248, - "loss": 6.5842, - "step": 797 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002317757009345794, - "loss": 6.8756, - "step": 798 - }, - { - "epoch": 0.45, - "learning_rate": 0.00023154205607476633, - "loss": 6.8269, - "step": 799 - }, - { - "epoch": 0.45, - "learning_rate": 0.00023130841121495326, - "loss": 6.6393, - "step": 800 - }, - { - "epoch": 0.45, - "learning_rate": 0.00023107476635514016, - "loss": 7.2537, - "step": 801 - }, - { - "epoch": 0.45, - "learning_rate": 0.00023084112149532709, - "loss": 7.1376, - "step": 802 - }, - { - "epoch": 0.45, - "learning_rate": 0.000230607476635514, - "loss": 7.1778, - "step": 803 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002303738317757009, - "loss": 7.0017, - "step": 804 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002301401869158878, - "loss": 7.2491, - "step": 805 - }, - { - "epoch": 0.45, - "learning_rate": 0.00022990654205607474, - "loss": 7.0915, - "step": 806 - }, - { - "epoch": 0.45, - "learning_rate": 0.00022967289719626167, - "loss": 6.9941, - "step": 807 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002294392523364486, - "loss": 6.9645, - "step": 808 - }, - { - "epoch": 0.45, - "learning_rate": 0.00022920560747663547, - "loss": 7.1642, - "step": 809 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002289719626168224, - "loss": 7.192, - "step": 810 - }, - { - "epoch": 0.45, - "learning_rate": 0.00022873831775700932, - "loss": 7.0638, - "step": 811 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022850467289719625, - "loss": 7.1556, - "step": 812 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022827102803738315, - "loss": 7.0088, - "step": 813 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022803738317757008, - "loss": 6.9954, - "step": 814 - }, - { - "epoch": 0.46, - "learning_rate": 0.000227803738317757, - "loss": 7.0348, - "step": 815 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002275700934579439, - "loss": 7.0125, - "step": 816 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002273364485981308, - "loss": 6.9714, - "step": 817 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022710280373831773, - "loss": 7.0442, - "step": 818 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022686915887850466, - "loss": 7.1709, - "step": 819 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022663551401869158, - "loss": 6.9699, - "step": 820 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022640186915887846, - "loss": 7.1657, - "step": 821 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022616822429906539, - "loss": 7.006, - "step": 822 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002259345794392523, - "loss": 7.1522, - "step": 823 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022570093457943924, - "loss": 7.0628, - "step": 824 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022546728971962614, - "loss": 7.0572, - "step": 825 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022523364485981307, - "loss": 6.9715, - "step": 826 - }, - { - "epoch": 0.46, - "learning_rate": 0.000225, - "loss": 7.048, - "step": 827 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002247663551401869, - "loss": 6.8395, - "step": 828 - }, - { - "epoch": 0.46, - "learning_rate": 0.00022453271028037382, - "loss": 7.0266, - "step": 829 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022429906542056072, - "loss": 6.9933, - "step": 830 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022406542056074765, - "loss": 7.2324, - "step": 831 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022383177570093458, - "loss": 7.1431, - "step": 832 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002235981308411215, - "loss": 7.0766, - "step": 833 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022336448598130838, - "loss": 6.7211, - "step": 834 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002231308411214953, - "loss": 6.8478, - "step": 835 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022289719626168223, - "loss": 7.0115, - "step": 836 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022266355140186916, - "loss": 7.0017, - "step": 837 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022242990654205606, - "loss": 6.8967, - "step": 838 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022219626168224296, - "loss": 6.8296, - "step": 839 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022196261682242988, - "loss": 6.9712, - "step": 840 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002217289719626168, - "loss": 7.1148, - "step": 841 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002214953271028037, - "loss": 7.0691, - "step": 842 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022126168224299064, - "loss": 6.9233, - "step": 843 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022102803738317757, - "loss": 7.2002, - "step": 844 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002207943925233645, - "loss": 6.7511, - "step": 845 - }, - { - "epoch": 0.47, - "learning_rate": 0.00022056074766355137, - "loss": 6.7363, - "step": 846 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002203271028037383, - "loss": 6.9408, - "step": 847 - }, - { - "epoch": 0.48, - "learning_rate": 0.00022009345794392522, - "loss": 6.9343, - "step": 848 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021985981308411215, - "loss": 6.6526, - "step": 849 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021962616822429905, - "loss": 6.603, - "step": 850 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021939252336448595, - "loss": 7.0871, - "step": 851 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021915887850467288, - "loss": 7.0657, - "step": 852 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002189252336448598, - "loss": 6.977, - "step": 853 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002186915887850467, - "loss": 7.0226, - "step": 854 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021845794392523363, - "loss": 7.0437, - "step": 855 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021822429906542056, - "loss": 7.0118, - "step": 856 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021799065420560748, - "loss": 7.0241, - "step": 857 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021775700934579436, - "loss": 7.1028, - "step": 858 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021752336448598128, - "loss": 7.1038, - "step": 859 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002172897196261682, - "loss": 7.0728, - "step": 860 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021705607476635514, - "loss": 6.9278, - "step": 861 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021682242990654204, - "loss": 6.8426, - "step": 862 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021658878504672894, - "loss": 6.8437, - "step": 863 - }, - { - "epoch": 0.48, - "learning_rate": 0.00021635514018691587, - "loss": 6.7931, - "step": 864 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002161214953271028, - "loss": 6.8879, - "step": 865 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002158878504672897, - "loss": 7.0654, - "step": 866 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021565420560747662, - "loss": 6.8924, - "step": 867 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021542056074766355, - "loss": 6.9473, - "step": 868 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021518691588785047, - "loss": 7.0184, - "step": 869 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021495327102803735, - "loss": 6.9939, - "step": 870 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021471962616822427, - "loss": 6.9466, - "step": 871 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002144859813084112, - "loss": 6.9636, - "step": 872 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021425233644859813, - "loss": 7.0039, - "step": 873 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021401869158878503, - "loss": 7.0092, - "step": 874 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021378504672897193, - "loss": 7.3061, - "step": 875 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021355140186915886, - "loss": 7.0345, - "step": 876 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021331775700934578, - "loss": 7.189, - "step": 877 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021308411214953268, - "loss": 6.9818, - "step": 878 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002128504672897196, - "loss": 6.8523, - "step": 879 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021261682242990654, - "loss": 7.124, - "step": 880 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021238317757009347, - "loss": 7.0963, - "step": 881 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021214953271028034, - "loss": 6.784, - "step": 882 - }, - { - "epoch": 0.49, - "learning_rate": 0.00021191588785046727, - "loss": 7.1054, - "step": 883 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002116822429906542, - "loss": 6.9615, - "step": 884 - }, - { - "epoch": 0.5, - "learning_rate": 0.00021144859813084112, - "loss": 7.0952, - "step": 885 - }, - { - "epoch": 0.5, - "learning_rate": 0.00021121495327102802, - "loss": 7.2212, - "step": 886 - }, - { - "epoch": 0.5, - "learning_rate": 0.00021098130841121492, - "loss": 7.0847, - "step": 887 - }, - { - "epoch": 0.5, - "learning_rate": 0.00021074766355140185, - "loss": 6.9394, - "step": 888 - }, - { - "epoch": 0.5, - "learning_rate": 0.00021051401869158877, - "loss": 6.9019, - "step": 889 - }, - { - "epoch": 0.5, - "learning_rate": 0.00021028037383177567, - "loss": 6.9829, - "step": 890 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002100467289719626, - "loss": 7.0558, - "step": 891 - }, - { - "epoch": 0.5, - "learning_rate": 0.00020981308411214953, - "loss": 7.0388, - "step": 892 - }, - { - "epoch": 0.5, - "learning_rate": 0.00020957943925233646, - "loss": 6.9971, - "step": 893 - }, - { - "epoch": 0.5, - "learning_rate": 0.00020934579439252333, - "loss": 6.9198, - "step": 894 - }, - { - "epoch": 0.5, - "learning_rate": 0.00020911214953271026, - "loss": 6.7717, - "step": 895 - }, - { - "epoch": 0.5, - "learning_rate": 0.00020887850467289718, - "loss": 6.8944, - "step": 896 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002086448598130841, - "loss": 6.6609, - "step": 897 - }, - { - "epoch": 0.5, - "learning_rate": 0.000208411214953271, - "loss": 7.0289, - "step": 898 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002081775700934579, - "loss": 6.5158, - "step": 899 - }, - { - "epoch": 0.5, - "learning_rate": 0.00020794392523364484, - "loss": 6.6221, - "step": 900 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020771028037383176, - "loss": 6.9249, - "step": 901 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020747663551401867, - "loss": 7.1439, - "step": 902 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002072429906542056, - "loss": 7.2176, - "step": 903 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020700934579439252, - "loss": 6.9676, - "step": 904 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020677570093457945, - "loss": 7.2003, - "step": 905 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020654205607476632, - "loss": 7.2292, - "step": 906 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020630841121495325, - "loss": 7.0781, - "step": 907 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020607476635514017, - "loss": 6.9036, - "step": 908 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002058411214953271, - "loss": 7.0435, - "step": 909 - }, - { - "epoch": 0.51, - "learning_rate": 0.000205607476635514, - "loss": 7.0006, - "step": 910 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002053738317757009, - "loss": 7.1955, - "step": 911 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020514018691588783, - "loss": 6.9229, - "step": 912 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020490654205607476, - "loss": 6.8387, - "step": 913 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020467289719626166, - "loss": 7.0758, - "step": 914 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020443925233644858, - "loss": 6.9897, - "step": 915 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002042056074766355, - "loss": 6.9014, - "step": 916 - }, - { - "epoch": 0.51, - "learning_rate": 0.00020397196261682244, - "loss": 6.9445, - "step": 917 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002037383177570093, - "loss": 7.0651, - "step": 918 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020350467289719624, - "loss": 6.9525, - "step": 919 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020327102803738316, - "loss": 6.9574, - "step": 920 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002030373831775701, - "loss": 7.0692, - "step": 921 - }, - { - "epoch": 0.52, - "learning_rate": 0.000202803738317757, - "loss": 6.8209, - "step": 922 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002025700934579439, - "loss": 6.8429, - "step": 923 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020233644859813082, - "loss": 7.1023, - "step": 924 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020210280373831775, - "loss": 7.0788, - "step": 925 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020186915887850465, - "loss": 7.0857, - "step": 926 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020163551401869157, - "loss": 6.9488, - "step": 927 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002014018691588785, - "loss": 6.8913, - "step": 928 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020116822429906543, - "loss": 6.9058, - "step": 929 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002009345794392523, - "loss": 7.0986, - "step": 930 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020070093457943923, - "loss": 7.0135, - "step": 931 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020046728971962616, - "loss": 7.0849, - "step": 932 - }, - { - "epoch": 0.52, - "learning_rate": 0.00020023364485981308, - "loss": 6.846, - "step": 933 - }, - { - "epoch": 0.52, - "learning_rate": 0.00019999999999999998, - "loss": 7.0911, - "step": 934 - }, - { - "epoch": 0.52, - "learning_rate": 0.00019976635514018688, - "loss": 6.9965, - "step": 935 - }, - { - "epoch": 0.52, - "learning_rate": 0.0001995327102803738, - "loss": 7.0619, - "step": 936 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019929906542056074, - "loss": 7.0449, - "step": 937 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019906542056074764, - "loss": 6.9337, - "step": 938 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019883177570093456, - "loss": 6.7139, - "step": 939 - }, - { - "epoch": 0.53, - "learning_rate": 0.0001985981308411215, - "loss": 7.1301, - "step": 940 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019836448598130842, - "loss": 7.1145, - "step": 941 - }, - { - "epoch": 0.53, - "learning_rate": 0.0001981308411214953, - "loss": 6.9858, - "step": 942 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019789719626168222, - "loss": 7.1614, - "step": 943 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019766355140186915, - "loss": 6.8699, - "step": 944 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019742990654205607, - "loss": 7.1797, - "step": 945 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019719626168224297, - "loss": 6.9562, - "step": 946 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019696261682242987, - "loss": 6.5535, - "step": 947 - }, - { - "epoch": 0.53, - "learning_rate": 0.0001967289719626168, - "loss": 6.9624, - "step": 948 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019649532710280373, - "loss": 7.1892, - "step": 949 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019626168224299063, - "loss": 6.4138, - "step": 950 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019602803738317755, - "loss": 7.1464, - "step": 951 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019579439252336448, - "loss": 7.0997, - "step": 952 - }, - { - "epoch": 0.53, - "learning_rate": 0.0001955607476635514, - "loss": 7.0904, - "step": 953 - }, - { - "epoch": 0.53, - "learning_rate": 0.00019532710280373828, - "loss": 6.9221, - "step": 954 - }, - { - "epoch": 0.54, - "learning_rate": 0.0001950934579439252, - "loss": 6.9857, - "step": 955 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019485981308411214, - "loss": 7.0046, - "step": 956 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019462616822429906, - "loss": 7.0042, - "step": 957 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019439252336448596, - "loss": 6.9648, - "step": 958 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019415887850467286, - "loss": 7.15, - "step": 959 - }, - { - "epoch": 0.54, - "learning_rate": 0.0001939252336448598, - "loss": 6.8918, - "step": 960 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019369158878504672, - "loss": 6.9253, - "step": 961 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019345794392523362, - "loss": 7.052, - "step": 962 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019322429906542055, - "loss": 6.9016, - "step": 963 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019299065420560747, - "loss": 6.9363, - "step": 964 - }, - { - "epoch": 0.54, - "learning_rate": 0.0001927570093457944, - "loss": 6.9766, - "step": 965 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019252336448598127, - "loss": 6.876, - "step": 966 - }, - { - "epoch": 0.54, - "learning_rate": 0.0001922897196261682, - "loss": 7.0262, - "step": 967 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019205607476635513, - "loss": 7.0085, - "step": 968 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019182242990654205, - "loss": 7.1267, - "step": 969 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019158878504672895, - "loss": 6.8912, - "step": 970 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019135514018691585, - "loss": 7.0677, - "step": 971 - }, - { - "epoch": 0.54, - "learning_rate": 0.00019112149532710278, - "loss": 6.9549, - "step": 972 - }, - { - "epoch": 0.55, - "learning_rate": 0.0001908878504672897, - "loss": 7.0509, - "step": 973 - }, - { - "epoch": 0.55, - "learning_rate": 0.0001906542056074766, - "loss": 7.1364, - "step": 974 - }, - { - "epoch": 0.55, - "learning_rate": 0.00019042056074766354, - "loss": 6.9843, - "step": 975 - }, - { - "epoch": 0.55, - "learning_rate": 0.00019018691588785046, - "loss": 6.9203, - "step": 976 - }, - { - "epoch": 0.55, - "learning_rate": 0.0001899532710280374, - "loss": 7.2871, - "step": 977 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018971962616822426, - "loss": 6.9906, - "step": 978 - }, - { - "epoch": 0.55, - "learning_rate": 0.0001894859813084112, - "loss": 6.8969, - "step": 979 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018925233644859812, - "loss": 6.8843, - "step": 980 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018901869158878504, - "loss": 6.7843, - "step": 981 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018878504672897195, - "loss": 6.8731, - "step": 982 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018855140186915885, - "loss": 7.1465, - "step": 983 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018831775700934577, - "loss": 7.0724, - "step": 984 - }, - { - "epoch": 0.55, - "learning_rate": 0.0001880841121495327, - "loss": 6.8224, - "step": 985 - }, - { - "epoch": 0.55, - "learning_rate": 0.0001878504672897196, - "loss": 7.098, - "step": 986 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018761682242990653, - "loss": 7.0772, - "step": 987 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018738317757009345, - "loss": 6.9383, - "step": 988 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018714953271028038, - "loss": 7.2851, - "step": 989 - }, - { - "epoch": 0.55, - "learning_rate": 0.00018691588785046725, - "loss": 6.9013, - "step": 990 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018668224299065418, - "loss": 6.7685, - "step": 991 - }, - { - "epoch": 0.56, - "learning_rate": 0.0001864485981308411, - "loss": 7.1633, - "step": 992 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018621495327102804, - "loss": 6.8629, - "step": 993 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018598130841121494, - "loss": 6.9612, - "step": 994 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018574766355140184, - "loss": 6.8707, - "step": 995 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018551401869158876, - "loss": 6.5812, - "step": 996 - }, - { - "epoch": 0.56, - "learning_rate": 0.0001852803738317757, - "loss": 6.7197, - "step": 997 - }, - { - "epoch": 0.56, - "learning_rate": 0.0001850467289719626, - "loss": 7.4113, - "step": 998 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018481308411214952, - "loss": 6.9108, - "step": 999 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018457943925233644, - "loss": 6.74, - "step": 1000 - }, - { - "epoch": 0.56, - "eval_loss": 6.988526821136475, - "eval_runtime": 890.4048, - "eval_samples_per_second": 2.967, - "eval_steps_per_second": 0.372, - "eval_wer": 2.0097029585094353, - "step": 1000 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018434579439252337, - "loss": 7.2423, - "step": 1001 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018411214953271024, - "loss": 7.0703, - "step": 1002 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018387850467289717, - "loss": 7.0276, - "step": 1003 - }, - { - "epoch": 0.56, - "learning_rate": 0.0001836448598130841, - "loss": 7.1216, - "step": 1004 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018341121495327103, - "loss": 7.2049, - "step": 1005 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018317757009345793, - "loss": 7.1602, - "step": 1006 - }, - { - "epoch": 0.56, - "learning_rate": 0.00018294392523364483, - "loss": 7.0232, - "step": 1007 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018271028037383175, - "loss": 6.996, - "step": 1008 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018247663551401868, - "loss": 6.9787, - "step": 1009 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018224299065420558, - "loss": 6.9444, - "step": 1010 - }, - { - "epoch": 0.57, - "learning_rate": 0.0001820093457943925, - "loss": 6.9225, - "step": 1011 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018177570093457944, - "loss": 7.1478, - "step": 1012 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018154205607476636, - "loss": 6.9185, - "step": 1013 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018130841121495324, - "loss": 7.1473, - "step": 1014 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018107476635514016, - "loss": 7.1079, - "step": 1015 - }, - { - "epoch": 0.57, - "learning_rate": 0.0001808411214953271, - "loss": 6.8766, - "step": 1016 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018060747663551402, - "loss": 6.929, - "step": 1017 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018037383177570092, - "loss": 6.9298, - "step": 1018 - }, - { - "epoch": 0.57, - "learning_rate": 0.00018014018691588782, - "loss": 7.2166, - "step": 1019 - }, - { - "epoch": 0.57, - "learning_rate": 0.00017990654205607474, - "loss": 7.1188, - "step": 1020 - }, - { - "epoch": 0.57, - "learning_rate": 0.00017967289719626167, - "loss": 7.027, - "step": 1021 - }, - { - "epoch": 0.57, - "learning_rate": 0.00017943925233644857, - "loss": 6.9638, - "step": 1022 - }, - { - "epoch": 0.57, - "learning_rate": 0.0001792056074766355, - "loss": 6.8388, - "step": 1023 - }, - { - "epoch": 0.57, - "learning_rate": 0.00017897196261682243, - "loss": 6.9376, - "step": 1024 - }, - { - "epoch": 0.57, - "learning_rate": 0.00017873831775700935, - "loss": 7.1039, - "step": 1025 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017850467289719623, - "loss": 6.9258, - "step": 1026 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017827102803738315, - "loss": 7.1009, - "step": 1027 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017803738317757008, - "loss": 7.2111, - "step": 1028 - }, - { - "epoch": 0.58, - "learning_rate": 0.000177803738317757, - "loss": 6.9808, - "step": 1029 - }, - { - "epoch": 0.58, - "learning_rate": 0.0001775700934579439, - "loss": 7.07, - "step": 1030 - }, - { - "epoch": 0.58, - "learning_rate": 0.0001773364485981308, - "loss": 7.0645, - "step": 1031 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017710280373831773, - "loss": 7.0076, - "step": 1032 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017686915887850466, - "loss": 6.9341, - "step": 1033 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017663551401869156, - "loss": 6.7951, - "step": 1034 - }, - { - "epoch": 0.58, - "learning_rate": 0.0001764018691588785, - "loss": 6.9385, - "step": 1035 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017616822429906542, - "loss": 7.0999, - "step": 1036 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017593457943925234, - "loss": 6.7708, - "step": 1037 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017570093457943922, - "loss": 7.0748, - "step": 1038 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017546728971962614, - "loss": 7.0044, - "step": 1039 - }, - { - "epoch": 0.58, - "learning_rate": 0.00017523364485981307, - "loss": 7.0355, - "step": 1040 - }, - { - "epoch": 0.58, - "learning_rate": 0.000175, - "loss": 7.1551, - "step": 1041 - }, - { - "epoch": 0.58, - "learning_rate": 0.0001747663551401869, - "loss": 6.8403, - "step": 1042 - }, - { - "epoch": 0.58, - "learning_rate": 0.0001745327102803738, - "loss": 6.8854, - "step": 1043 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017429906542056073, - "loss": 6.8326, - "step": 1044 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017406542056074765, - "loss": 7.0009, - "step": 1045 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017383177570093455, - "loss": 6.9063, - "step": 1046 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017359813084112148, - "loss": 6.9923, - "step": 1047 - }, - { - "epoch": 0.59, - "learning_rate": 0.0001733644859813084, - "loss": 6.6022, - "step": 1048 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017313084112149533, - "loss": 6.5359, - "step": 1049 - }, - { - "epoch": 0.59, - "learning_rate": 0.0001728971962616822, - "loss": 7.0852, - "step": 1050 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017266355140186913, - "loss": 7.37, - "step": 1051 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017242990654205606, - "loss": 6.9386, - "step": 1052 - }, - { - "epoch": 0.59, - "learning_rate": 0.000172196261682243, - "loss": 7.014, - "step": 1053 - }, - { - "epoch": 0.59, - "learning_rate": 0.0001719626168224299, - "loss": 7.0895, - "step": 1054 - }, - { - "epoch": 0.59, - "learning_rate": 0.0001717289719626168, - "loss": 7.1071, - "step": 1055 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017149532710280372, - "loss": 7.0649, - "step": 1056 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017126168224299064, - "loss": 7.1938, - "step": 1057 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017102803738317754, - "loss": 7.0419, - "step": 1058 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017079439252336447, - "loss": 7.0406, - "step": 1059 - }, - { - "epoch": 0.59, - "learning_rate": 0.0001705607476635514, - "loss": 7.0155, - "step": 1060 - }, - { - "epoch": 0.59, - "learning_rate": 0.00017032710280373833, - "loss": 7.0306, - "step": 1061 - }, - { - "epoch": 0.6, - "learning_rate": 0.0001700934579439252, - "loss": 7.0, - "step": 1062 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016985981308411213, - "loss": 7.2515, - "step": 1063 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016962616822429905, - "loss": 6.923, - "step": 1064 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016939252336448598, - "loss": 7.1924, - "step": 1065 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016915887850467288, - "loss": 6.7962, - "step": 1066 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016892523364485978, - "loss": 7.2572, - "step": 1067 - }, - { - "epoch": 0.6, - "learning_rate": 0.0001686915887850467, - "loss": 6.9445, - "step": 1068 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016845794392523363, - "loss": 6.9963, - "step": 1069 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016822429906542053, - "loss": 7.0373, - "step": 1070 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016799065420560746, - "loss": 6.9157, - "step": 1071 - }, - { - "epoch": 0.6, - "learning_rate": 0.0001677570093457944, - "loss": 6.7637, - "step": 1072 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016752336448598132, - "loss": 7.0241, - "step": 1073 - }, - { - "epoch": 0.6, - "learning_rate": 0.0001672897196261682, - "loss": 7.0438, - "step": 1074 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016705607476635512, - "loss": 6.8616, - "step": 1075 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016682242990654204, - "loss": 7.0244, - "step": 1076 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016658878504672897, - "loss": 6.8898, - "step": 1077 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016635514018691587, - "loss": 6.955, - "step": 1078 - }, - { - "epoch": 0.6, - "learning_rate": 0.00016612149532710277, - "loss": 7.1004, - "step": 1079 - }, - { - "epoch": 0.61, - "learning_rate": 0.0001658878504672897, - "loss": 6.9628, - "step": 1080 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016565420560747662, - "loss": 6.9673, - "step": 1081 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016542056074766352, - "loss": 6.9567, - "step": 1082 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016518691588785045, - "loss": 6.9401, - "step": 1083 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016495327102803738, - "loss": 7.1168, - "step": 1084 - }, - { - "epoch": 0.61, - "learning_rate": 0.0001647196261682243, - "loss": 6.9336, - "step": 1085 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016448598130841118, - "loss": 7.1096, - "step": 1086 - }, - { - "epoch": 0.61, - "learning_rate": 0.0001642523364485981, - "loss": 7.2345, - "step": 1087 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016401869158878503, - "loss": 7.1436, - "step": 1088 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016378504672897196, - "loss": 7.1528, - "step": 1089 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016355140186915886, - "loss": 6.9956, - "step": 1090 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016331775700934576, - "loss": 7.1095, - "step": 1091 - }, - { - "epoch": 0.61, - "learning_rate": 0.0001630841121495327, - "loss": 7.0951, - "step": 1092 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016285046728971962, - "loss": 6.9334, - "step": 1093 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016261682242990652, - "loss": 6.766, - "step": 1094 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016238317757009344, - "loss": 6.9135, - "step": 1095 - }, - { - "epoch": 0.61, - "learning_rate": 0.00016214953271028037, - "loss": 6.7706, - "step": 1096 - }, - { - "epoch": 0.61, - "learning_rate": 0.0001619158878504673, - "loss": 7.025, - "step": 1097 - }, - { - "epoch": 0.62, - "learning_rate": 0.00016168224299065417, - "loss": 7.1551, - "step": 1098 - }, - { - "epoch": 0.62, - "learning_rate": 0.0001614485981308411, - "loss": 6.86, - "step": 1099 - }, - { - "epoch": 0.62, - "learning_rate": 0.00016121495327102802, - "loss": 6.4933, - "step": 1100 - }, - { - "epoch": 0.62, - "learning_rate": 0.00016098130841121495, - "loss": 6.9036, - "step": 1101 - }, - { - "epoch": 0.62, - "learning_rate": 0.00016074766355140185, - "loss": 7.1383, - "step": 1102 - }, - { - "epoch": 0.62, - "learning_rate": 0.00016051401869158875, - "loss": 7.1216, - "step": 1103 - }, - { - "epoch": 0.62, - "learning_rate": 0.00016028037383177568, - "loss": 6.9468, - "step": 1104 - }, - { - "epoch": 0.62, - "learning_rate": 0.0001600467289719626, - "loss": 6.9912, - "step": 1105 - }, - { - "epoch": 0.62, - "learning_rate": 0.0001598130841121495, - "loss": 6.865, - "step": 1106 - }, - { - "epoch": 0.62, - "learning_rate": 0.00015957943925233643, - "loss": 6.9954, - "step": 1107 - }, - { - "epoch": 0.62, - "learning_rate": 0.00015934579439252336, - "loss": 6.958, - "step": 1108 - }, - { - "epoch": 0.62, - "learning_rate": 0.0001591121495327103, - "loss": 7.0301, - "step": 1109 - }, - { - "epoch": 0.62, - "learning_rate": 0.00015887850467289716, - "loss": 7.0107, - "step": 1110 - }, - { - "epoch": 0.62, - "learning_rate": 0.0001586448598130841, - "loss": 6.8857, - "step": 1111 - }, - { - "epoch": 0.62, - "learning_rate": 0.00015841121495327101, - "loss": 6.9603, - "step": 1112 - }, - { - "epoch": 0.62, - "learning_rate": 0.00015817757009345794, - "loss": 6.9502, - "step": 1113 - }, - { - "epoch": 0.62, - "learning_rate": 0.00015794392523364484, - "loss": 7.1344, - "step": 1114 - }, - { - "epoch": 0.62, - "learning_rate": 0.00015771028037383174, - "loss": 7.0907, - "step": 1115 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015747663551401867, - "loss": 6.7842, - "step": 1116 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001572429906542056, - "loss": 6.919, - "step": 1117 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001570093457943925, - "loss": 7.0003, - "step": 1118 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015677570093457942, - "loss": 7.0514, - "step": 1119 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015654205607476635, - "loss": 6.9718, - "step": 1120 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015630841121495328, - "loss": 6.9891, - "step": 1121 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015607476635514015, - "loss": 6.9222, - "step": 1122 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015584112149532708, - "loss": 6.8753, - "step": 1123 - }, - { - "epoch": 0.63, - "learning_rate": 0.000155607476635514, - "loss": 6.941, - "step": 1124 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015537383177570093, - "loss": 6.9554, - "step": 1125 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015514018691588783, - "loss": 6.7981, - "step": 1126 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015490654205607473, - "loss": 6.9189, - "step": 1127 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015467289719626166, - "loss": 7.2442, - "step": 1128 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001544392523364486, - "loss": 7.0163, - "step": 1129 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001542056074766355, - "loss": 7.1331, - "step": 1130 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015397196261682241, - "loss": 7.128, - "step": 1131 - }, - { - "epoch": 0.63, - "learning_rate": 0.00015373831775700934, - "loss": 7.1329, - "step": 1132 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015350467289719627, - "loss": 6.8574, - "step": 1133 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015327102803738314, - "loss": 6.9467, - "step": 1134 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015303738317757007, - "loss": 7.0593, - "step": 1135 - }, - { - "epoch": 0.64, - "learning_rate": 0.000152803738317757, - "loss": 6.9699, - "step": 1136 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015257009345794392, - "loss": 6.9434, - "step": 1137 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015233644859813082, - "loss": 7.056, - "step": 1138 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015210280373831772, - "loss": 6.827, - "step": 1139 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015186915887850465, - "loss": 7.016, - "step": 1140 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015163551401869158, - "loss": 6.9792, - "step": 1141 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015140186915887848, - "loss": 7.2732, - "step": 1142 - }, - { - "epoch": 0.64, - "learning_rate": 0.0001511682242990654, - "loss": 7.0662, - "step": 1143 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015093457943925233, - "loss": 7.0129, - "step": 1144 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015070093457943926, - "loss": 6.9208, - "step": 1145 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015046728971962613, - "loss": 6.9565, - "step": 1146 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015023364485981306, - "loss": 6.9095, - "step": 1147 - }, - { - "epoch": 0.64, - "learning_rate": 0.00015, - "loss": 6.8028, - "step": 1148 - }, - { - "epoch": 0.64, - "learning_rate": 0.00014976635514018691, - "loss": 6.7528, - "step": 1149 - }, - { - "epoch": 0.64, - "learning_rate": 0.00014953271028037381, - "loss": 6.6378, - "step": 1150 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014929906542056074, - "loss": 7.0801, - "step": 1151 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014906542056074764, - "loss": 7.175, - "step": 1152 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014883177570093457, - "loss": 7.0276, - "step": 1153 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014859813084112147, - "loss": 7.1764, - "step": 1154 - }, - { - "epoch": 0.65, - "learning_rate": 0.0001483644859813084, - "loss": 7.1214, - "step": 1155 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014813084112149532, - "loss": 7.1066, - "step": 1156 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014789719626168225, - "loss": 6.9861, - "step": 1157 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014766355140186915, - "loss": 7.1374, - "step": 1158 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014742990654205608, - "loss": 6.8755, - "step": 1159 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014719626168224298, - "loss": 6.9121, - "step": 1160 - }, - { - "epoch": 0.65, - "learning_rate": 0.0001469626168224299, - "loss": 7.1185, - "step": 1161 - }, - { - "epoch": 0.65, - "learning_rate": 0.0001467289719626168, - "loss": 6.8981, - "step": 1162 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014649532710280373, - "loss": 6.9337, - "step": 1163 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014626168224299063, - "loss": 6.886, - "step": 1164 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014602803738317756, - "loss": 7.0434, - "step": 1165 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014579439252336446, - "loss": 7.0345, - "step": 1166 - }, - { - "epoch": 0.65, - "learning_rate": 0.0001455607476635514, - "loss": 7.1817, - "step": 1167 - }, - { - "epoch": 0.65, - "learning_rate": 0.00014532710280373831, - "loss": 6.6841, - "step": 1168 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014509345794392524, - "loss": 6.9224, - "step": 1169 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014485981308411214, - "loss": 7.0135, - "step": 1170 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014462616822429907, - "loss": 7.0044, - "step": 1171 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014439252336448597, - "loss": 6.9385, - "step": 1172 - }, - { - "epoch": 0.66, - "learning_rate": 0.0001441588785046729, - "loss": 7.2891, - "step": 1173 - }, - { - "epoch": 0.66, - "learning_rate": 0.0001439252336448598, - "loss": 6.9103, - "step": 1174 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014369158878504672, - "loss": 6.866, - "step": 1175 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014345794392523362, - "loss": 6.9995, - "step": 1176 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014322429906542055, - "loss": 6.9609, - "step": 1177 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014299065420560745, - "loss": 6.906, - "step": 1178 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014275700934579438, - "loss": 6.8465, - "step": 1179 - }, - { - "epoch": 0.66, - "learning_rate": 0.0001425233644859813, - "loss": 7.1573, - "step": 1180 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014228971962616823, - "loss": 6.9587, - "step": 1181 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014205607476635513, - "loss": 6.9477, - "step": 1182 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014182242990654206, - "loss": 6.9239, - "step": 1183 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014158878504672896, - "loss": 6.9525, - "step": 1184 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014135514018691589, - "loss": 7.3453, - "step": 1185 - }, - { - "epoch": 0.66, - "learning_rate": 0.00014112149532710279, - "loss": 6.8569, - "step": 1186 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001408878504672897, - "loss": 7.0619, - "step": 1187 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001406542056074766, - "loss": 7.0111, - "step": 1188 - }, - { - "epoch": 0.67, - "learning_rate": 0.00014042056074766354, - "loss": 6.9286, - "step": 1189 - }, - { - "epoch": 0.67, - "learning_rate": 0.00014018691588785044, - "loss": 6.9576, - "step": 1190 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013995327102803737, - "loss": 6.9818, - "step": 1191 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001397196261682243, - "loss": 6.78, - "step": 1192 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013948598130841122, - "loss": 6.9119, - "step": 1193 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013925233644859812, - "loss": 6.9228, - "step": 1194 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013901869158878505, - "loss": 6.7023, - "step": 1195 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013878504672897195, - "loss": 6.7319, - "step": 1196 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013855140186915888, - "loss": 6.6737, - "step": 1197 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013831775700934578, - "loss": 6.865, - "step": 1198 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001380841121495327, - "loss": 6.4257, - "step": 1199 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001378504672897196, - "loss": 6.2315, - "step": 1200 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013761682242990653, - "loss": 7.0852, - "step": 1201 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013738317757009343, - "loss": 7.0461, - "step": 1202 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013714953271028036, - "loss": 7.013, - "step": 1203 - }, - { - "epoch": 0.67, - "learning_rate": 0.00013691588785046729, - "loss": 6.8305, - "step": 1204 - }, - { - "epoch": 0.68, - "learning_rate": 0.0001366822429906542, - "loss": 6.936, - "step": 1205 - }, - { - "epoch": 0.68, - "learning_rate": 0.0001364485981308411, - "loss": 6.9314, - "step": 1206 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013621495327102804, - "loss": 7.023, - "step": 1207 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013598130841121494, - "loss": 7.0537, - "step": 1208 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013574766355140187, - "loss": 6.9336, - "step": 1209 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013551401869158877, - "loss": 6.8911, - "step": 1210 - }, - { - "epoch": 0.68, - "learning_rate": 0.0001352803738317757, - "loss": 7.0679, - "step": 1211 - }, - { - "epoch": 0.68, - "learning_rate": 0.0001350467289719626, - "loss": 6.8457, - "step": 1212 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013481308411214952, - "loss": 6.9401, - "step": 1213 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013457943925233642, - "loss": 7.1918, - "step": 1214 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013434579439252335, - "loss": 6.9589, - "step": 1215 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013411214953271028, - "loss": 6.9271, - "step": 1216 - }, - { - "epoch": 0.68, - "learning_rate": 0.0001338785046728972, - "loss": 6.8912, - "step": 1217 - }, - { - "epoch": 0.68, - "learning_rate": 0.0001336448598130841, - "loss": 6.7201, - "step": 1218 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013341121495327103, - "loss": 6.9223, - "step": 1219 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013317757009345793, - "loss": 7.1514, - "step": 1220 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013294392523364486, - "loss": 7.0156, - "step": 1221 - }, - { - "epoch": 0.68, - "learning_rate": 0.00013271028037383176, - "loss": 7.1662, - "step": 1222 - }, - { - "epoch": 0.69, - "learning_rate": 0.00013247663551401869, - "loss": 7.1794, - "step": 1223 - }, - { - "epoch": 0.69, - "learning_rate": 0.00013224299065420559, - "loss": 6.8795, - "step": 1224 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001320093457943925, - "loss": 6.8894, - "step": 1225 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001317757009345794, - "loss": 6.9371, - "step": 1226 - }, - { - "epoch": 0.69, - "learning_rate": 0.00013154205607476634, - "loss": 6.9505, - "step": 1227 - }, - { - "epoch": 0.69, - "learning_rate": 0.00013130841121495327, - "loss": 7.1028, - "step": 1228 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001310747663551402, - "loss": 7.042, - "step": 1229 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001308411214953271, - "loss": 7.2691, - "step": 1230 - }, - { - "epoch": 0.69, - "learning_rate": 0.00013060747663551402, - "loss": 6.9947, - "step": 1231 - }, - { - "epoch": 0.69, - "learning_rate": 0.00013037383177570092, - "loss": 6.7231, - "step": 1232 - }, - { - "epoch": 0.69, - "learning_rate": 0.00013014018691588785, - "loss": 7.0804, - "step": 1233 - }, - { - "epoch": 0.69, - "learning_rate": 0.00012990654205607475, - "loss": 6.9197, - "step": 1234 - }, - { - "epoch": 0.69, - "learning_rate": 0.00012967289719626168, - "loss": 7.1474, - "step": 1235 - }, - { - "epoch": 0.69, - "learning_rate": 0.00012943925233644858, - "loss": 7.0081, - "step": 1236 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001292056074766355, - "loss": 7.034, - "step": 1237 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001289719626168224, - "loss": 6.8294, - "step": 1238 - }, - { - "epoch": 0.69, - "learning_rate": 0.00012873831775700933, - "loss": 7.1582, - "step": 1239 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012850467289719626, - "loss": 7.1427, - "step": 1240 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012827102803738318, - "loss": 6.9733, - "step": 1241 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012803738317757008, - "loss": 7.1433, - "step": 1242 - }, - { - "epoch": 0.7, - "learning_rate": 0.000127803738317757, - "loss": 7.008, - "step": 1243 - }, - { - "epoch": 0.7, - "learning_rate": 0.0001275700934579439, - "loss": 6.7795, - "step": 1244 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012733644859813084, - "loss": 6.841, - "step": 1245 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012710280373831774, - "loss": 7.1104, - "step": 1246 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012686915887850467, - "loss": 6.9427, - "step": 1247 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012663551401869157, - "loss": 6.8738, - "step": 1248 - }, - { - "epoch": 0.7, - "learning_rate": 0.0001264018691588785, - "loss": 6.7209, - "step": 1249 - }, - { - "epoch": 0.7, - "learning_rate": 0.0001261682242990654, - "loss": 6.9355, - "step": 1250 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012593457943925232, - "loss": 7.1563, - "step": 1251 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012570093457943925, - "loss": 6.9835, - "step": 1252 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012546728971962618, - "loss": 7.0192, - "step": 1253 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012523364485981308, - "loss": 7.0741, - "step": 1254 - }, - { - "epoch": 0.7, - "learning_rate": 0.000125, - "loss": 7.236, - "step": 1255 - }, - { - "epoch": 0.7, - "learning_rate": 0.0001247663551401869, - "loss": 6.9092, - "step": 1256 - }, - { - "epoch": 0.7, - "learning_rate": 0.00012453271028037383, - "loss": 6.8516, - "step": 1257 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012429906542056073, - "loss": 6.9645, - "step": 1258 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012406542056074766, - "loss": 6.9515, - "step": 1259 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012383177570093456, - "loss": 7.0824, - "step": 1260 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012359813084112148, - "loss": 7.0192, - "step": 1261 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012336448598130838, - "loss": 6.9977, - "step": 1262 - }, - { - "epoch": 0.71, - "learning_rate": 0.0001231308411214953, - "loss": 6.9787, - "step": 1263 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012289719626168224, - "loss": 6.8725, - "step": 1264 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012266355140186917, - "loss": 7.0092, - "step": 1265 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012242990654205607, - "loss": 7.0743, - "step": 1266 - }, - { - "epoch": 0.71, - "learning_rate": 0.000122196261682243, - "loss": 6.9171, - "step": 1267 - }, - { - "epoch": 0.71, - "learning_rate": 0.0001219626168224299, - "loss": 7.009, - "step": 1268 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012172897196261682, - "loss": 6.8043, - "step": 1269 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012149532710280372, - "loss": 6.9266, - "step": 1270 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012126168224299065, - "loss": 6.8346, - "step": 1271 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012102803738317756, - "loss": 6.8009, - "step": 1272 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012079439252336448, - "loss": 7.0138, - "step": 1273 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012056074766355139, - "loss": 6.9029, - "step": 1274 - }, - { - "epoch": 0.71, - "learning_rate": 0.00012032710280373832, - "loss": 7.0285, - "step": 1275 - }, - { - "epoch": 0.72, - "learning_rate": 0.00012009345794392522, - "loss": 7.121, - "step": 1276 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011985981308411214, - "loss": 7.208, - "step": 1277 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011962616822429906, - "loss": 6.9973, - "step": 1278 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011939252336448597, - "loss": 7.0714, - "step": 1279 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011915887850467288, - "loss": 6.9593, - "step": 1280 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011892523364485981, - "loss": 6.8453, - "step": 1281 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011869158878504671, - "loss": 6.9101, - "step": 1282 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011845794392523364, - "loss": 7.0606, - "step": 1283 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011822429906542055, - "loss": 6.8706, - "step": 1284 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011799065420560747, - "loss": 6.9919, - "step": 1285 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011775700934579438, - "loss": 7.1027, - "step": 1286 - }, - { - "epoch": 0.72, - "learning_rate": 0.0001175233644859813, - "loss": 6.9927, - "step": 1287 - }, - { - "epoch": 0.72, - "learning_rate": 0.0001172897196261682, - "loss": 6.8993, - "step": 1288 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011705607476635513, - "loss": 7.0197, - "step": 1289 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011682242990654205, - "loss": 6.8385, - "step": 1290 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011658878504672896, - "loss": 6.9411, - "step": 1291 - }, - { - "epoch": 0.72, - "learning_rate": 0.00011635514018691587, - "loss": 7.0041, - "step": 1292 - }, - { - "epoch": 0.72, - "learning_rate": 0.0001161214953271028, - "loss": 6.9982, - "step": 1293 - }, - { - "epoch": 0.73, - "learning_rate": 0.0001158878504672897, - "loss": 6.8026, - "step": 1294 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011565420560747663, - "loss": 6.8446, - "step": 1295 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011542056074766354, - "loss": 6.7453, - "step": 1296 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011518691588785046, - "loss": 6.6583, - "step": 1297 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011495327102803737, - "loss": 6.5291, - "step": 1298 - }, - { - "epoch": 0.73, - "learning_rate": 0.0001147196261682243, - "loss": 6.9623, - "step": 1299 - }, - { - "epoch": 0.73, - "learning_rate": 0.0001144859813084112, - "loss": 6.3422, - "step": 1300 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011425233644859812, - "loss": 7.0665, - "step": 1301 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011401869158878504, - "loss": 7.2059, - "step": 1302 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011378504672897195, - "loss": 7.0901, - "step": 1303 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011355140186915887, - "loss": 7.065, - "step": 1304 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011331775700934579, - "loss": 7.0094, - "step": 1305 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011308411214953269, - "loss": 7.1075, - "step": 1306 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011285046728971962, - "loss": 6.845, - "step": 1307 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011261682242990653, - "loss": 7.105, - "step": 1308 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011238317757009345, - "loss": 6.7908, - "step": 1309 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011214953271028036, - "loss": 6.9517, - "step": 1310 - }, - { - "epoch": 0.73, - "learning_rate": 0.00011191588785046729, - "loss": 7.0393, - "step": 1311 - }, - { - "epoch": 0.74, - "learning_rate": 0.00011168224299065419, - "loss": 6.9529, - "step": 1312 - }, - { - "epoch": 0.74, - "learning_rate": 0.00011144859813084112, - "loss": 6.9206, - "step": 1313 - }, - { - "epoch": 0.74, - "learning_rate": 0.00011121495327102803, - "loss": 6.8287, - "step": 1314 - }, - { - "epoch": 0.74, - "learning_rate": 0.00011098130841121494, - "loss": 7.2535, - "step": 1315 - }, - { - "epoch": 0.74, - "learning_rate": 0.00011074766355140186, - "loss": 6.9465, - "step": 1316 - }, - { - "epoch": 0.74, - "learning_rate": 0.00011051401869158878, - "loss": 6.7518, - "step": 1317 - }, - { - "epoch": 0.74, - "learning_rate": 0.00011028037383177568, - "loss": 7.0312, - "step": 1318 - }, - { - "epoch": 0.74, - "learning_rate": 0.00011004672897196261, - "loss": 6.868, - "step": 1319 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010981308411214952, - "loss": 6.9033, - "step": 1320 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010957943925233644, - "loss": 6.819, - "step": 1321 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010934579439252335, - "loss": 6.9306, - "step": 1322 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010911214953271028, - "loss": 7.079, - "step": 1323 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010887850467289718, - "loss": 7.0751, - "step": 1324 - }, - { - "epoch": 0.74, - "learning_rate": 0.0001086448598130841, - "loss": 7.0397, - "step": 1325 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010841121495327102, - "loss": 6.9291, - "step": 1326 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010817757009345793, - "loss": 6.7671, - "step": 1327 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010794392523364485, - "loss": 7.0475, - "step": 1328 - }, - { - "epoch": 0.74, - "learning_rate": 0.00010771028037383177, - "loss": 7.0572, - "step": 1329 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010747663551401867, - "loss": 7.0204, - "step": 1330 - }, - { - "epoch": 0.75, - "learning_rate": 0.0001072429906542056, - "loss": 6.8385, - "step": 1331 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010700934579439251, - "loss": 6.9145, - "step": 1332 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010677570093457943, - "loss": 7.1778, - "step": 1333 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010654205607476634, - "loss": 7.0957, - "step": 1334 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010630841121495327, - "loss": 6.7778, - "step": 1335 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010607476635514017, - "loss": 6.9218, - "step": 1336 - }, - { - "epoch": 0.75, - "learning_rate": 0.0001058411214953271, - "loss": 7.0399, - "step": 1337 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010560747663551401, - "loss": 7.0019, - "step": 1338 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010537383177570092, - "loss": 7.0394, - "step": 1339 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010514018691588784, - "loss": 7.1063, - "step": 1340 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010490654205607476, - "loss": 7.1153, - "step": 1341 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010467289719626166, - "loss": 6.9461, - "step": 1342 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010443925233644859, - "loss": 7.0037, - "step": 1343 - }, - { - "epoch": 0.75, - "learning_rate": 0.0001042056074766355, - "loss": 6.8876, - "step": 1344 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010397196261682242, - "loss": 6.7956, - "step": 1345 - }, - { - "epoch": 0.75, - "learning_rate": 0.00010373831775700933, - "loss": 6.8709, - "step": 1346 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010350467289719626, - "loss": 6.9288, - "step": 1347 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010327102803738316, - "loss": 6.5108, - "step": 1348 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010303738317757009, - "loss": 6.7721, - "step": 1349 - }, - { - "epoch": 0.76, - "learning_rate": 0.000102803738317757, - "loss": 6.2563, - "step": 1350 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010257009345794391, - "loss": 7.1633, - "step": 1351 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010233644859813083, - "loss": 7.1344, - "step": 1352 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010210280373831776, - "loss": 7.0726, - "step": 1353 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010186915887850466, - "loss": 6.8862, - "step": 1354 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010163551401869158, - "loss": 7.0405, - "step": 1355 - }, - { - "epoch": 0.76, - "learning_rate": 0.0001014018691588785, - "loss": 7.0804, - "step": 1356 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010116822429906541, - "loss": 6.8674, - "step": 1357 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010093457943925232, - "loss": 7.1048, - "step": 1358 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010070093457943925, - "loss": 7.0674, - "step": 1359 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010046728971962615, - "loss": 7.0487, - "step": 1360 - }, - { - "epoch": 0.76, - "learning_rate": 0.00010023364485981308, - "loss": 6.9764, - "step": 1361 - }, - { - "epoch": 0.76, - "learning_rate": 9.999999999999999e-05, - "loss": 6.9977, - "step": 1362 - }, - { - "epoch": 0.76, - "learning_rate": 9.97663551401869e-05, - "loss": 6.9231, - "step": 1363 - }, - { - "epoch": 0.76, - "learning_rate": 9.953271028037382e-05, - "loss": 7.1122, - "step": 1364 - }, - { - "epoch": 0.77, - "learning_rate": 9.929906542056075e-05, - "loss": 6.842, - "step": 1365 - }, - { - "epoch": 0.77, - "learning_rate": 9.906542056074765e-05, - "loss": 6.9758, - "step": 1366 - }, - { - "epoch": 0.77, - "learning_rate": 9.883177570093457e-05, - "loss": 6.9318, - "step": 1367 - }, - { - "epoch": 0.77, - "learning_rate": 9.859813084112149e-05, - "loss": 6.909, - "step": 1368 - }, - { - "epoch": 0.77, - "learning_rate": 9.83644859813084e-05, - "loss": 6.9903, - "step": 1369 - }, - { - "epoch": 0.77, - "learning_rate": 9.813084112149531e-05, - "loss": 6.9978, - "step": 1370 - }, - { - "epoch": 0.77, - "learning_rate": 9.789719626168224e-05, - "loss": 7.169, - "step": 1371 - }, - { - "epoch": 0.77, - "learning_rate": 9.766355140186914e-05, - "loss": 6.8041, - "step": 1372 - }, - { - "epoch": 0.77, - "learning_rate": 9.742990654205607e-05, - "loss": 6.9999, - "step": 1373 - }, - { - "epoch": 0.77, - "learning_rate": 9.719626168224298e-05, - "loss": 7.1388, - "step": 1374 - }, - { - "epoch": 0.77, - "learning_rate": 9.69626168224299e-05, - "loss": 6.9293, - "step": 1375 - }, - { - "epoch": 0.77, - "learning_rate": 9.672897196261681e-05, - "loss": 6.7652, - "step": 1376 - }, - { - "epoch": 0.77, - "learning_rate": 9.649532710280374e-05, - "loss": 7.0033, - "step": 1377 - }, - { - "epoch": 0.77, - "learning_rate": 9.626168224299064e-05, - "loss": 7.0077, - "step": 1378 - }, - { - "epoch": 0.77, - "learning_rate": 9.602803738317756e-05, - "loss": 6.8923, - "step": 1379 - }, - { - "epoch": 0.77, - "learning_rate": 9.579439252336448e-05, - "loss": 7.0931, - "step": 1380 - }, - { - "epoch": 0.77, - "learning_rate": 9.556074766355139e-05, - "loss": 6.8055, - "step": 1381 - }, - { - "epoch": 0.77, - "learning_rate": 9.53271028037383e-05, - "loss": 7.0076, - "step": 1382 - }, - { - "epoch": 0.78, - "learning_rate": 9.509345794392523e-05, - "loss": 6.953, - "step": 1383 - }, - { - "epoch": 0.78, - "learning_rate": 9.485981308411213e-05, - "loss": 7.0701, - "step": 1384 - }, - { - "epoch": 0.78, - "learning_rate": 9.462616822429906e-05, - "loss": 6.9867, - "step": 1385 - }, - { - "epoch": 0.78, - "learning_rate": 9.439252336448597e-05, - "loss": 7.0534, - "step": 1386 - }, - { - "epoch": 0.78, - "learning_rate": 9.415887850467289e-05, - "loss": 7.0223, - "step": 1387 - }, - { - "epoch": 0.78, - "learning_rate": 9.39252336448598e-05, - "loss": 6.9209, - "step": 1388 - }, - { - "epoch": 0.78, - "learning_rate": 9.369158878504673e-05, - "loss": 7.0354, - "step": 1389 - }, - { - "epoch": 0.78, - "learning_rate": 9.345794392523363e-05, - "loss": 6.9512, - "step": 1390 - }, - { - "epoch": 0.78, - "learning_rate": 9.322429906542055e-05, - "loss": 6.6991, - "step": 1391 - }, - { - "epoch": 0.78, - "learning_rate": 9.299065420560747e-05, - "loss": 7.0627, - "step": 1392 - }, - { - "epoch": 0.78, - "learning_rate": 9.275700934579438e-05, - "loss": 6.8125, - "step": 1393 - }, - { - "epoch": 0.78, - "learning_rate": 9.25233644859813e-05, - "loss": 6.7808, - "step": 1394 - }, - { - "epoch": 0.78, - "learning_rate": 9.228971962616822e-05, - "loss": 7.0037, - "step": 1395 - }, - { - "epoch": 0.78, - "learning_rate": 9.205607476635512e-05, - "loss": 6.8952, - "step": 1396 - }, - { - "epoch": 0.78, - "learning_rate": 9.182242990654205e-05, - "loss": 6.6638, - "step": 1397 - }, - { - "epoch": 0.78, - "learning_rate": 9.158878504672896e-05, - "loss": 6.6949, - "step": 1398 - }, - { - "epoch": 0.78, - "learning_rate": 9.135514018691588e-05, - "loss": 6.6607, - "step": 1399 - }, - { - "epoch": 0.78, - "learning_rate": 9.112149532710279e-05, - "loss": 6.8109, - "step": 1400 - }, - { - "epoch": 0.79, - "learning_rate": 9.088785046728972e-05, - "loss": 7.0519, - "step": 1401 - }, - { - "epoch": 0.79, - "learning_rate": 9.065420560747662e-05, - "loss": 7.0084, - "step": 1402 - }, - { - "epoch": 0.79, - "learning_rate": 9.042056074766354e-05, - "loss": 6.905, - "step": 1403 - }, - { - "epoch": 0.79, - "learning_rate": 9.018691588785046e-05, - "loss": 6.9186, - "step": 1404 - }, - { - "epoch": 0.79, - "learning_rate": 8.995327102803737e-05, - "loss": 6.8921, - "step": 1405 - }, - { - "epoch": 0.79, - "learning_rate": 8.971962616822429e-05, - "loss": 7.2629, - "step": 1406 - }, - { - "epoch": 0.79, - "learning_rate": 8.948598130841121e-05, - "loss": 7.0932, - "step": 1407 - }, - { - "epoch": 0.79, - "learning_rate": 8.925233644859811e-05, - "loss": 6.9424, - "step": 1408 - }, - { - "epoch": 0.79, - "learning_rate": 8.901869158878504e-05, - "loss": 6.9297, - "step": 1409 - }, - { - "epoch": 0.79, - "learning_rate": 8.878504672897195e-05, - "loss": 6.8927, - "step": 1410 - }, - { - "epoch": 0.79, - "learning_rate": 8.855140186915887e-05, - "loss": 6.8569, - "step": 1411 - }, - { - "epoch": 0.79, - "learning_rate": 8.831775700934578e-05, - "loss": 6.8343, - "step": 1412 - }, - { - "epoch": 0.79, - "learning_rate": 8.808411214953271e-05, - "loss": 7.1072, - "step": 1413 - }, - { - "epoch": 0.79, - "learning_rate": 8.785046728971961e-05, - "loss": 6.9749, - "step": 1414 - }, - { - "epoch": 0.79, - "learning_rate": 8.761682242990654e-05, - "loss": 7.0596, - "step": 1415 - }, - { - "epoch": 0.79, - "learning_rate": 8.738317757009345e-05, - "loss": 7.2487, - "step": 1416 - }, - { - "epoch": 0.79, - "learning_rate": 8.714953271028036e-05, - "loss": 7.1493, - "step": 1417 - }, - { - "epoch": 0.79, - "learning_rate": 8.691588785046728e-05, - "loss": 6.9719, - "step": 1418 - }, - { - "epoch": 0.8, - "learning_rate": 8.66822429906542e-05, - "loss": 6.9269, - "step": 1419 - }, - { - "epoch": 0.8, - "learning_rate": 8.64485981308411e-05, - "loss": 6.8417, - "step": 1420 - }, - { - "epoch": 0.8, - "learning_rate": 8.621495327102803e-05, - "loss": 7.1814, - "step": 1421 - }, - { - "epoch": 0.8, - "learning_rate": 8.598130841121494e-05, - "loss": 7.072, - "step": 1422 - }, - { - "epoch": 0.8, - "learning_rate": 8.574766355140186e-05, - "loss": 6.8622, - "step": 1423 - }, - { - "epoch": 0.8, - "learning_rate": 8.551401869158877e-05, - "loss": 6.9659, - "step": 1424 - }, - { - "epoch": 0.8, - "learning_rate": 8.52803738317757e-05, - "loss": 7.071, - "step": 1425 - }, - { - "epoch": 0.8, - "learning_rate": 8.50467289719626e-05, - "loss": 7.096, - "step": 1426 - }, - { - "epoch": 0.8, - "learning_rate": 8.481308411214953e-05, - "loss": 6.9033, - "step": 1427 - }, - { - "epoch": 0.8, - "learning_rate": 8.457943925233644e-05, - "loss": 6.9965, - "step": 1428 - }, - { - "epoch": 0.8, - "learning_rate": 8.434579439252335e-05, - "loss": 6.9812, - "step": 1429 - }, - { - "epoch": 0.8, - "learning_rate": 8.411214953271027e-05, - "loss": 7.2081, - "step": 1430 - }, - { - "epoch": 0.8, - "learning_rate": 8.38785046728972e-05, - "loss": 7.0615, - "step": 1431 - }, - { - "epoch": 0.8, - "learning_rate": 8.36448598130841e-05, - "loss": 7.0771, - "step": 1432 - }, - { - "epoch": 0.8, - "learning_rate": 8.341121495327102e-05, - "loss": 7.1155, - "step": 1433 - }, - { - "epoch": 0.8, - "learning_rate": 8.317757009345794e-05, - "loss": 6.9879, - "step": 1434 - }, - { - "epoch": 0.8, - "learning_rate": 8.294392523364485e-05, - "loss": 7.1873, - "step": 1435 - }, - { - "epoch": 0.8, - "learning_rate": 8.271028037383176e-05, - "loss": 6.833, - "step": 1436 - }, - { - "epoch": 0.81, - "learning_rate": 8.247663551401869e-05, - "loss": 7.1303, - "step": 1437 - }, - { - "epoch": 0.81, - "learning_rate": 8.224299065420559e-05, - "loss": 6.9964, - "step": 1438 - }, - { - "epoch": 0.81, - "learning_rate": 8.200934579439252e-05, - "loss": 6.8957, - "step": 1439 - }, - { - "epoch": 0.81, - "learning_rate": 8.177570093457943e-05, - "loss": 6.8843, - "step": 1440 - }, - { - "epoch": 0.81, - "learning_rate": 8.154205607476634e-05, - "loss": 7.0924, - "step": 1441 - }, - { - "epoch": 0.81, - "learning_rate": 8.130841121495326e-05, - "loss": 7.0105, - "step": 1442 - }, - { - "epoch": 0.81, - "learning_rate": 8.107476635514018e-05, - "loss": 6.9324, - "step": 1443 - }, - { - "epoch": 0.81, - "learning_rate": 8.084112149532708e-05, - "loss": 6.8268, - "step": 1444 - }, - { - "epoch": 0.81, - "learning_rate": 8.060747663551401e-05, - "loss": 6.8697, - "step": 1445 - }, - { - "epoch": 0.81, - "learning_rate": 8.037383177570093e-05, - "loss": 7.0405, - "step": 1446 - }, - { - "epoch": 0.81, - "learning_rate": 8.014018691588784e-05, - "loss": 6.8525, - "step": 1447 - }, - { - "epoch": 0.81, - "learning_rate": 7.990654205607475e-05, - "loss": 6.6054, - "step": 1448 - }, - { - "epoch": 0.81, - "learning_rate": 7.967289719626168e-05, - "loss": 6.6658, - "step": 1449 - }, - { - "epoch": 0.81, - "learning_rate": 7.943925233644858e-05, - "loss": 6.6179, - "step": 1450 - }, - { - "epoch": 0.81, - "learning_rate": 7.920560747663551e-05, - "loss": 7.1583, - "step": 1451 - }, - { - "epoch": 0.81, - "learning_rate": 7.897196261682242e-05, - "loss": 7.0736, - "step": 1452 - }, - { - "epoch": 0.81, - "learning_rate": 7.873831775700933e-05, - "loss": 6.901, - "step": 1453 - }, - { - "epoch": 0.82, - "learning_rate": 7.850467289719625e-05, - "loss": 6.8665, - "step": 1454 - }, - { - "epoch": 0.82, - "learning_rate": 7.827102803738318e-05, - "loss": 6.929, - "step": 1455 - }, - { - "epoch": 0.82, - "learning_rate": 7.803738317757008e-05, - "loss": 6.9802, - "step": 1456 - }, - { - "epoch": 0.82, - "learning_rate": 7.7803738317757e-05, - "loss": 7.0804, - "step": 1457 - }, - { - "epoch": 0.82, - "learning_rate": 7.757009345794392e-05, - "loss": 7.0749, - "step": 1458 - }, - { - "epoch": 0.82, - "learning_rate": 7.733644859813083e-05, - "loss": 6.9733, - "step": 1459 - }, - { - "epoch": 0.82, - "learning_rate": 7.710280373831774e-05, - "loss": 6.9622, - "step": 1460 - }, - { - "epoch": 0.82, - "learning_rate": 7.686915887850467e-05, - "loss": 7.0163, - "step": 1461 - }, - { - "epoch": 0.82, - "learning_rate": 7.663551401869157e-05, - "loss": 7.0044, - "step": 1462 - }, - { - "epoch": 0.82, - "learning_rate": 7.64018691588785e-05, - "loss": 6.8313, - "step": 1463 - }, - { - "epoch": 0.82, - "learning_rate": 7.616822429906541e-05, - "loss": 6.8882, - "step": 1464 - }, - { - "epoch": 0.82, - "learning_rate": 7.593457943925233e-05, - "loss": 7.0497, - "step": 1465 - }, - { - "epoch": 0.82, - "learning_rate": 7.570093457943924e-05, - "loss": 6.9395, - "step": 1466 - }, - { - "epoch": 0.82, - "learning_rate": 7.546728971962617e-05, - "loss": 7.0603, - "step": 1467 - }, - { - "epoch": 0.82, - "learning_rate": 7.523364485981307e-05, - "loss": 6.9133, - "step": 1468 - }, - { - "epoch": 0.82, - "learning_rate": 7.5e-05, - "loss": 6.7417, - "step": 1469 - }, - { - "epoch": 0.82, - "learning_rate": 7.476635514018691e-05, - "loss": 6.8393, - "step": 1470 - }, - { - "epoch": 0.82, - "learning_rate": 7.453271028037382e-05, - "loss": 7.0191, - "step": 1471 - }, - { - "epoch": 0.83, - "learning_rate": 7.429906542056073e-05, - "loss": 6.8629, - "step": 1472 - }, - { - "epoch": 0.83, - "learning_rate": 7.406542056074766e-05, - "loss": 6.8455, - "step": 1473 - }, - { - "epoch": 0.83, - "learning_rate": 7.383177570093458e-05, - "loss": 6.7303, - "step": 1474 - }, - { - "epoch": 0.83, - "learning_rate": 7.359813084112149e-05, - "loss": 7.0396, - "step": 1475 - }, - { - "epoch": 0.83, - "learning_rate": 7.33644859813084e-05, - "loss": 7.0671, - "step": 1476 - }, - { - "epoch": 0.83, - "learning_rate": 7.313084112149532e-05, - "loss": 7.2388, - "step": 1477 - }, - { - "epoch": 0.83, - "learning_rate": 7.289719626168223e-05, - "loss": 6.7881, - "step": 1478 - }, - { - "epoch": 0.83, - "learning_rate": 7.266355140186916e-05, - "loss": 6.9322, - "step": 1479 - }, - { - "epoch": 0.83, - "learning_rate": 7.242990654205607e-05, - "loss": 6.8778, - "step": 1480 - }, - { - "epoch": 0.83, - "learning_rate": 7.219626168224298e-05, - "loss": 7.0878, - "step": 1481 - }, - { - "epoch": 0.83, - "learning_rate": 7.19626168224299e-05, - "loss": 7.1191, - "step": 1482 - }, - { - "epoch": 0.83, - "learning_rate": 7.172897196261681e-05, - "loss": 6.991, - "step": 1483 - }, - { - "epoch": 0.83, - "learning_rate": 7.149532710280372e-05, - "loss": 6.7618, - "step": 1484 - }, - { - "epoch": 0.83, - "learning_rate": 7.126168224299065e-05, - "loss": 7.102, - "step": 1485 - }, - { - "epoch": 0.83, - "learning_rate": 7.102803738317757e-05, - "loss": 7.0795, - "step": 1486 - }, - { - "epoch": 0.83, - "learning_rate": 7.079439252336448e-05, - "loss": 6.9457, - "step": 1487 - }, - { - "epoch": 0.83, - "learning_rate": 7.056074766355139e-05, - "loss": 6.9655, - "step": 1488 - }, - { - "epoch": 0.83, - "learning_rate": 7.03271028037383e-05, - "loss": 7.067, - "step": 1489 - }, - { - "epoch": 0.84, - "learning_rate": 7.009345794392522e-05, - "loss": 6.7452, - "step": 1490 - }, - { - "epoch": 0.84, - "learning_rate": 6.985981308411215e-05, - "loss": 7.0299, - "step": 1491 - }, - { - "epoch": 0.84, - "learning_rate": 6.962616822429906e-05, - "loss": 7.0919, - "step": 1492 - }, - { - "epoch": 0.84, - "learning_rate": 6.939252336448597e-05, - "loss": 6.7958, - "step": 1493 - }, - { - "epoch": 0.84, - "learning_rate": 6.915887850467289e-05, - "loss": 7.053, - "step": 1494 - }, - { - "epoch": 0.84, - "learning_rate": 6.89252336448598e-05, - "loss": 6.8927, - "step": 1495 - }, - { - "epoch": 0.84, - "learning_rate": 6.869158878504672e-05, - "loss": 7.127, - "step": 1496 - }, - { - "epoch": 0.84, - "learning_rate": 6.845794392523364e-05, - "loss": 6.8373, - "step": 1497 - }, - { - "epoch": 0.84, - "learning_rate": 6.822429906542056e-05, - "loss": 6.7983, - "step": 1498 - }, - { - "epoch": 0.84, - "learning_rate": 6.799065420560747e-05, - "loss": 6.6486, - "step": 1499 - }, - { - "epoch": 0.84, - "learning_rate": 6.775700934579438e-05, - "loss": 6.5927, - "step": 1500 - }, - { - "epoch": 0.84, - "eval_loss": 7.354469299316406, - "eval_runtime": 892.5545, - "eval_samples_per_second": 2.96, - "eval_steps_per_second": 0.371, - "eval_wer": 2.0380578209020377, - "step": 1500 - }, - { - "epoch": 0.84, - "learning_rate": 6.75233644859813e-05, - "loss": 6.8504, - "step": 1501 - }, - { - "epoch": 0.84, - "learning_rate": 6.728971962616821e-05, - "loss": 7.1311, - "step": 1502 - }, - { - "epoch": 0.84, - "learning_rate": 6.705607476635514e-05, - "loss": 6.9482, - "step": 1503 - }, - { - "epoch": 0.84, - "learning_rate": 6.682242990654205e-05, - "loss": 7.0233, - "step": 1504 - }, - { - "epoch": 0.84, - "learning_rate": 6.658878504672897e-05, - "loss": 6.9965, - "step": 1505 - }, - { - "epoch": 0.84, - "learning_rate": 6.635514018691588e-05, - "loss": 7.0362, - "step": 1506 - }, - { - "epoch": 0.84, - "learning_rate": 6.612149532710279e-05, - "loss": 6.9526, - "step": 1507 - }, - { - "epoch": 0.85, - "learning_rate": 6.58878504672897e-05, - "loss": 6.8416, - "step": 1508 - }, - { - "epoch": 0.85, - "learning_rate": 6.565420560747663e-05, - "loss": 7.0551, - "step": 1509 - }, - { - "epoch": 0.85, - "learning_rate": 6.542056074766355e-05, - "loss": 6.9162, - "step": 1510 - }, - { - "epoch": 0.85, - "learning_rate": 6.518691588785046e-05, - "loss": 6.8999, - "step": 1511 - }, - { - "epoch": 0.85, - "learning_rate": 6.495327102803737e-05, - "loss": 6.8981, - "step": 1512 - }, - { - "epoch": 0.85, - "learning_rate": 6.471962616822429e-05, - "loss": 6.7938, - "step": 1513 - }, - { - "epoch": 0.85, - "learning_rate": 6.44859813084112e-05, - "loss": 6.9569, - "step": 1514 - }, - { - "epoch": 0.85, - "learning_rate": 6.425233644859813e-05, - "loss": 6.7673, - "step": 1515 - }, - { - "epoch": 0.85, - "learning_rate": 6.401869158878504e-05, - "loss": 7.0115, - "step": 1516 - }, - { - "epoch": 0.85, - "learning_rate": 6.378504672897196e-05, - "loss": 6.9042, - "step": 1517 - }, - { - "epoch": 0.85, - "learning_rate": 6.355140186915887e-05, - "loss": 6.8785, - "step": 1518 - }, - { - "epoch": 0.85, - "learning_rate": 6.331775700934578e-05, - "loss": 6.926, - "step": 1519 - }, - { - "epoch": 0.85, - "learning_rate": 6.30841121495327e-05, - "loss": 7.0099, - "step": 1520 - }, - { - "epoch": 0.85, - "learning_rate": 6.285046728971962e-05, - "loss": 6.8627, - "step": 1521 - }, - { - "epoch": 0.85, - "learning_rate": 6.261682242990654e-05, - "loss": 7.0151, - "step": 1522 - }, - { - "epoch": 0.85, - "learning_rate": 6.238317757009345e-05, - "loss": 7.094, - "step": 1523 - }, - { - "epoch": 0.85, - "learning_rate": 6.214953271028036e-05, - "loss": 7.0216, - "step": 1524 - }, - { - "epoch": 0.85, - "learning_rate": 6.191588785046728e-05, - "loss": 6.8508, - "step": 1525 - }, - { - "epoch": 0.86, - "learning_rate": 6.168224299065419e-05, - "loss": 6.7578, - "step": 1526 - }, - { - "epoch": 0.86, - "learning_rate": 6.144859813084112e-05, - "loss": 7.0334, - "step": 1527 - }, - { - "epoch": 0.86, - "learning_rate": 6.121495327102803e-05, - "loss": 6.8264, - "step": 1528 - }, - { - "epoch": 0.86, - "learning_rate": 6.098130841121495e-05, - "loss": 6.9025, - "step": 1529 - }, - { - "epoch": 0.86, - "learning_rate": 6.074766355140186e-05, - "loss": 6.9113, - "step": 1530 - }, - { - "epoch": 0.86, - "learning_rate": 6.051401869158878e-05, - "loss": 6.9003, - "step": 1531 - }, - { - "epoch": 0.86, - "learning_rate": 6.0280373831775694e-05, - "loss": 6.6228, - "step": 1532 - }, - { - "epoch": 0.86, - "learning_rate": 6.004672897196261e-05, - "loss": 7.0121, - "step": 1533 - }, - { - "epoch": 0.86, - "learning_rate": 5.981308411214953e-05, - "loss": 6.9171, - "step": 1534 - }, - { - "epoch": 0.86, - "learning_rate": 5.957943925233644e-05, - "loss": 7.0514, - "step": 1535 - }, - { - "epoch": 0.86, - "learning_rate": 5.9345794392523356e-05, - "loss": 6.9852, - "step": 1536 - }, - { - "epoch": 0.86, - "learning_rate": 5.9112149532710276e-05, - "loss": 6.8433, - "step": 1537 - }, - { - "epoch": 0.86, - "learning_rate": 5.887850467289719e-05, - "loss": 7.1214, - "step": 1538 - }, - { - "epoch": 0.86, - "learning_rate": 5.86448598130841e-05, - "loss": 6.9136, - "step": 1539 - }, - { - "epoch": 0.86, - "learning_rate": 5.8411214953271024e-05, - "loss": 6.9088, - "step": 1540 - }, - { - "epoch": 0.86, - "learning_rate": 5.817757009345794e-05, - "loss": 6.9068, - "step": 1541 - }, - { - "epoch": 0.86, - "learning_rate": 5.794392523364485e-05, - "loss": 6.7177, - "step": 1542 - }, - { - "epoch": 0.86, - "learning_rate": 5.771028037383177e-05, - "loss": 6.9991, - "step": 1543 - }, - { - "epoch": 0.87, - "learning_rate": 5.7476635514018685e-05, - "loss": 6.8745, - "step": 1544 - }, - { - "epoch": 0.87, - "learning_rate": 5.72429906542056e-05, - "loss": 6.9885, - "step": 1545 - }, - { - "epoch": 0.87, - "learning_rate": 5.700934579439252e-05, - "loss": 6.8174, - "step": 1546 - }, - { - "epoch": 0.87, - "learning_rate": 5.677570093457943e-05, - "loss": 6.9149, - "step": 1547 - }, - { - "epoch": 0.87, - "learning_rate": 5.6542056074766346e-05, - "loss": 6.7636, - "step": 1548 - }, - { - "epoch": 0.87, - "learning_rate": 5.630841121495327e-05, - "loss": 6.8673, - "step": 1549 - }, - { - "epoch": 0.87, - "learning_rate": 5.607476635514018e-05, - "loss": 6.5537, - "step": 1550 - }, - { - "epoch": 0.87, - "learning_rate": 5.5841121495327094e-05, - "loss": 7.0685, - "step": 1551 - }, - { - "epoch": 0.87, - "learning_rate": 5.5607476635514014e-05, - "loss": 7.0549, - "step": 1552 - }, - { - "epoch": 0.87, - "learning_rate": 5.537383177570093e-05, - "loss": 6.9184, - "step": 1553 - }, - { - "epoch": 0.87, - "learning_rate": 5.514018691588784e-05, - "loss": 6.7083, - "step": 1554 - }, - { - "epoch": 0.87, - "learning_rate": 5.490654205607476e-05, - "loss": 7.03, - "step": 1555 - }, - { - "epoch": 0.87, - "learning_rate": 5.4672897196261676e-05, - "loss": 7.1109, - "step": 1556 - }, - { - "epoch": 0.87, - "learning_rate": 5.443925233644859e-05, - "loss": 6.8695, - "step": 1557 - }, - { - "epoch": 0.87, - "learning_rate": 5.420560747663551e-05, - "loss": 6.9423, - "step": 1558 - }, - { - "epoch": 0.87, - "learning_rate": 5.397196261682242e-05, - "loss": 6.8934, - "step": 1559 - }, - { - "epoch": 0.87, - "learning_rate": 5.373831775700934e-05, - "loss": 7.0202, - "step": 1560 - }, - { - "epoch": 0.88, - "learning_rate": 5.350467289719626e-05, - "loss": 7.0105, - "step": 1561 - }, - { - "epoch": 0.88, - "learning_rate": 5.327102803738317e-05, - "loss": 7.0227, - "step": 1562 - }, - { - "epoch": 0.88, - "learning_rate": 5.3037383177570085e-05, - "loss": 6.8664, - "step": 1563 - }, - { - "epoch": 0.88, - "learning_rate": 5.2803738317757005e-05, - "loss": 6.7332, - "step": 1564 - }, - { - "epoch": 0.88, - "learning_rate": 5.257009345794392e-05, - "loss": 6.9855, - "step": 1565 - }, - { - "epoch": 0.88, - "learning_rate": 5.233644859813083e-05, - "loss": 7.0127, - "step": 1566 - }, - { - "epoch": 0.88, - "learning_rate": 5.210280373831775e-05, - "loss": 6.9085, - "step": 1567 - }, - { - "epoch": 0.88, - "learning_rate": 5.1869158878504666e-05, - "loss": 6.9852, - "step": 1568 - }, - { - "epoch": 0.88, - "learning_rate": 5.163551401869158e-05, - "loss": 7.0122, - "step": 1569 - }, - { - "epoch": 0.88, - "learning_rate": 5.14018691588785e-05, - "loss": 6.9711, - "step": 1570 - }, - { - "epoch": 0.88, - "learning_rate": 5.1168224299065414e-05, - "loss": 7.0087, - "step": 1571 - }, - { - "epoch": 0.88, - "learning_rate": 5.093457943925233e-05, - "loss": 7.272, - "step": 1572 - }, - { - "epoch": 0.88, - "learning_rate": 5.070093457943925e-05, - "loss": 6.9771, - "step": 1573 - }, - { - "epoch": 0.88, - "learning_rate": 5.046728971962616e-05, - "loss": 6.739, - "step": 1574 - }, - { - "epoch": 0.88, - "learning_rate": 5.0233644859813075e-05, - "loss": 6.6974, - "step": 1575 - }, - { - "epoch": 0.88, - "learning_rate": 4.9999999999999996e-05, - "loss": 6.7734, - "step": 1576 - }, - { - "epoch": 0.88, - "learning_rate": 4.976635514018691e-05, - "loss": 6.7853, - "step": 1577 - }, - { - "epoch": 0.88, - "learning_rate": 4.953271028037382e-05, - "loss": 6.966, - "step": 1578 - }, - { - "epoch": 0.89, - "learning_rate": 4.929906542056074e-05, - "loss": 6.8902, - "step": 1579 - }, - { - "epoch": 0.89, - "learning_rate": 4.906542056074766e-05, - "loss": 7.0419, - "step": 1580 - }, - { - "epoch": 0.89, - "learning_rate": 4.883177570093457e-05, - "loss": 7.1364, - "step": 1581 - }, - { - "epoch": 0.89, - "learning_rate": 4.859813084112149e-05, - "loss": 7.0964, - "step": 1582 - }, - { - "epoch": 0.89, - "learning_rate": 4.8364485981308405e-05, - "loss": 7.0061, - "step": 1583 - }, - { - "epoch": 0.89, - "learning_rate": 4.813084112149532e-05, - "loss": 6.8292, - "step": 1584 - }, - { - "epoch": 0.89, - "learning_rate": 4.789719626168224e-05, - "loss": 6.947, - "step": 1585 - }, - { - "epoch": 0.89, - "learning_rate": 4.766355140186915e-05, - "loss": 6.9463, - "step": 1586 - }, - { - "epoch": 0.89, - "learning_rate": 4.7429906542056066e-05, - "loss": 6.9593, - "step": 1587 - }, { "epoch": 0.89, - "learning_rate": 4.7196261682242986e-05, - "loss": 6.8991, - "step": 1588 + "learning_rate": 8.188775510204081e-05, + "loss": 6.7076, + "step": 790 }, { "epoch": 0.89, - "learning_rate": 4.69626168224299e-05, - "loss": 6.8887, - "step": 1589 + "learning_rate": 8.112244897959184e-05, + "loss": 6.5998, + "step": 791 }, { "epoch": 0.89, - "learning_rate": 4.6728971962616814e-05, - "loss": 6.9038, - "step": 1590 + "learning_rate": 8.035714285714285e-05, + "loss": 6.5586, + "step": 792 }, { "epoch": 0.89, - "learning_rate": 4.6495327102803734e-05, - "loss": 6.8638, - "step": 1591 + "learning_rate": 7.959183673469388e-05, + "loss": 6.649, + "step": 793 }, { "epoch": 0.89, - "learning_rate": 4.626168224299065e-05, - "loss": 6.8827, - "step": 1592 + "learning_rate": 7.882653061224489e-05, + "loss": 6.5968, + "step": 794 }, { "epoch": 0.89, - "learning_rate": 4.602803738317756e-05, - "loss": 6.8068, - "step": 1593 + "learning_rate": 7.806122448979592e-05, + "loss": 6.7391, + "step": 795 }, { "epoch": 0.89, - "learning_rate": 4.579439252336448e-05, - "loss": 6.6732, - "step": 1594 + "learning_rate": 7.729591836734693e-05, + "loss": 6.5806, + "step": 796 }, { "epoch": 0.89, - "learning_rate": 4.5560747663551395e-05, - "loss": 6.926, - "step": 1595 + "learning_rate": 7.653061224489796e-05, + "loss": 6.6103, + "step": 797 }, { "epoch": 0.89, - "learning_rate": 4.532710280373831e-05, - "loss": 6.8826, - "step": 1596 - }, - { - "epoch": 0.9, - "learning_rate": 4.509345794392523e-05, - "loss": 6.7865, - "step": 1597 - }, - { - "epoch": 0.9, - "learning_rate": 4.485981308411214e-05, - "loss": 6.9715, - "step": 1598 - }, - { - "epoch": 0.9, - "learning_rate": 4.4626168224299057e-05, - "loss": 7.3356, - "step": 1599 - }, - { - "epoch": 0.9, - "learning_rate": 4.439252336448598e-05, - "loss": 6.5279, - "step": 1600 - }, - { - "epoch": 0.9, - "learning_rate": 4.415887850467289e-05, - "loss": 7.0714, - "step": 1601 - }, - { - "epoch": 0.9, - "learning_rate": 4.3925233644859804e-05, - "loss": 6.8224, - "step": 1602 - }, - { - "epoch": 0.9, - "learning_rate": 4.3691588785046725e-05, - "loss": 7.1106, - "step": 1603 - }, - { - "epoch": 0.9, - "learning_rate": 4.345794392523364e-05, - "loss": 7.1471, - "step": 1604 - }, - { - "epoch": 0.9, - "learning_rate": 4.322429906542055e-05, - "loss": 6.8026, - "step": 1605 + "learning_rate": 7.576530612244897e-05, + "loss": 6.5708, + "step": 798 }, { "epoch": 0.9, - "learning_rate": 4.299065420560747e-05, - "loss": 6.9473, - "step": 1606 + "learning_rate": 7.5e-05, + "loss": 6.6839, + "step": 799 }, { "epoch": 0.9, - "learning_rate": 4.2757009345794386e-05, - "loss": 6.9513, - "step": 1607 + "learning_rate": 7.423469387755102e-05, + "loss": 6.2081, + "step": 800 }, { "epoch": 0.9, - "learning_rate": 4.25233644859813e-05, - "loss": 6.8104, - "step": 1608 + "learning_rate": 7.346938775510203e-05, + "loss": 6.8204, + "step": 801 }, { "epoch": 0.9, - "learning_rate": 4.228971962616822e-05, - "loss": 7.0781, - "step": 1609 + "learning_rate": 7.270408163265306e-05, + "loss": 6.7398, + "step": 802 }, { "epoch": 0.9, - "learning_rate": 4.2056074766355134e-05, - "loss": 6.9259, - "step": 1610 + "learning_rate": 7.193877551020407e-05, + "loss": 6.8149, + "step": 803 }, { "epoch": 0.9, - "learning_rate": 4.182242990654205e-05, - "loss": 6.9721, - "step": 1611 + "learning_rate": 7.11734693877551e-05, + "loss": 6.821, + "step": 804 }, { "epoch": 0.9, - "learning_rate": 4.158878504672897e-05, - "loss": 7.1136, - "step": 1612 + "learning_rate": 7.040816326530612e-05, + "loss": 6.5948, + "step": 805 }, { "epoch": 0.9, - "learning_rate": 4.135514018691588e-05, - "loss": 7.0782, - "step": 1613 + "learning_rate": 6.964285714285713e-05, + "loss": 6.8221, + "step": 806 }, { "epoch": 0.9, - "learning_rate": 4.1121495327102795e-05, - "loss": 6.8828, - "step": 1614 - }, - { - "epoch": 0.91, - "learning_rate": 4.0887850467289715e-05, - "loss": 6.9101, - "step": 1615 - }, - { - "epoch": 0.91, - "learning_rate": 4.065420560747663e-05, - "loss": 6.9111, - "step": 1616 - }, - { - "epoch": 0.91, - "learning_rate": 4.042056074766354e-05, - "loss": 7.0139, - "step": 1617 - }, - { - "epoch": 0.91, - "learning_rate": 4.018691588785046e-05, - "loss": 7.0112, - "step": 1618 - }, - { - "epoch": 0.91, - "learning_rate": 3.9953271028037377e-05, - "loss": 6.9344, - "step": 1619 - }, - { - "epoch": 0.91, - "learning_rate": 3.971962616822429e-05, - "loss": 6.9783, - "step": 1620 - }, - { - "epoch": 0.91, - "learning_rate": 3.948598130841121e-05, - "loss": 7.1215, - "step": 1621 - }, - { - "epoch": 0.91, - "learning_rate": 3.9252336448598124e-05, - "loss": 6.8193, - "step": 1622 - }, - { - "epoch": 0.91, - "learning_rate": 3.901869158878504e-05, - "loss": 6.9952, - "step": 1623 + "learning_rate": 6.887755102040816e-05, + "loss": 6.8279, + "step": 807 }, { "epoch": 0.91, - "learning_rate": 3.878504672897196e-05, - "loss": 6.7212, - "step": 1624 + "learning_rate": 6.811224489795917e-05, + "loss": 6.6116, + "step": 808 }, { "epoch": 0.91, - "learning_rate": 3.855140186915887e-05, - "loss": 7.0557, - "step": 1625 + "learning_rate": 6.73469387755102e-05, + "loss": 6.7841, + "step": 809 }, { "epoch": 0.91, - "learning_rate": 3.8317757009345785e-05, - "loss": 7.0742, - "step": 1626 + "learning_rate": 6.658163265306122e-05, + "loss": 6.6926, + "step": 810 }, { "epoch": 0.91, - "learning_rate": 3.8084112149532706e-05, - "loss": 7.0449, - "step": 1627 + "learning_rate": 6.581632653061224e-05, + "loss": 6.8074, + "step": 811 }, { "epoch": 0.91, - "learning_rate": 3.785046728971962e-05, - "loss": 6.8781, - "step": 1628 + "learning_rate": 6.505102040816326e-05, + "loss": 6.8235, + "step": 812 }, { "epoch": 0.91, - "learning_rate": 3.761682242990653e-05, - "loss": 6.98, - "step": 1629 + "learning_rate": 6.428571428571427e-05, + "loss": 6.6634, + "step": 813 }, { "epoch": 0.91, - "learning_rate": 3.7383177570093454e-05, - "loss": 6.9161, - "step": 1630 + "learning_rate": 6.35204081632653e-05, + "loss": 6.7408, + "step": 814 }, { "epoch": 0.91, - "learning_rate": 3.714953271028037e-05, - "loss": 6.846, - "step": 1631 + "learning_rate": 6.275510204081633e-05, + "loss": 6.7874, + "step": 815 }, { "epoch": 0.91, - "learning_rate": 3.691588785046729e-05, - "loss": 7.1471, - "step": 1632 - }, - { - "epoch": 0.92, - "learning_rate": 3.66822429906542e-05, - "loss": 6.7796, - "step": 1633 - }, - { - "epoch": 0.92, - "learning_rate": 3.6448598130841115e-05, - "loss": 6.9572, - "step": 1634 - }, - { - "epoch": 0.92, - "learning_rate": 3.6214953271028035e-05, - "loss": 6.9526, - "step": 1635 - }, - { - "epoch": 0.92, - "learning_rate": 3.598130841121495e-05, - "loss": 6.8952, - "step": 1636 - }, - { - "epoch": 0.92, - "learning_rate": 3.574766355140186e-05, - "loss": 6.8766, - "step": 1637 - }, - { - "epoch": 0.92, - "learning_rate": 3.551401869158878e-05, - "loss": 7.1076, - "step": 1638 - }, - { - "epoch": 0.92, - "learning_rate": 3.5280373831775697e-05, - "loss": 6.7836, - "step": 1639 - }, - { - "epoch": 0.92, - "learning_rate": 3.504672897196261e-05, - "loss": 6.8504, - "step": 1640 - }, - { - "epoch": 0.92, - "learning_rate": 3.481308411214953e-05, - "loss": 7.072, - "step": 1641 - }, - { - "epoch": 0.92, - "learning_rate": 3.4579439252336444e-05, - "loss": 6.8679, - "step": 1642 - }, - { - "epoch": 0.92, - "learning_rate": 3.434579439252336e-05, - "loss": 6.9, - "step": 1643 - }, - { - "epoch": 0.92, - "learning_rate": 3.411214953271028e-05, - "loss": 6.813, - "step": 1644 - }, - { - "epoch": 0.92, - "learning_rate": 3.387850467289719e-05, - "loss": 6.9564, - "step": 1645 + "learning_rate": 6.198979591836734e-05, + "loss": 6.6599, + "step": 816 }, { "epoch": 0.92, - "learning_rate": 3.3644859813084105e-05, - "loss": 7.0088, - "step": 1646 + "learning_rate": 6.122448979591836e-05, + "loss": 6.8251, + "step": 817 }, { "epoch": 0.92, - "learning_rate": 3.3411214953271026e-05, - "loss": 6.6815, - "step": 1647 + "learning_rate": 6.045918367346938e-05, + "loss": 6.7198, + "step": 818 }, { "epoch": 0.92, - "learning_rate": 3.317757009345794e-05, - "loss": 6.97, - "step": 1648 + "learning_rate": 5.96938775510204e-05, + "loss": 6.649, + "step": 819 }, { "epoch": 0.92, - "learning_rate": 3.294392523364485e-05, - "loss": 6.8449, - "step": 1649 + "learning_rate": 5.892857142857142e-05, + "loss": 6.8378, + "step": 820 }, { "epoch": 0.92, - "learning_rate": 3.2710280373831774e-05, - "loss": 6.3524, - "step": 1650 - }, - { - "epoch": 0.93, - "learning_rate": 3.247663551401869e-05, - "loss": 6.8893, - "step": 1651 - }, - { - "epoch": 0.93, - "learning_rate": 3.22429906542056e-05, - "loss": 6.9998, - "step": 1652 - }, - { - "epoch": 0.93, - "learning_rate": 3.200934579439252e-05, - "loss": 6.9749, - "step": 1653 - }, - { - "epoch": 0.93, - "learning_rate": 3.1775700934579435e-05, - "loss": 6.8593, - "step": 1654 - }, - { - "epoch": 0.93, - "learning_rate": 3.154205607476635e-05, - "loss": 6.7728, - "step": 1655 - }, - { - "epoch": 0.93, - "learning_rate": 3.130841121495327e-05, - "loss": 6.9486, - "step": 1656 - }, - { - "epoch": 0.93, - "learning_rate": 3.107476635514018e-05, - "loss": 7.0639, - "step": 1657 - }, - { - "epoch": 0.93, - "learning_rate": 3.0841121495327096e-05, - "loss": 6.8194, - "step": 1658 - }, - { - "epoch": 0.93, - "learning_rate": 3.0607476635514017e-05, - "loss": 6.8089, - "step": 1659 - }, - { - "epoch": 0.93, - "learning_rate": 3.037383177570093e-05, - "loss": 7.0252, - "step": 1660 - }, - { - "epoch": 0.93, - "learning_rate": 3.0140186915887847e-05, - "loss": 6.9422, - "step": 1661 - }, - { - "epoch": 0.93, - "learning_rate": 2.9906542056074764e-05, - "loss": 6.7673, - "step": 1662 - }, - { - "epoch": 0.93, - "learning_rate": 2.9672897196261678e-05, - "loss": 6.8956, - "step": 1663 - }, - { - "epoch": 0.93, - "learning_rate": 2.9439252336448595e-05, - "loss": 7.0141, - "step": 1664 - }, - { - "epoch": 0.93, - "learning_rate": 2.9205607476635512e-05, - "loss": 7.0156, - "step": 1665 - }, - { - "epoch": 0.93, - "learning_rate": 2.8971962616822425e-05, - "loss": 6.9334, - "step": 1666 - }, - { - "epoch": 0.93, - "learning_rate": 2.8738317757009342e-05, - "loss": 6.969, - "step": 1667 - }, - { - "epoch": 0.93, - "learning_rate": 2.850467289719626e-05, - "loss": 6.9254, - "step": 1668 - }, - { - "epoch": 0.94, - "learning_rate": 2.8271028037383173e-05, - "loss": 6.8406, - "step": 1669 - }, - { - "epoch": 0.94, - "learning_rate": 2.803738317757009e-05, - "loss": 7.0885, - "step": 1670 - }, - { - "epoch": 0.94, - "learning_rate": 2.7803738317757007e-05, - "loss": 6.9049, - "step": 1671 - }, - { - "epoch": 0.94, - "learning_rate": 2.757009345794392e-05, - "loss": 7.1204, - "step": 1672 - }, - { - "epoch": 0.94, - "learning_rate": 2.7336448598130838e-05, - "loss": 7.0046, - "step": 1673 + "learning_rate": 5.816326530612244e-05, + "loss": 6.8143, + "step": 821 }, { - "epoch": 0.94, - "learning_rate": 2.7102803738317755e-05, - "loss": 7.0819, - "step": 1674 + "epoch": 0.92, + "learning_rate": 5.739795918367346e-05, + "loss": 6.6955, + "step": 822 }, { - "epoch": 0.94, - "learning_rate": 2.686915887850467e-05, - "loss": 7.2802, - "step": 1675 + "epoch": 0.92, + "learning_rate": 5.663265306122448e-05, + "loss": 6.8389, + "step": 823 }, { - "epoch": 0.94, - "learning_rate": 2.6635514018691585e-05, - "loss": 6.8029, - "step": 1676 + "epoch": 0.92, + "learning_rate": 5.586734693877551e-05, + "loss": 6.6973, + "step": 824 }, { - "epoch": 0.94, - "learning_rate": 2.6401869158878502e-05, - "loss": 6.8034, - "step": 1677 + "epoch": 0.92, + "learning_rate": 5.510204081632653e-05, + "loss": 6.9625, + "step": 825 }, { - "epoch": 0.94, - "learning_rate": 2.6168224299065416e-05, - "loss": 6.8992, - "step": 1678 + "epoch": 0.93, + "learning_rate": 5.433673469387755e-05, + "loss": 6.8445, + "step": 826 }, { - "epoch": 0.94, - "learning_rate": 2.5934579439252333e-05, - "loss": 7.0654, - "step": 1679 + "epoch": 0.93, + "learning_rate": 5.357142857142857e-05, + "loss": 6.7165, + "step": 827 }, { - "epoch": 0.94, - "learning_rate": 2.570093457943925e-05, - "loss": 7.0052, - "step": 1680 + "epoch": 0.93, + "learning_rate": 5.280612244897959e-05, + "loss": 6.632, + "step": 828 }, { - "epoch": 0.94, - "learning_rate": 2.5467289719626164e-05, - "loss": 6.7841, - "step": 1681 + "epoch": 0.93, + "learning_rate": 5.204081632653061e-05, + "loss": 6.6587, + "step": 829 }, { - "epoch": 0.94, - "learning_rate": 2.523364485981308e-05, - "loss": 6.792, - "step": 1682 + "epoch": 0.93, + "learning_rate": 5.1275510204081626e-05, + "loss": 6.6842, + "step": 830 }, { - "epoch": 0.94, - "learning_rate": 2.4999999999999998e-05, - "loss": 6.9385, - "step": 1683 + "epoch": 0.93, + "learning_rate": 5.0510204081632645e-05, + "loss": 6.6688, + "step": 831 }, { - "epoch": 0.94, - "learning_rate": 2.476635514018691e-05, - "loss": 6.9271, - "step": 1684 + "epoch": 0.93, + "learning_rate": 4.9744897959183664e-05, + "loss": 6.699, + "step": 832 }, { - "epoch": 0.94, - "learning_rate": 2.453271028037383e-05, - "loss": 7.0395, - "step": 1685 + "epoch": 0.93, + "learning_rate": 4.897959183673468e-05, + "loss": 6.7232, + "step": 833 }, { - "epoch": 0.95, - "learning_rate": 2.4299065420560745e-05, - "loss": 6.9648, - "step": 1686 + "epoch": 0.93, + "learning_rate": 4.8214285714285716e-05, + "loss": 6.7536, + "step": 834 }, { - "epoch": 0.95, - "learning_rate": 2.406542056074766e-05, - "loss": 7.0288, - "step": 1687 + "epoch": 0.94, + "learning_rate": 4.7448979591836735e-05, + "loss": 6.6919, + "step": 835 }, { - "epoch": 0.95, - "learning_rate": 2.3831775700934576e-05, - "loss": 7.1777, - "step": 1688 + "epoch": 0.94, + "learning_rate": 4.6683673469387754e-05, + "loss": 6.8364, + "step": 836 }, { - "epoch": 0.95, - "learning_rate": 2.3598130841121493e-05, - "loss": 7.0212, - "step": 1689 + "epoch": 0.94, + "learning_rate": 4.591836734693877e-05, + "loss": 6.7028, + "step": 837 }, { - "epoch": 0.95, - "learning_rate": 2.3364485981308407e-05, - "loss": 6.7181, - "step": 1690 + "epoch": 0.94, + "learning_rate": 4.515306122448979e-05, + "loss": 6.8399, + "step": 838 }, { - "epoch": 0.95, - "learning_rate": 2.3130841121495324e-05, - "loss": 7.0158, - "step": 1691 + "epoch": 0.94, + "learning_rate": 4.438775510204081e-05, + "loss": 6.6795, + "step": 839 }, { - "epoch": 0.95, - "learning_rate": 2.289719626168224e-05, - "loss": 6.9559, - "step": 1692 + "epoch": 0.94, + "learning_rate": 4.362244897959183e-05, + "loss": 6.6221, + "step": 840 }, { - "epoch": 0.95, - "learning_rate": 2.2663551401869154e-05, - "loss": 7.114, - "step": 1693 + "epoch": 0.94, + "learning_rate": 4.285714285714285e-05, + "loss": 6.7963, + "step": 841 }, { - "epoch": 0.95, - "learning_rate": 2.242990654205607e-05, - "loss": 7.0529, - "step": 1694 + "epoch": 0.94, + "learning_rate": 4.209183673469387e-05, + "loss": 6.6097, + "step": 842 }, { "epoch": 0.95, - "learning_rate": 2.219626168224299e-05, - "loss": 6.9065, - "step": 1695 + "learning_rate": 4.132653061224489e-05, + "loss": 6.8284, + "step": 843 }, { "epoch": 0.95, - "learning_rate": 2.1962616822429902e-05, - "loss": 6.9567, - "step": 1696 + "learning_rate": 4.056122448979592e-05, + "loss": 6.6447, + "step": 844 }, { "epoch": 0.95, - "learning_rate": 2.172897196261682e-05, - "loss": 6.6178, - "step": 1697 + "learning_rate": 3.979591836734694e-05, + "loss": 6.6868, + "step": 845 }, { "epoch": 0.95, - "learning_rate": 2.1495327102803736e-05, - "loss": 6.992, - "step": 1698 + "learning_rate": 3.903061224489796e-05, + "loss": 6.5554, + "step": 846 }, { "epoch": 0.95, - "learning_rate": 2.126168224299065e-05, - "loss": 6.6052, - "step": 1699 + "learning_rate": 3.826530612244898e-05, + "loss": 6.413, + "step": 847 }, { "epoch": 0.95, - "learning_rate": 2.1028037383177567e-05, - "loss": 6.653, - "step": 1700 + "learning_rate": 3.75e-05, + "loss": 6.6788, + "step": 848 }, { "epoch": 0.95, - "learning_rate": 2.0794392523364484e-05, - "loss": 7.0506, - "step": 1701 + "learning_rate": 3.6734693877551016e-05, + "loss": 6.4647, + "step": 849 }, { "epoch": 0.95, - "learning_rate": 2.0560747663551397e-05, - "loss": 7.0843, - "step": 1702 + "learning_rate": 3.5969387755102035e-05, + "loss": 6.2421, + "step": 850 }, { "epoch": 0.95, - "learning_rate": 2.0327102803738314e-05, - "loss": 6.9721, - "step": 1703 - }, - { - "epoch": 0.96, - "learning_rate": 2.009345794392523e-05, - "loss": 7.1, - "step": 1704 - }, - { - "epoch": 0.96, - "learning_rate": 1.9859813084112145e-05, - "loss": 6.9992, - "step": 1705 - }, - { - "epoch": 0.96, - "learning_rate": 1.9626168224299062e-05, - "loss": 6.9213, - "step": 1706 - }, - { - "epoch": 0.96, - "learning_rate": 1.939252336448598e-05, - "loss": 7.0191, - "step": 1707 - }, - { - "epoch": 0.96, - "learning_rate": 1.9158878504672893e-05, - "loss": 7.0637, - "step": 1708 - }, - { - "epoch": 0.96, - "learning_rate": 1.892523364485981e-05, - "loss": 7.1791, - "step": 1709 - }, - { - "epoch": 0.96, - "learning_rate": 1.8691588785046727e-05, - "loss": 7.2129, - "step": 1710 - }, - { - "epoch": 0.96, - "learning_rate": 1.8457943925233644e-05, - "loss": 6.7878, - "step": 1711 - }, - { - "epoch": 0.96, - "learning_rate": 1.8224299065420557e-05, - "loss": 6.9851, - "step": 1712 + "learning_rate": 3.520408163265306e-05, + "loss": 6.8049, + "step": 851 }, { "epoch": 0.96, - "learning_rate": 1.7990654205607474e-05, - "loss": 7.1056, - "step": 1713 + "learning_rate": 3.443877551020408e-05, + "loss": 6.6672, + "step": 852 }, { "epoch": 0.96, - "learning_rate": 1.775700934579439e-05, - "loss": 7.0482, - "step": 1714 + "learning_rate": 3.36734693877551e-05, + "loss": 6.9027, + "step": 853 }, { "epoch": 0.96, - "learning_rate": 1.7523364485981305e-05, - "loss": 6.8617, - "step": 1715 + "learning_rate": 3.290816326530612e-05, + "loss": 6.8127, + "step": 854 }, { "epoch": 0.96, - "learning_rate": 1.7289719626168222e-05, - "loss": 6.9986, - "step": 1716 + "learning_rate": 3.214285714285714e-05, + "loss": 6.7657, + "step": 855 }, { "epoch": 0.96, - "learning_rate": 1.705607476635514e-05, - "loss": 7.1376, - "step": 1717 + "learning_rate": 3.137755102040816e-05, + "loss": 6.8388, + "step": 856 }, { "epoch": 0.96, - "learning_rate": 1.6822429906542053e-05, - "loss": 7.1549, - "step": 1718 + "learning_rate": 3.061224489795918e-05, + "loss": 6.8636, + "step": 857 }, { "epoch": 0.96, - "learning_rate": 1.658878504672897e-05, - "loss": 6.8296, - "step": 1719 + "learning_rate": 2.98469387755102e-05, + "loss": 6.7972, + "step": 858 }, { "epoch": 0.96, - "learning_rate": 1.6355140186915887e-05, - "loss": 7.1262, - "step": 1720 + "learning_rate": 2.908163265306122e-05, + "loss": 6.6306, + "step": 859 }, { "epoch": 0.96, - "learning_rate": 1.61214953271028e-05, - "loss": 6.819, - "step": 1721 - }, - { - "epoch": 0.97, - "learning_rate": 1.5887850467289717e-05, - "loss": 6.9176, - "step": 1722 - }, - { - "epoch": 0.97, - "learning_rate": 1.5654205607476634e-05, - "loss": 7.0317, - "step": 1723 - }, - { - "epoch": 0.97, - "learning_rate": 1.5420560747663548e-05, - "loss": 7.0113, - "step": 1724 - }, - { - "epoch": 0.97, - "learning_rate": 1.5186915887850465e-05, - "loss": 6.9838, - "step": 1725 - }, - { - "epoch": 0.97, - "learning_rate": 1.4953271028037382e-05, - "loss": 6.9093, - "step": 1726 - }, - { - "epoch": 0.97, - "learning_rate": 1.4719626168224297e-05, - "loss": 7.0229, - "step": 1727 - }, - { - "epoch": 0.97, - "learning_rate": 1.4485981308411213e-05, - "loss": 7.1636, - "step": 1728 - }, - { - "epoch": 0.97, - "learning_rate": 1.425233644859813e-05, - "loss": 7.0861, - "step": 1729 - }, - { - "epoch": 0.97, - "learning_rate": 1.4018691588785045e-05, - "loss": 7.0024, - "step": 1730 + "learning_rate": 2.831632653061224e-05, + "loss": 6.7981, + "step": 860 }, { "epoch": 0.97, - "learning_rate": 1.378504672897196e-05, - "loss": 6.7001, - "step": 1731 + "learning_rate": 2.7551020408163265e-05, + "loss": 6.8927, + "step": 861 }, { "epoch": 0.97, - "learning_rate": 1.3551401869158877e-05, - "loss": 6.878, - "step": 1732 + "learning_rate": 2.6785714285714284e-05, + "loss": 6.7818, + "step": 862 }, { "epoch": 0.97, - "learning_rate": 1.3317757009345793e-05, - "loss": 6.8191, - "step": 1733 + "learning_rate": 2.6020408163265303e-05, + "loss": 6.7381, + "step": 863 }, { "epoch": 0.97, - "learning_rate": 1.3084112149532708e-05, - "loss": 6.9826, - "step": 1734 + "learning_rate": 2.5255102040816323e-05, + "loss": 6.9252, + "step": 864 }, { "epoch": 0.97, - "learning_rate": 1.2850467289719625e-05, - "loss": 7.0692, - "step": 1735 + "learning_rate": 2.448979591836734e-05, + "loss": 6.8262, + "step": 865 }, { "epoch": 0.97, - "learning_rate": 1.261682242990654e-05, - "loss": 7.0486, - "step": 1736 + "learning_rate": 2.3724489795918367e-05, + "loss": 6.657, + "step": 866 }, { "epoch": 0.97, - "learning_rate": 1.2383177570093456e-05, - "loss": 6.696, - "step": 1737 + "learning_rate": 2.2959183673469387e-05, + "loss": 6.8904, + "step": 867 }, { "epoch": 0.97, - "learning_rate": 1.2149532710280373e-05, - "loss": 6.9566, - "step": 1738 + "learning_rate": 2.2193877551020406e-05, + "loss": 6.6568, + "step": 868 }, { "epoch": 0.97, - "learning_rate": 1.1915887850467288e-05, - "loss": 6.9232, - "step": 1739 - }, - { - "epoch": 0.98, - "learning_rate": 1.1682242990654203e-05, - "loss": 6.9279, - "step": 1740 - }, - { - "epoch": 0.98, - "learning_rate": 1.144859813084112e-05, - "loss": 6.8637, - "step": 1741 - }, - { - "epoch": 0.98, - "learning_rate": 1.1214953271028036e-05, - "loss": 6.9554, - "step": 1742 - }, - { - "epoch": 0.98, - "learning_rate": 1.0981308411214951e-05, - "loss": 6.9159, - "step": 1743 - }, - { - "epoch": 0.98, - "learning_rate": 1.0747663551401868e-05, - "loss": 6.6131, - "step": 1744 - }, - { - "epoch": 0.98, - "learning_rate": 1.0514018691588783e-05, - "loss": 6.7268, - "step": 1745 - }, - { - "epoch": 0.98, - "learning_rate": 1.0280373831775699e-05, - "loss": 6.7947, - "step": 1746 - }, - { - "epoch": 0.98, - "learning_rate": 1.0046728971962616e-05, - "loss": 6.9218, - "step": 1747 - }, - { - "epoch": 0.98, - "learning_rate": 9.813084112149531e-06, - "loss": 6.6222, - "step": 1748 + "learning_rate": 2.1428571428571425e-05, + "loss": 6.7322, + "step": 869 }, { "epoch": 0.98, - "learning_rate": 9.579439252336446e-06, - "loss": 6.6173, - "step": 1749 + "learning_rate": 2.0663265306122444e-05, + "loss": 6.7062, + "step": 870 }, { "epoch": 0.98, - "learning_rate": 9.345794392523363e-06, - "loss": 6.6376, - "step": 1750 + "learning_rate": 1.989795918367347e-05, + "loss": 6.6488, + "step": 871 }, { "epoch": 0.98, - "learning_rate": 9.112149532710279e-06, - "loss": 6.9141, - "step": 1751 + "learning_rate": 1.913265306122449e-05, + "loss": 6.7146, + "step": 872 }, { "epoch": 0.98, - "learning_rate": 8.878504672897196e-06, - "loss": 6.8905, - "step": 1752 + "learning_rate": 1.8367346938775508e-05, + "loss": 6.7157, + "step": 873 }, { "epoch": 0.98, - "learning_rate": 8.644859813084111e-06, - "loss": 7.1515, - "step": 1753 + "learning_rate": 1.760204081632653e-05, + "loss": 6.7805, + "step": 874 }, { "epoch": 0.98, - "learning_rate": 8.411214953271026e-06, - "loss": 6.8692, - "step": 1754 + "learning_rate": 1.683673469387755e-05, + "loss": 6.8813, + "step": 875 }, { "epoch": 0.98, - "learning_rate": 8.177570093457943e-06, - "loss": 7.1454, - "step": 1755 + "learning_rate": 1.607142857142857e-05, + "loss": 6.6575, + "step": 876 }, { "epoch": 0.98, - "learning_rate": 7.943925233644859e-06, - "loss": 6.8043, - "step": 1756 + "learning_rate": 1.530612244897959e-05, + "loss": 6.5684, + "step": 877 }, { "epoch": 0.98, - "learning_rate": 7.710280373831774e-06, - "loss": 6.838, - "step": 1757 - }, - { - "epoch": 0.99, - "learning_rate": 7.476635514018691e-06, - "loss": 6.8728, - "step": 1758 - }, - { - "epoch": 0.99, - "learning_rate": 7.242990654205606e-06, - "loss": 7.1731, - "step": 1759 - }, - { - "epoch": 0.99, - "learning_rate": 7.0093457943925225e-06, - "loss": 6.9907, - "step": 1760 - }, - { - "epoch": 0.99, - "learning_rate": 6.775700934579439e-06, - "loss": 7.0496, - "step": 1761 - }, - { - "epoch": 0.99, - "learning_rate": 6.542056074766354e-06, - "loss": 6.8904, - "step": 1762 - }, - { - "epoch": 0.99, - "learning_rate": 6.30841121495327e-06, - "loss": 6.9486, - "step": 1763 - }, - { - "epoch": 0.99, - "learning_rate": 6.074766355140186e-06, - "loss": 7.0383, - "step": 1764 - }, - { - "epoch": 0.99, - "learning_rate": 5.841121495327102e-06, - "loss": 6.803, - "step": 1765 - }, - { - "epoch": 0.99, - "learning_rate": 5.607476635514018e-06, - "loss": 7.0063, - "step": 1766 + "learning_rate": 1.454081632653061e-05, + "loss": 6.6932, + "step": 878 }, { "epoch": 0.99, - "learning_rate": 5.373831775700934e-06, - "loss": 6.819, - "step": 1767 + "learning_rate": 1.3775510204081633e-05, + "loss": 6.7408, + "step": 879 }, { "epoch": 0.99, - "learning_rate": 5.140186915887849e-06, - "loss": 6.7526, - "step": 1768 + "learning_rate": 1.3010204081632652e-05, + "loss": 6.8176, + "step": 880 }, { "epoch": 0.99, - "learning_rate": 4.9065420560747655e-06, - "loss": 6.8813, - "step": 1769 + "learning_rate": 1.224489795918367e-05, + "loss": 6.5252, + "step": 881 }, { "epoch": 0.99, - "learning_rate": 4.672897196261682e-06, - "loss": 6.8935, - "step": 1770 + "learning_rate": 1.1479591836734693e-05, + "loss": 6.7663, + "step": 882 }, { "epoch": 0.99, - "learning_rate": 4.439252336448598e-06, - "loss": 7.0451, - "step": 1771 + "learning_rate": 1.0714285714285712e-05, + "loss": 6.6353, + "step": 883 }, { "epoch": 0.99, - "learning_rate": 4.205607476635513e-06, - "loss": 6.7961, - "step": 1772 + "learning_rate": 9.948979591836735e-06, + "loss": 6.6783, + "step": 884 }, { "epoch": 0.99, - "learning_rate": 3.971962616822429e-06, - "loss": 6.9412, - "step": 1773 + "learning_rate": 9.183673469387754e-06, + "loss": 6.6815, + "step": 885 }, { "epoch": 0.99, - "learning_rate": 3.7383177570093455e-06, - "loss": 6.882, - "step": 1774 + "learning_rate": 8.418367346938775e-06, + "loss": 6.7086, + "step": 886 }, { "epoch": 0.99, - "learning_rate": 3.5046728971962613e-06, - "loss": 6.7775, - "step": 1775 - }, - { - "epoch": 1.0, - "learning_rate": 3.271028037383177e-06, - "loss": 6.9474, - "step": 1776 - }, - { - "epoch": 1.0, - "learning_rate": 3.037383177570093e-06, - "loss": 6.7826, - "step": 1777 - }, - { - "epoch": 1.0, - "learning_rate": 2.803738317757009e-06, - "loss": 7.0686, - "step": 1778 - }, - { - "epoch": 1.0, - "learning_rate": 2.5700934579439247e-06, - "loss": 7.0071, - "step": 1779 + "learning_rate": 7.653061224489796e-06, + "loss": 6.4744, + "step": 887 }, { "epoch": 1.0, - "learning_rate": 2.336448598130841e-06, - "loss": 6.8633, - "step": 1780 + "learning_rate": 6.887755102040816e-06, + "loss": 6.642, + "step": 888 }, { "epoch": 1.0, - "learning_rate": 2.1028037383177566e-06, - "loss": 6.7112, - "step": 1781 + "learning_rate": 6.122448979591835e-06, + "loss": 6.4769, + "step": 889 }, { "epoch": 1.0, - "learning_rate": 1.8691588785046728e-06, - "loss": 6.9282, - "step": 1782 + "learning_rate": 5.357142857142856e-06, + "loss": 6.4305, + "step": 890 }, { "epoch": 1.0, - "learning_rate": 1.6355140186915885e-06, - "loss": 6.451, - "step": 1783 + "learning_rate": 4.591836734693877e-06, + "loss": 6.3352, + "step": 891 }, { "epoch": 1.0, - "learning_rate": 1.4018691588785045e-06, - "loss": 6.0712, - "step": 1784 + "learning_rate": 3.826530612244898e-06, + "loss": 6.0518, + "step": 892 }, { "epoch": 1.0, - "step": 1784, + "step": 892, "total_flos": 0.0, - "train_loss": 7.010292288433811, - "train_runtime": 7556.1894, - "train_samples_per_second": 3.777, - "train_steps_per_second": 0.236 + "train_loss": 6.883699467363914, + "train_runtime": 5554.433, + "train_samples_per_second": 5.138, + "train_steps_per_second": 0.161 } ], - "max_steps": 1784, + "max_steps": 892, "num_train_epochs": 1, "total_flos": 0.0, "trial_name": null,