diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -1,8 +1,8 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 0.9995796553173603, - "global_step": 1189, + "epoch": 0.9987389659520807, + "global_step": 594, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, @@ -10,7166 +10,3587 @@ { "epoch": 0.0, "learning_rate": 0.0, - "loss": 4.878, + "loss": 4.8251, "step": 1 }, { "epoch": 0.0, - "learning_rate": 0.0, - "loss": 5.0676, + "learning_rate": 6e-07, + "loss": 4.9877, "step": 2 }, { - "epoch": 0.0, - "learning_rate": 6e-07, - "loss": 5.0744, + "epoch": 0.01, + "learning_rate": 1.2e-06, + "loss": 4.8754, "step": 3 }, { - "epoch": 0.0, - "learning_rate": 1.2e-06, - "loss": 4.8887, + "epoch": 0.01, + "learning_rate": 1.8e-06, + "loss": 4.8216, "step": 4 }, { - "epoch": 0.0, - "learning_rate": 1.8e-06, - "loss": 4.6155, + "epoch": 0.01, + "learning_rate": 2.4e-06, + "loss": 4.7633, "step": 5 }, { "epoch": 0.01, - "learning_rate": 2.4e-06, - "loss": 4.9689, + "learning_rate": 2.9999999999999997e-06, + "loss": 4.7562, "step": 6 }, { "epoch": 0.01, - "learning_rate": 2.9999999999999997e-06, - "loss": 4.7161, + "learning_rate": 3.6e-06, + "loss": 4.8158, "step": 7 }, { "epoch": 0.01, - "learning_rate": 3.6e-06, - "loss": 4.6837, + "learning_rate": 4.2e-06, + "loss": 4.6936, "step": 8 }, { - "epoch": 0.01, - "learning_rate": 3.6e-06, - "loss": 4.8697, + "epoch": 0.02, + "learning_rate": 4.8e-06, + "loss": 4.7747, "step": 9 }, { - "epoch": 0.01, - "learning_rate": 4.2e-06, - "loss": 4.7117, + "epoch": 0.02, + "learning_rate": 5.399999999999999e-06, + "loss": 4.6495, "step": 10 }, { - "epoch": 0.01, - "learning_rate": 4.8e-06, - "loss": 4.7449, + "epoch": 0.02, + "learning_rate": 5.999999999999999e-06, + "loss": 4.6263, "step": 11 }, { - "epoch": 0.01, - "learning_rate": 5.399999999999999e-06, - "loss": 4.5915, + "epoch": 0.02, + "learning_rate": 6.599999999999999e-06, + "loss": 4.6435, "step": 12 }, { - "epoch": 0.01, - "learning_rate": 5.999999999999999e-06, - "loss": 4.8094, + "epoch": 0.02, + "learning_rate": 7.2e-06, + "loss": 4.5393, "step": 13 }, { - "epoch": 0.01, - "learning_rate": 6.599999999999999e-06, - "loss": 4.5704, + "epoch": 0.02, + "learning_rate": 7.799999999999998e-06, + "loss": 4.4934, "step": 14 }, { - "epoch": 0.01, - "learning_rate": 7.2e-06, - "loss": 4.4314, + "epoch": 0.03, + "learning_rate": 8.4e-06, + "loss": 4.3827, "step": 15 }, { - "epoch": 0.01, - "learning_rate": 7.799999999999998e-06, - "loss": 4.6685, + "epoch": 0.03, + "learning_rate": 8.999999999999999e-06, + "loss": 4.528, "step": 16 }, { - "epoch": 0.01, - "learning_rate": 8.4e-06, - "loss": 4.584, + "epoch": 0.03, + "learning_rate": 9.6e-06, + "loss": 4.5121, "step": 17 }, { - "epoch": 0.02, - "learning_rate": 8.999999999999999e-06, - "loss": 4.6994, + "epoch": 0.03, + "learning_rate": 1.02e-05, + "loss": 4.5744, "step": 18 }, { - "epoch": 0.02, - "learning_rate": 9.6e-06, - "loss": 4.44, + "epoch": 0.03, + "learning_rate": 1.0799999999999998e-05, + "loss": 4.3183, "step": 19 }, { - "epoch": 0.02, - "learning_rate": 1.02e-05, - "loss": 4.6258, + "epoch": 0.03, + "learning_rate": 1.14e-05, + "loss": 4.5438, "step": 20 }, { - "epoch": 0.02, - "learning_rate": 1.0799999999999998e-05, - "loss": 4.5593, + "epoch": 0.04, + "learning_rate": 1.1999999999999999e-05, + "loss": 4.4622, "step": 21 }, { - "epoch": 0.02, - "learning_rate": 1.14e-05, - "loss": 4.4087, + "epoch": 0.04, + "learning_rate": 1.26e-05, + "loss": 4.4327, "step": 22 }, { - "epoch": 0.02, - "learning_rate": 1.1999999999999999e-05, - "loss": 4.5185, + "epoch": 0.04, + "learning_rate": 1.3199999999999997e-05, + "loss": 4.4276, "step": 23 }, { - "epoch": 0.02, - "learning_rate": 1.26e-05, - "loss": 4.5389, + "epoch": 0.04, + "learning_rate": 1.3799999999999998e-05, + "loss": 4.4719, "step": 24 }, { - "epoch": 0.02, - "learning_rate": 1.3199999999999997e-05, - "loss": 4.3627, + "epoch": 0.04, + "learning_rate": 1.44e-05, + "loss": 4.2464, "step": 25 }, { - "epoch": 0.02, - "learning_rate": 1.3799999999999998e-05, - "loss": 4.5219, + "epoch": 0.04, + "learning_rate": 1.4999999999999999e-05, + "loss": 4.664, "step": 26 }, { - "epoch": 0.02, - "learning_rate": 1.44e-05, - "loss": 4.3728, + "epoch": 0.05, + "learning_rate": 1.5599999999999996e-05, + "loss": 4.3188, "step": 27 }, { - "epoch": 0.02, - "learning_rate": 1.4999999999999999e-05, - "loss": 4.1913, + "epoch": 0.05, + "learning_rate": 1.6199999999999997e-05, + "loss": 4.202, "step": 28 }, { - "epoch": 0.02, - "learning_rate": 1.5599999999999996e-05, - "loss": 4.2714, + "epoch": 0.05, + "learning_rate": 1.68e-05, + "loss": 4.226, "step": 29 }, { - "epoch": 0.03, - "learning_rate": 1.6199999999999997e-05, - "loss": 4.3536, + "epoch": 0.05, + "learning_rate": 1.74e-05, + "loss": 4.3916, "step": 30 }, { - "epoch": 0.03, - "learning_rate": 1.68e-05, - "loss": 4.4772, + "epoch": 0.05, + "learning_rate": 1.7999999999999997e-05, + "loss": 4.4153, "step": 31 }, { - "epoch": 0.03, - "learning_rate": 1.74e-05, - "loss": 4.2838, + "epoch": 0.05, + "learning_rate": 1.8599999999999998e-05, + "loss": 4.2902, "step": 32 }, { - "epoch": 0.03, - "learning_rate": 1.7999999999999997e-05, - "loss": 4.3969, + "epoch": 0.06, + "learning_rate": 1.92e-05, + "loss": 4.2797, "step": 33 }, { - "epoch": 0.03, - "learning_rate": 1.8599999999999998e-05, - "loss": 4.1069, + "epoch": 0.06, + "learning_rate": 1.98e-05, + "loss": 4.2387, "step": 34 }, { - "epoch": 0.03, - "learning_rate": 1.92e-05, - "loss": 4.3024, + "epoch": 0.06, + "learning_rate": 2.04e-05, + "loss": 4.2934, "step": 35 }, { - "epoch": 0.03, - "learning_rate": 1.98e-05, - "loss": 4.6005, + "epoch": 0.06, + "learning_rate": 2.1e-05, + "loss": 4.4521, "step": 36 }, { - "epoch": 0.03, - "learning_rate": 2.04e-05, - "loss": 4.4158, + "epoch": 0.06, + "learning_rate": 2.1599999999999996e-05, + "loss": 4.2516, "step": 37 }, { - "epoch": 0.03, - "learning_rate": 2.1e-05, - "loss": 4.2882, + "epoch": 0.06, + "learning_rate": 2.2199999999999998e-05, + "loss": 4.4088, "step": 38 }, { - "epoch": 0.03, - "learning_rate": 2.1599999999999996e-05, - "loss": 4.7542, + "epoch": 0.07, + "learning_rate": 2.28e-05, + "loss": 4.5498, "step": 39 }, { - "epoch": 0.03, - "learning_rate": 2.2199999999999998e-05, - "loss": 4.4229, + "epoch": 0.07, + "learning_rate": 2.34e-05, + "loss": 4.2121, "step": 40 }, { - "epoch": 0.03, - "learning_rate": 2.28e-05, - "loss": 4.2078, + "epoch": 0.07, + "learning_rate": 2.3999999999999997e-05, + "loss": 4.2337, "step": 41 }, { - "epoch": 0.04, - "learning_rate": 2.34e-05, - "loss": 4.2834, + "epoch": 0.07, + "learning_rate": 2.4599999999999998e-05, + "loss": 4.4752, "step": 42 }, { - "epoch": 0.04, - "learning_rate": 2.3999999999999997e-05, - "loss": 4.2819, + "epoch": 0.07, + "learning_rate": 2.52e-05, + "loss": 4.1934, "step": 43 }, { - "epoch": 0.04, - "learning_rate": 2.4599999999999998e-05, - "loss": 4.309, + "epoch": 0.07, + "learning_rate": 2.5799999999999997e-05, + "loss": 4.3419, "step": 44 }, { - "epoch": 0.04, - "learning_rate": 2.52e-05, - "loss": 4.5016, + "epoch": 0.08, + "learning_rate": 2.6399999999999995e-05, + "loss": 4.4241, "step": 45 }, { - "epoch": 0.04, - "learning_rate": 2.5799999999999997e-05, - "loss": 4.4434, + "epoch": 0.08, + "learning_rate": 2.6999999999999996e-05, + "loss": 4.4737, "step": 46 }, { - "epoch": 0.04, - "learning_rate": 2.6399999999999995e-05, - "loss": 4.6236, + "epoch": 0.08, + "learning_rate": 2.7599999999999997e-05, + "loss": 4.3846, "step": 47 }, { - "epoch": 0.04, - "learning_rate": 2.6999999999999996e-05, - "loss": 4.8033, + "epoch": 0.08, + "learning_rate": 2.8199999999999998e-05, + "loss": 4.7292, "step": 48 }, { - "epoch": 0.04, - "learning_rate": 2.7599999999999997e-05, - "loss": 4.972, + "epoch": 0.08, + "learning_rate": 2.8199999999999998e-05, + "loss": 4.7684, "step": 49 }, { - "epoch": 0.04, - "learning_rate": 2.7599999999999997e-05, - "loss": 4.7127, + "epoch": 0.08, + "learning_rate": 2.88e-05, + "loss": 4.4745, "step": 50 }, { - "epoch": 0.04, - "learning_rate": 2.8199999999999998e-05, - "loss": 4.2836, + "epoch": 0.09, + "learning_rate": 2.94e-05, + "loss": 4.193, "step": 51 }, { - "epoch": 0.04, - "learning_rate": 2.88e-05, - "loss": 4.2811, + "epoch": 0.09, + "learning_rate": 2.9999999999999997e-05, + "loss": 4.2115, "step": 52 }, { - "epoch": 0.04, - "learning_rate": 2.94e-05, - "loss": 4.1493, + "epoch": 0.09, + "learning_rate": 3.06e-05, + "loss": 4.2324, "step": 53 }, { - "epoch": 0.05, - "learning_rate": 2.9999999999999997e-05, - "loss": 4.2219, + "epoch": 0.09, + "learning_rate": 3.119999999999999e-05, + "loss": 4.3067, "step": 54 }, { - "epoch": 0.05, - "learning_rate": 3.06e-05, - "loss": 4.3674, + "epoch": 0.09, + "learning_rate": 3.1799999999999994e-05, + "loss": 4.167, "step": 55 }, { - "epoch": 0.05, - "learning_rate": 3.119999999999999e-05, - "loss": 4.1651, + "epoch": 0.09, + "learning_rate": 3.2399999999999995e-05, + "loss": 4.3703, "step": 56 }, { - "epoch": 0.05, - "learning_rate": 3.1799999999999994e-05, - "loss": 4.4474, + "epoch": 0.1, + "learning_rate": 3.2999999999999996e-05, + "loss": 4.1812, "step": 57 }, { - "epoch": 0.05, - "learning_rate": 3.2399999999999995e-05, - "loss": 4.3298, + "epoch": 0.1, + "learning_rate": 3.36e-05, + "loss": 4.4778, "step": 58 }, { - "epoch": 0.05, - "learning_rate": 3.2999999999999996e-05, - "loss": 4.2132, + "epoch": 0.1, + "learning_rate": 3.42e-05, + "loss": 4.2518, "step": 59 }, { - "epoch": 0.05, - "learning_rate": 3.36e-05, - "loss": 4.3708, + "epoch": 0.1, + "learning_rate": 3.48e-05, + "loss": 4.0458, "step": 60 }, { - "epoch": 0.05, - "learning_rate": 3.42e-05, - "loss": 4.1803, + "epoch": 0.1, + "learning_rate": 3.539999999999999e-05, + "loss": 4.2996, "step": 61 }, { - "epoch": 0.05, - "learning_rate": 3.48e-05, - "loss": 4.3545, + "epoch": 0.1, + "learning_rate": 3.5999999999999994e-05, + "loss": 4.3556, "step": 62 }, { - "epoch": 0.05, - "learning_rate": 3.539999999999999e-05, - "loss": 4.2327, + "epoch": 0.11, + "learning_rate": 3.6599999999999995e-05, + "loss": 4.2308, "step": 63 }, { - "epoch": 0.05, - "learning_rate": 3.5999999999999994e-05, - "loss": 4.1038, + "epoch": 0.11, + "learning_rate": 3.7199999999999996e-05, + "loss": 4.2086, "step": 64 }, { - "epoch": 0.05, - "learning_rate": 3.6599999999999995e-05, - "loss": 4.1128, + "epoch": 0.11, + "learning_rate": 3.78e-05, + "loss": 4.2349, "step": 65 }, { - "epoch": 0.06, - "learning_rate": 3.7199999999999996e-05, - "loss": 4.2349, + "epoch": 0.11, + "learning_rate": 3.84e-05, + "loss": 4.1813, "step": 66 }, { - "epoch": 0.06, - "learning_rate": 3.78e-05, - "loss": 4.3056, + "epoch": 0.11, + "learning_rate": 3.9e-05, + "loss": 4.1444, "step": 67 }, { - "epoch": 0.06, - "learning_rate": 3.84e-05, - "loss": 4.2165, + "epoch": 0.11, + "learning_rate": 3.96e-05, + "loss": 4.1388, "step": 68 }, { - "epoch": 0.06, - "learning_rate": 3.9e-05, - "loss": 4.2028, + "epoch": 0.12, + "learning_rate": 4.02e-05, + "loss": 4.2267, "step": 69 }, { - "epoch": 0.06, - "learning_rate": 3.96e-05, - "loss": 4.2546, + "epoch": 0.12, + "learning_rate": 4.08e-05, + "loss": 4.2124, "step": 70 }, { - "epoch": 0.06, - "learning_rate": 4.02e-05, - "loss": 4.4175, + "epoch": 0.12, + "learning_rate": 4.14e-05, + "loss": 4.3128, "step": 71 }, { - "epoch": 0.06, - "learning_rate": 4.08e-05, - "loss": 4.3118, + "epoch": 0.12, + "learning_rate": 4.2e-05, + "loss": 4.2821, "step": 72 }, { - "epoch": 0.06, - "learning_rate": 4.14e-05, - "loss": 4.2295, + "epoch": 0.12, + "learning_rate": 4.259999999999999e-05, + "loss": 4.0836, "step": 73 }, { - "epoch": 0.06, - "learning_rate": 4.2e-05, - "loss": 4.2973, + "epoch": 0.12, + "learning_rate": 4.319999999999999e-05, + "loss": 4.2887, "step": 74 }, { - "epoch": 0.06, - "learning_rate": 4.259999999999999e-05, - "loss": 4.1686, + "epoch": 0.13, + "learning_rate": 4.3799999999999994e-05, + "loss": 4.2636, "step": 75 }, { - "epoch": 0.06, - "learning_rate": 4.319999999999999e-05, - "loss": 4.5328, + "epoch": 0.13, + "learning_rate": 4.4399999999999995e-05, + "loss": 4.2405, "step": 76 }, { - "epoch": 0.06, - "learning_rate": 4.3799999999999994e-05, - "loss": 4.3378, + "epoch": 0.13, + "learning_rate": 4.4999999999999996e-05, + "loss": 4.2182, "step": 77 }, { - "epoch": 0.07, - "learning_rate": 4.4399999999999995e-05, - "loss": 4.1686, + "epoch": 0.13, + "learning_rate": 4.56e-05, + "loss": 4.2002, "step": 78 }, { - "epoch": 0.07, - "learning_rate": 4.4999999999999996e-05, - "loss": 4.1852, + "epoch": 0.13, + "learning_rate": 4.62e-05, + "loss": 4.1308, "step": 79 }, { - "epoch": 0.07, - "learning_rate": 4.56e-05, - "loss": 4.3177, + "epoch": 0.13, + "learning_rate": 4.68e-05, + "loss": 4.3264, "step": 80 }, { - "epoch": 0.07, - "learning_rate": 4.62e-05, - "loss": 4.3517, + "epoch": 0.14, + "learning_rate": 4.7399999999999993e-05, + "loss": 4.2728, "step": 81 }, { - "epoch": 0.07, - "learning_rate": 4.68e-05, - "loss": 4.2729, + "epoch": 0.14, + "learning_rate": 4.7999999999999994e-05, + "loss": 4.2262, "step": 82 }, { - "epoch": 0.07, - "learning_rate": 4.7399999999999993e-05, - "loss": 4.2086, + "epoch": 0.14, + "learning_rate": 4.8599999999999995e-05, + "loss": 4.3323, "step": 83 }, { - "epoch": 0.07, - "learning_rate": 4.7999999999999994e-05, - "loss": 4.3288, + "epoch": 0.14, + "learning_rate": 4.9199999999999997e-05, + "loss": 4.2886, "step": 84 }, { - "epoch": 0.07, - "learning_rate": 4.8599999999999995e-05, - "loss": 4.2414, + "epoch": 0.14, + "learning_rate": 4.98e-05, + "loss": 4.2094, "step": 85 }, { - "epoch": 0.07, - "learning_rate": 4.9199999999999997e-05, - "loss": 4.2314, + "epoch": 0.14, + "learning_rate": 5.04e-05, + "loss": 4.1616, "step": 86 }, { - "epoch": 0.07, - "learning_rate": 4.98e-05, - "loss": 4.119, + "epoch": 0.15, + "learning_rate": 5.1e-05, + "loss": 4.4014, "step": 87 }, { - "epoch": 0.07, - "learning_rate": 5.04e-05, - "loss": 4.3444, + "epoch": 0.15, + "learning_rate": 5.1599999999999994e-05, + "loss": 4.2411, "step": 88 }, { - "epoch": 0.07, - "learning_rate": 5.1e-05, - "loss": 4.4085, + "epoch": 0.15, + "learning_rate": 5.2199999999999995e-05, + "loss": 4.2378, "step": 89 }, { - "epoch": 0.08, - "learning_rate": 5.1599999999999994e-05, - "loss": 4.2942, + "epoch": 0.15, + "learning_rate": 5.279999999999999e-05, + "loss": 4.2782, "step": 90 }, { - "epoch": 0.08, - "learning_rate": 5.2199999999999995e-05, - "loss": 4.381, + "epoch": 0.15, + "learning_rate": 5.339999999999999e-05, + "loss": 4.2438, "step": 91 }, { - "epoch": 0.08, - "learning_rate": 5.279999999999999e-05, - "loss": 4.3951, + "epoch": 0.15, + "learning_rate": 5.399999999999999e-05, + "loss": 4.3678, "step": 92 }, { - "epoch": 0.08, - "learning_rate": 5.339999999999999e-05, - "loss": 4.0369, + "epoch": 0.16, + "learning_rate": 5.459999999999999e-05, + "loss": 4.2209, "step": 93 }, { - "epoch": 0.08, - "learning_rate": 5.399999999999999e-05, - "loss": 4.4064, + "epoch": 0.16, + "learning_rate": 5.519999999999999e-05, + "loss": 4.4583, "step": 94 }, { - "epoch": 0.08, - "learning_rate": 5.459999999999999e-05, - "loss": 4.5946, + "epoch": 0.16, + "learning_rate": 5.5799999999999994e-05, + "loss": 4.3543, "step": 95 }, { - "epoch": 0.08, - "learning_rate": 5.519999999999999e-05, - "loss": 4.4538, + "epoch": 0.16, + "learning_rate": 5.6399999999999995e-05, + "loss": 4.3987, "step": 96 }, { - "epoch": 0.08, - "learning_rate": 5.5799999999999994e-05, - "loss": 4.464, + "epoch": 0.16, + "learning_rate": 5.6999999999999996e-05, + "loss": 4.3792, "step": 97 }, { - "epoch": 0.08, - "learning_rate": 5.6399999999999995e-05, - "loss": 4.5133, + "epoch": 0.16, + "learning_rate": 5.76e-05, + "loss": 4.2989, "step": 98 }, { - "epoch": 0.08, - "learning_rate": 5.6999999999999996e-05, - "loss": 4.5579, + "epoch": 0.17, + "learning_rate": 5.82e-05, + "loss": 4.5377, "step": 99 }, { - "epoch": 0.08, - "learning_rate": 5.76e-05, - "loss": 4.2087, + "epoch": 0.17, + "learning_rate": 5.88e-05, + "loss": 4.1863, "step": 100 }, { - "epoch": 0.08, - "learning_rate": 5.82e-05, - "loss": 4.0703, + "epoch": 0.17, + "learning_rate": 5.94e-05, + "loss": 4.3579, "step": 101 }, { - "epoch": 0.09, - "learning_rate": 5.88e-05, - "loss": 4.4371, + "epoch": 0.17, + "learning_rate": 5.9999999999999995e-05, + "loss": 4.1935, "step": 102 }, { - "epoch": 0.09, - "learning_rate": 5.94e-05, - "loss": 4.2404, + "epoch": 0.17, + "learning_rate": 6.0599999999999996e-05, + "loss": 4.206, "step": 103 }, { - "epoch": 0.09, - "learning_rate": 5.9999999999999995e-05, - "loss": 4.3118, + "epoch": 0.17, + "learning_rate": 6.12e-05, + "loss": 4.1263, "step": 104 }, { - "epoch": 0.09, - "learning_rate": 6.0599999999999996e-05, - "loss": 4.253, + "epoch": 0.18, + "learning_rate": 6.18e-05, + "loss": 4.1788, "step": 105 }, { - "epoch": 0.09, - "learning_rate": 6.12e-05, - "loss": 4.3638, + "epoch": 0.18, + "learning_rate": 6.239999999999999e-05, + "loss": 4.1409, "step": 106 }, { - "epoch": 0.09, - "learning_rate": 6.18e-05, - "loss": 4.4452, + "epoch": 0.18, + "learning_rate": 6.299999999999999e-05, + "loss": 4.2692, "step": 107 }, { - "epoch": 0.09, - "learning_rate": 6.239999999999999e-05, - "loss": 4.4478, + "epoch": 0.18, + "learning_rate": 6.359999999999999e-05, + "loss": 4.3054, "step": 108 }, { - "epoch": 0.09, - "learning_rate": 6.299999999999999e-05, - "loss": 4.3519, + "epoch": 0.18, + "learning_rate": 6.419999999999999e-05, + "loss": 4.4333, "step": 109 }, { - "epoch": 0.09, - "learning_rate": 6.359999999999999e-05, - "loss": 4.1316, + "epoch": 0.18, + "learning_rate": 6.479999999999999e-05, + "loss": 4.2029, "step": 110 }, { - "epoch": 0.09, - "learning_rate": 6.419999999999999e-05, - "loss": 4.3832, + "epoch": 0.19, + "learning_rate": 6.539999999999999e-05, + "loss": 4.2235, "step": 111 }, { - "epoch": 0.09, - "learning_rate": 6.479999999999999e-05, - "loss": 4.4707, + "epoch": 0.19, + "learning_rate": 6.599999999999999e-05, + "loss": 4.256, "step": 112 }, { - "epoch": 0.09, - "learning_rate": 6.539999999999999e-05, - "loss": 4.2211, + "epoch": 0.19, + "learning_rate": 6.659999999999999e-05, + "loss": 4.2251, "step": 113 }, { - "epoch": 0.1, - "learning_rate": 6.599999999999999e-05, - "loss": 4.4116, + "epoch": 0.19, + "learning_rate": 6.72e-05, + "loss": 4.2755, "step": 114 }, { - "epoch": 0.1, - "learning_rate": 6.659999999999999e-05, - "loss": 4.359, + "epoch": 0.19, + "learning_rate": 6.78e-05, + "loss": 4.269, "step": 115 }, { - "epoch": 0.1, - "learning_rate": 6.72e-05, - "loss": 4.4433, + "epoch": 0.2, + "learning_rate": 6.84e-05, + "loss": 4.2454, "step": 116 }, { - "epoch": 0.1, - "learning_rate": 6.78e-05, - "loss": 4.2172, + "epoch": 0.2, + "learning_rate": 6.9e-05, + "loss": 4.3376, "step": 117 }, { - "epoch": 0.1, - "learning_rate": 6.84e-05, - "loss": 4.0784, + "epoch": 0.2, + "learning_rate": 6.96e-05, + "loss": 4.1226, "step": 118 }, { - "epoch": 0.1, - "learning_rate": 6.9e-05, - "loss": 4.1843, + "epoch": 0.2, + "learning_rate": 7.02e-05, + "loss": 4.1287, "step": 119 }, { - "epoch": 0.1, - "learning_rate": 6.96e-05, - "loss": 4.1996, + "epoch": 0.2, + "learning_rate": 7.079999999999999e-05, + "loss": 4.1267, "step": 120 }, { - "epoch": 0.1, - "learning_rate": 7.02e-05, - "loss": 4.5407, + "epoch": 0.2, + "learning_rate": 7.139999999999999e-05, + "loss": 4.2983, "step": 121 }, { - "epoch": 0.1, - "learning_rate": 7.079999999999999e-05, - "loss": 4.3013, + "epoch": 0.21, + "learning_rate": 7.199999999999999e-05, + "loss": 4.4717, "step": 122 }, { - "epoch": 0.1, - "learning_rate": 7.139999999999999e-05, - "loss": 4.3319, + "epoch": 0.21, + "learning_rate": 7.259999999999999e-05, + "loss": 4.193, "step": 123 }, { - "epoch": 0.1, - "learning_rate": 7.199999999999999e-05, - "loss": 4.422, + "epoch": 0.21, + "learning_rate": 7.319999999999999e-05, + "loss": 4.2596, "step": 124 }, { - "epoch": 0.11, - "learning_rate": 7.259999999999999e-05, - "loss": 4.1384, + "epoch": 0.21, + "learning_rate": 7.379999999999999e-05, + "loss": 4.2809, "step": 125 }, { - "epoch": 0.11, - "learning_rate": 7.319999999999999e-05, - "loss": 4.2991, + "epoch": 0.21, + "learning_rate": 7.439999999999999e-05, + "loss": 4.1158, "step": 126 }, { - "epoch": 0.11, - "learning_rate": 7.379999999999999e-05, - "loss": 4.4639, + "epoch": 0.21, + "learning_rate": 7.5e-05, + "loss": 4.1375, "step": 127 }, { - "epoch": 0.11, - "learning_rate": 7.439999999999999e-05, - "loss": 4.1505, + "epoch": 0.22, + "learning_rate": 7.56e-05, + "loss": 4.1286, "step": 128 }, { - "epoch": 0.11, - "learning_rate": 7.5e-05, - "loss": 4.106, + "epoch": 0.22, + "learning_rate": 7.62e-05, + "loss": 4.2348, "step": 129 }, { - "epoch": 0.11, - "learning_rate": 7.56e-05, - "loss": 4.3523, + "epoch": 0.22, + "learning_rate": 7.68e-05, + "loss": 4.2981, "step": 130 }, { - "epoch": 0.11, - "learning_rate": 7.62e-05, - "loss": 4.3679, + "epoch": 0.22, + "learning_rate": 7.74e-05, + "loss": 4.2467, "step": 131 }, { - "epoch": 0.11, - "learning_rate": 7.68e-05, - "loss": 4.2785, + "epoch": 0.22, + "learning_rate": 7.8e-05, + "loss": 4.2083, "step": 132 }, { - "epoch": 0.11, - "learning_rate": 7.74e-05, - "loss": 4.5031, + "epoch": 0.22, + "learning_rate": 7.86e-05, + "loss": 4.2554, "step": 133 }, { - "epoch": 0.11, - "learning_rate": 7.8e-05, - "loss": 4.3363, + "epoch": 0.23, + "learning_rate": 7.92e-05, + "loss": 4.3453, "step": 134 }, { - "epoch": 0.11, - "learning_rate": 7.86e-05, - "loss": 4.4382, + "epoch": 0.23, + "learning_rate": 7.98e-05, + "loss": 4.3405, "step": 135 }, { - "epoch": 0.11, - "learning_rate": 7.92e-05, - "loss": 4.1609, + "epoch": 0.23, + "learning_rate": 8.04e-05, + "loss": 4.2022, "step": 136 }, { - "epoch": 0.12, - "learning_rate": 7.98e-05, - "loss": 4.456, + "epoch": 0.23, + "learning_rate": 8.1e-05, + "loss": 4.182, "step": 137 }, { - "epoch": 0.12, - "learning_rate": 8.04e-05, - "loss": 4.432, + "epoch": 0.23, + "learning_rate": 8.16e-05, + "loss": 4.284, "step": 138 }, { - "epoch": 0.12, - "learning_rate": 8.1e-05, - "loss": 4.2593, + "epoch": 0.23, + "learning_rate": 8.22e-05, + "loss": 4.224, "step": 139 }, { - "epoch": 0.12, - "learning_rate": 8.16e-05, - "loss": 4.4301, + "epoch": 0.24, + "learning_rate": 8.28e-05, + "loss": 4.4008, "step": 140 }, { - "epoch": 0.12, - "learning_rate": 8.22e-05, - "loss": 4.4305, + "epoch": 0.24, + "learning_rate": 8.34e-05, + "loss": 4.2829, "step": 141 }, { - "epoch": 0.12, - "learning_rate": 8.28e-05, - "loss": 4.2845, + "epoch": 0.24, + "learning_rate": 8.4e-05, + "loss": 4.3976, "step": 142 }, { - "epoch": 0.12, - "learning_rate": 8.34e-05, - "loss": 4.1929, + "epoch": 0.24, + "learning_rate": 8.459999999999998e-05, + "loss": 4.4763, "step": 143 }, { - "epoch": 0.12, - "learning_rate": 8.4e-05, - "loss": 4.5414, + "epoch": 0.24, + "learning_rate": 8.519999999999998e-05, + "loss": 4.3198, "step": 144 }, { - "epoch": 0.12, - "learning_rate": 8.459999999999998e-05, - "loss": 4.4126, + "epoch": 0.24, + "learning_rate": 8.579999999999998e-05, + "loss": 4.3517, "step": 145 }, { - "epoch": 0.12, - "learning_rate": 8.519999999999998e-05, - "loss": 4.5683, + "epoch": 0.25, + "learning_rate": 8.639999999999999e-05, + "loss": 4.7129, "step": 146 }, { - "epoch": 0.12, - "learning_rate": 8.579999999999998e-05, - "loss": 4.4539, + "epoch": 0.25, + "learning_rate": 8.699999999999999e-05, + "loss": 4.6215, "step": 147 }, { - "epoch": 0.12, - "learning_rate": 8.639999999999999e-05, - "loss": 4.2047, + "epoch": 0.25, + "learning_rate": 8.759999999999999e-05, + "loss": 4.5996, "step": 148 }, { - "epoch": 0.13, - "learning_rate": 8.699999999999999e-05, - "loss": 4.5413, + "epoch": 0.25, + "learning_rate": 8.819999999999999e-05, + "loss": 4.6987, "step": 149 }, { - "epoch": 0.13, - "learning_rate": 8.759999999999999e-05, - "loss": 4.5694, + "epoch": 0.25, + "learning_rate": 8.879999999999999e-05, + "loss": 4.577, "step": 150 }, { - "epoch": 0.13, - "learning_rate": 8.819999999999999e-05, - "loss": 4.3819, + "epoch": 0.25, + "learning_rate": 8.939999999999999e-05, + "loss": 4.3062, "step": 151 }, { - "epoch": 0.13, - "learning_rate": 8.879999999999999e-05, - "loss": 4.3633, + "epoch": 0.26, + "learning_rate": 8.999999999999999e-05, + "loss": 4.2302, "step": 152 }, { - "epoch": 0.13, - "learning_rate": 8.939999999999999e-05, - "loss": 4.3485, + "epoch": 0.26, + "learning_rate": 9.059999999999999e-05, + "loss": 4.2216, "step": 153 }, { - "epoch": 0.13, - "learning_rate": 8.999999999999999e-05, - "loss": 4.428, + "epoch": 0.26, + "learning_rate": 9.12e-05, + "loss": 4.12, "step": 154 }, { - "epoch": 0.13, - "learning_rate": 9.059999999999999e-05, - "loss": 4.3302, + "epoch": 0.26, + "learning_rate": 9.18e-05, + "loss": 4.1832, "step": 155 }, { - "epoch": 0.13, - "learning_rate": 9.12e-05, - "loss": 4.1732, + "epoch": 0.26, + "learning_rate": 9.24e-05, + "loss": 4.1754, "step": 156 }, { - "epoch": 0.13, - "learning_rate": 9.18e-05, - "loss": 4.3756, + "epoch": 0.26, + "learning_rate": 9.3e-05, + "loss": 4.274, "step": 157 }, { - "epoch": 0.13, - "learning_rate": 9.24e-05, - "loss": 4.5436, + "epoch": 0.27, + "learning_rate": 9.36e-05, + "loss": 4.1414, "step": 158 }, { - "epoch": 0.13, - "learning_rate": 9.3e-05, - "loss": 4.2929, + "epoch": 0.27, + "learning_rate": 9.419999999999999e-05, + "loss": 4.0708, "step": 159 }, { - "epoch": 0.13, - "learning_rate": 9.36e-05, - "loss": 4.0308, + "epoch": 0.27, + "learning_rate": 9.479999999999999e-05, + "loss": 4.2012, "step": 160 }, { - "epoch": 0.14, - "learning_rate": 9.419999999999999e-05, - "loss": 4.3842, + "epoch": 0.27, + "learning_rate": 9.539999999999999e-05, + "loss": 4.2286, "step": 161 }, { - "epoch": 0.14, - "learning_rate": 9.479999999999999e-05, - "loss": 4.3992, + "epoch": 0.27, + "learning_rate": 9.599999999999999e-05, + "loss": 4.1608, "step": 162 }, { - "epoch": 0.14, - "learning_rate": 9.539999999999999e-05, - "loss": 4.3533, + "epoch": 0.27, + "learning_rate": 9.659999999999999e-05, + "loss": 4.1439, "step": 163 }, { - "epoch": 0.14, - "learning_rate": 9.599999999999999e-05, - "loss": 4.0939, + "epoch": 0.28, + "learning_rate": 9.719999999999999e-05, + "loss": 4.1617, "step": 164 }, { - "epoch": 0.14, - "learning_rate": 9.659999999999999e-05, - "loss": 4.3203, + "epoch": 0.28, + "learning_rate": 9.779999999999999e-05, + "loss": 4.2501, "step": 165 }, { - "epoch": 0.14, - "learning_rate": 9.719999999999999e-05, - "loss": 3.944, + "epoch": 0.28, + "learning_rate": 9.839999999999999e-05, + "loss": 4.322, "step": 166 }, { - "epoch": 0.14, - "learning_rate": 9.779999999999999e-05, - "loss": 4.3161, + "epoch": 0.28, + "learning_rate": 9.9e-05, + "loss": 4.1637, "step": 167 }, { - "epoch": 0.14, - "learning_rate": 9.839999999999999e-05, - "loss": 4.2308, + "epoch": 0.28, + "learning_rate": 9.96e-05, + "loss": 4.3875, "step": 168 }, { - "epoch": 0.14, - "learning_rate": 9.9e-05, - "loss": 4.3824, + "epoch": 0.28, + "learning_rate": 0.0001002, + "loss": 4.1984, "step": 169 }, { - "epoch": 0.14, - "learning_rate": 9.96e-05, - "loss": 4.2629, + "epoch": 0.29, + "learning_rate": 0.0001008, + "loss": 4.2334, "step": 170 }, { - "epoch": 0.14, - "learning_rate": 0.0001002, - "loss": 4.3181, + "epoch": 0.29, + "learning_rate": 0.0001014, + "loss": 4.0895, "step": 171 }, { - "epoch": 0.14, - "learning_rate": 0.0001008, - "loss": 4.196, + "epoch": 0.29, + "learning_rate": 0.000102, + "loss": 4.1526, "step": 172 }, { - "epoch": 0.15, - "learning_rate": 0.0001014, - "loss": 4.0841, + "epoch": 0.29, + "learning_rate": 0.0001026, + "loss": 4.1554, "step": 173 }, { - "epoch": 0.15, - "learning_rate": 0.000102, - "loss": 4.514, + "epoch": 0.29, + "learning_rate": 0.00010319999999999999, + "loss": 4.174, "step": 174 }, { - "epoch": 0.15, - "learning_rate": 0.0001026, - "loss": 4.3829, + "epoch": 0.29, + "learning_rate": 0.00010379999999999999, + "loss": 4.3213, "step": 175 }, { - "epoch": 0.15, - "learning_rate": 0.00010319999999999999, - "loss": 4.3692, + "epoch": 0.3, + "learning_rate": 0.00010439999999999999, + "loss": 4.3565, "step": 176 }, { - "epoch": 0.15, - "learning_rate": 0.00010379999999999999, - "loss": 4.4161, + "epoch": 0.3, + "learning_rate": 0.00010499999999999999, + "loss": 4.2848, "step": 177 }, { - "epoch": 0.15, - "learning_rate": 0.00010439999999999999, - "loss": 4.2987, + "epoch": 0.3, + "learning_rate": 0.00010559999999999998, + "loss": 4.1378, "step": 178 }, { - "epoch": 0.15, - "learning_rate": 0.00010499999999999999, - "loss": 4.3005, + "epoch": 0.3, + "learning_rate": 0.00010619999999999998, + "loss": 4.2663, "step": 179 }, { - "epoch": 0.15, - "learning_rate": 0.00010559999999999998, - "loss": 4.4786, + "epoch": 0.3, + "learning_rate": 0.00010679999999999998, + "loss": 4.1552, "step": 180 }, { - "epoch": 0.15, - "learning_rate": 0.00010619999999999998, - "loss": 4.4301, + "epoch": 0.3, + "learning_rate": 0.00010739999999999998, + "loss": 4.4146, "step": 181 }, { - "epoch": 0.15, - "learning_rate": 0.00010679999999999998, - "loss": 4.4607, + "epoch": 0.31, + "learning_rate": 0.00010799999999999998, + "loss": 4.212, "step": 182 }, { - "epoch": 0.15, - "learning_rate": 0.00010739999999999998, - "loss": 4.3058, + "epoch": 0.31, + "learning_rate": 0.00010859999999999998, + "loss": 4.2198, "step": 183 }, { - "epoch": 0.15, - "learning_rate": 0.00010799999999999998, - "loss": 4.3893, + "epoch": 0.31, + "learning_rate": 0.00010919999999999998, + "loss": 4.0986, "step": 184 }, { - "epoch": 0.16, - "learning_rate": 0.00010859999999999998, - "loss": 4.1137, + "epoch": 0.31, + "learning_rate": 0.00010979999999999999, + "loss": 4.4803, "step": 185 }, { - "epoch": 0.16, - "learning_rate": 0.00010919999999999998, - "loss": 4.3025, + "epoch": 0.31, + "learning_rate": 0.00011039999999999999, + "loss": 4.2863, "step": 186 }, { - "epoch": 0.16, - "learning_rate": 0.00010979999999999999, - "loss": 4.5062, + "epoch": 0.31, + "learning_rate": 0.00011099999999999999, + "loss": 4.292, "step": 187 }, { - "epoch": 0.16, - "learning_rate": 0.00011039999999999999, - "loss": 4.3481, + "epoch": 0.32, + "learning_rate": 0.00011159999999999999, + "loss": 4.3438, "step": 188 }, { - "epoch": 0.16, - "learning_rate": 0.00011099999999999999, - "loss": 4.4715, + "epoch": 0.32, + "learning_rate": 0.00011219999999999999, + "loss": 4.4794, "step": 189 }, { - "epoch": 0.16, - "learning_rate": 0.00011159999999999999, - "loss": 4.1929, + "epoch": 0.32, + "learning_rate": 0.00011279999999999999, + "loss": 4.1904, "step": 190 }, { - "epoch": 0.16, - "learning_rate": 0.00011219999999999999, - "loss": 4.3975, + "epoch": 0.32, + "learning_rate": 0.00011339999999999999, + "loss": 4.4741, "step": 191 }, { - "epoch": 0.16, - "learning_rate": 0.00011279999999999999, - "loss": 4.6548, + "epoch": 0.32, + "learning_rate": 0.00011399999999999999, + "loss": 4.3827, "step": 192 }, { - "epoch": 0.16, - "learning_rate": 0.00011339999999999999, - "loss": 5.0553, + "epoch": 0.32, + "learning_rate": 0.0001146, + "loss": 4.5862, "step": 193 }, { - "epoch": 0.16, - "learning_rate": 0.00011399999999999999, - "loss": 4.3685, + "epoch": 0.33, + "learning_rate": 0.0001152, + "loss": 4.5182, "step": 194 }, { - "epoch": 0.16, - "learning_rate": 0.0001146, - "loss": 4.5626, + "epoch": 0.33, + "learning_rate": 0.0001158, + "loss": 4.306, "step": 195 }, { - "epoch": 0.16, - "learning_rate": 0.0001152, - "loss": 4.8323, + "epoch": 0.33, + "learning_rate": 0.0001164, + "loss": 4.3188, "step": 196 }, { - "epoch": 0.17, - "learning_rate": 0.0001158, - "loss": 4.5394, + "epoch": 0.33, + "learning_rate": 0.000117, + "loss": 4.7074, "step": 197 }, { - "epoch": 0.17, - "learning_rate": 0.0001164, - "loss": 4.5841, + "epoch": 0.33, + "learning_rate": 0.0001176, + "loss": 4.5491, "step": 198 }, { - "epoch": 0.17, - "learning_rate": 0.000117, - "loss": 4.652, + "epoch": 0.33, + "learning_rate": 0.0001182, + "loss": 4.8521, "step": 199 }, { - "epoch": 0.17, - "learning_rate": 0.0001176, - "loss": 4.4321, + "epoch": 0.34, + "learning_rate": 0.0001188, + "loss": 4.7298, "step": 200 }, { - "epoch": 0.17, - "learning_rate": 0.0001182, - "loss": 4.5498, + "epoch": 0.34, + "learning_rate": 0.0001194, + "loss": 4.2078, "step": 201 }, { - "epoch": 0.17, - "learning_rate": 0.0001188, - "loss": 4.3639, + "epoch": 0.34, + "learning_rate": 0.00011999999999999999, + "loss": 4.2913, "step": 202 }, { - "epoch": 0.17, - "learning_rate": 0.0001194, - "loss": 4.4626, + "epoch": 0.34, + "learning_rate": 0.00012059999999999999, + "loss": 4.4183, "step": 203 }, { - "epoch": 0.17, - "learning_rate": 0.00011999999999999999, - "loss": 4.3329, + "epoch": 0.34, + "learning_rate": 0.00012119999999999999, + "loss": 4.1983, "step": 204 }, { - "epoch": 0.17, - "learning_rate": 0.00012059999999999999, - "loss": 4.3249, + "epoch": 0.34, + "learning_rate": 0.00012179999999999999, + "loss": 4.2838, "step": 205 }, { - "epoch": 0.17, - "learning_rate": 0.00012119999999999999, - "loss": 4.318, + "epoch": 0.35, + "learning_rate": 0.0001224, + "loss": 4.2377, "step": 206 }, { - "epoch": 0.17, - "learning_rate": 0.00012179999999999999, - "loss": 4.3869, + "epoch": 0.35, + "learning_rate": 0.00012299999999999998, + "loss": 4.3051, "step": 207 }, { - "epoch": 0.17, - "learning_rate": 0.0001224, - "loss": 4.9422, + "epoch": 0.35, + "learning_rate": 0.0001236, + "loss": 4.681, "step": 208 }, { - "epoch": 0.18, - "learning_rate": 0.00012299999999999998, - "loss": 4.4169, + "epoch": 0.35, + "learning_rate": 0.00012419999999999998, + "loss": 4.3501, "step": 209 }, { - "epoch": 0.18, - "learning_rate": 0.0001236, - "loss": 4.8081, + "epoch": 0.35, + "learning_rate": 0.00012479999999999997, + "loss": 4.3868, "step": 210 }, { - "epoch": 0.18, - "learning_rate": 0.00012419999999999998, - "loss": 4.3622, + "epoch": 0.35, + "learning_rate": 0.00012539999999999999, + "loss": 4.3917, "step": 211 }, { - "epoch": 0.18, - "learning_rate": 0.00012479999999999997, - "loss": 4.4941, + "epoch": 0.36, + "learning_rate": 0.00012599999999999997, + "loss": 4.3078, "step": 212 }, { - "epoch": 0.18, - "learning_rate": 0.00012539999999999999, - "loss": 4.3123, + "epoch": 0.36, + "learning_rate": 0.0001266, + "loss": 4.3063, "step": 213 }, { - "epoch": 0.18, - "learning_rate": 0.00012599999999999997, - "loss": 4.5494, + "epoch": 0.36, + "learning_rate": 0.00012719999999999997, + "loss": 4.3981, "step": 214 }, { - "epoch": 0.18, - "learning_rate": 0.0001266, - "loss": 4.4142, + "epoch": 0.36, + "learning_rate": 0.0001278, + "loss": 4.2747, "step": 215 }, { - "epoch": 0.18, - "learning_rate": 0.00012719999999999997, - "loss": 4.4588, + "epoch": 0.36, + "learning_rate": 0.00012839999999999998, + "loss": 4.3503, "step": 216 }, { - "epoch": 0.18, - "learning_rate": 0.0001278, - "loss": 4.5321, + "epoch": 0.36, + "learning_rate": 0.000129, + "loss": 4.182, "step": 217 }, { - "epoch": 0.18, - "learning_rate": 0.00012839999999999998, - "loss": 4.3559, + "epoch": 0.37, + "learning_rate": 0.00012959999999999998, + "loss": 4.437, "step": 218 }, { - "epoch": 0.18, - "learning_rate": 0.000129, - "loss": 4.1084, + "epoch": 0.37, + "learning_rate": 0.0001302, + "loss": 4.2125, "step": 219 }, { - "epoch": 0.18, - "learning_rate": 0.00012959999999999998, - "loss": 4.5419, + "epoch": 0.37, + "learning_rate": 0.00013079999999999998, + "loss": 4.3465, "step": 220 }, { - "epoch": 0.19, - "learning_rate": 0.0001302, - "loss": 4.3222, + "epoch": 0.37, + "learning_rate": 0.0001314, + "loss": 4.3094, "step": 221 }, { - "epoch": 0.19, - "learning_rate": 0.00013079999999999998, - "loss": 4.7097, + "epoch": 0.37, + "learning_rate": 0.00013199999999999998, + "loss": 4.386, "step": 222 }, { - "epoch": 0.19, - "learning_rate": 0.0001314, - "loss": 4.626, + "epoch": 0.37, + "learning_rate": 0.0001326, + "loss": 4.3321, "step": 223 }, { - "epoch": 0.19, - "learning_rate": 0.00013199999999999998, - "loss": 4.4566, + "epoch": 0.38, + "learning_rate": 0.00013319999999999999, + "loss": 4.351, "step": 224 }, { - "epoch": 0.19, - "learning_rate": 0.0001326, - "loss": 4.5153, + "epoch": 0.38, + "learning_rate": 0.0001338, + "loss": 4.6107, "step": 225 }, { - "epoch": 0.19, - "learning_rate": 0.00013319999999999999, - "loss": 4.2298, + "epoch": 0.38, + "learning_rate": 0.0001344, + "loss": 4.2463, "step": 226 }, { - "epoch": 0.19, - "learning_rate": 0.0001338, - "loss": 4.4521, + "epoch": 0.38, + "learning_rate": 0.000135, + "loss": 4.1284, "step": 227 }, { - "epoch": 0.19, - "learning_rate": 0.0001344, - "loss": 4.6118, + "epoch": 0.38, + "learning_rate": 0.0001356, + "loss": 4.2243, "step": 228 }, { - "epoch": 0.19, - "learning_rate": 0.000135, - "loss": 4.4091, + "epoch": 0.39, + "learning_rate": 0.0001362, + "loss": 4.491, "step": 229 }, { - "epoch": 0.19, - "learning_rate": 0.0001356, - "loss": 4.5025, + "epoch": 0.39, + "learning_rate": 0.0001368, + "loss": 4.4179, "step": 230 }, { - "epoch": 0.19, - "learning_rate": 0.0001362, - "loss": 4.4919, + "epoch": 0.39, + "learning_rate": 0.0001374, + "loss": 4.1993, "step": 231 }, { - "epoch": 0.2, - "learning_rate": 0.0001368, - "loss": 4.3356, + "epoch": 0.39, + "learning_rate": 0.000138, + "loss": 4.291, "step": 232 }, { - "epoch": 0.2, - "learning_rate": 0.0001374, - "loss": 4.6275, + "epoch": 0.39, + "learning_rate": 0.0001386, + "loss": 4.7274, "step": 233 }, { - "epoch": 0.2, - "learning_rate": 0.000138, - "loss": 4.3615, + "epoch": 0.39, + "learning_rate": 0.0001392, + "loss": 4.2559, "step": 234 }, { - "epoch": 0.2, - "learning_rate": 0.0001386, - "loss": 4.5941, + "epoch": 0.4, + "learning_rate": 0.00013979999999999998, + "loss": 4.5068, "step": 235 }, { - "epoch": 0.2, - "learning_rate": 0.0001392, - "loss": 4.5838, + "epoch": 0.4, + "learning_rate": 0.0001404, + "loss": 4.2703, "step": 236 }, { - "epoch": 0.2, - "learning_rate": 0.00013979999999999998, - "loss": 4.2655, + "epoch": 0.4, + "learning_rate": 0.00014099999999999998, + "loss": 4.2862, "step": 237 }, { - "epoch": 0.2, - "learning_rate": 0.0001404, - "loss": 4.3444, + "epoch": 0.4, + "learning_rate": 0.00014159999999999997, + "loss": 4.2675, "step": 238 }, { - "epoch": 0.2, - "learning_rate": 0.00014099999999999998, - "loss": 4.3864, + "epoch": 0.4, + "learning_rate": 0.0001422, + "loss": 4.3457, "step": 239 }, { - "epoch": 0.2, - "learning_rate": 0.00014159999999999997, - "loss": 4.4261, + "epoch": 0.4, + "learning_rate": 0.00014279999999999997, + "loss": 4.3633, "step": 240 }, { - "epoch": 0.2, - "learning_rate": 0.0001422, - "loss": 4.7173, + "epoch": 0.41, + "learning_rate": 0.0001434, + "loss": 4.5277, "step": 241 }, { - "epoch": 0.2, - "learning_rate": 0.00014279999999999997, - "loss": 4.2811, + "epoch": 0.41, + "learning_rate": 0.00014399999999999998, + "loss": 4.4545, "step": 242 }, { - "epoch": 0.2, - "learning_rate": 0.0001434, - "loss": 4.5207, + "epoch": 0.41, + "learning_rate": 0.0001446, + "loss": 4.3021, "step": 243 }, { - "epoch": 0.21, - "learning_rate": 0.00014399999999999998, - "loss": 4.3863, + "epoch": 0.41, + "learning_rate": 0.00014519999999999998, + "loss": 4.6181, "step": 244 }, { - "epoch": 0.21, - "learning_rate": 0.0001446, - "loss": 4.7548, + "epoch": 0.41, + "learning_rate": 0.0001458, + "loss": 4.681, "step": 245 }, { - "epoch": 0.21, - "learning_rate": 0.00014519999999999998, - "loss": 4.2792, + "epoch": 0.41, + "learning_rate": 0.00014639999999999998, + "loss": 4.5904, "step": 246 }, { - "epoch": 0.21, - "learning_rate": 0.0001458, - "loss": 5.5172, + "epoch": 0.42, + "learning_rate": 0.000147, + "loss": 4.819, "step": 247 }, { - "epoch": 0.21, - "learning_rate": 0.00014639999999999998, - "loss": 4.6362, + "epoch": 0.42, + "learning_rate": 0.00014759999999999998, + "loss": 4.7076, "step": 248 }, { - "epoch": 0.21, - "learning_rate": 0.000147, - "loss": 7.8904, + "epoch": 0.42, + "learning_rate": 0.0001482, + "loss": 4.5274, "step": 249 }, { - "epoch": 0.21, - "learning_rate": 0.00014759999999999998, - "loss": 4.6706, + "epoch": 0.42, + "learning_rate": 0.00014879999999999998, + "loss": 4.6464, "step": 250 }, { - "epoch": 0.21, - "learning_rate": 0.0001482, - "loss": 4.6318, + "epoch": 0.42, + "learning_rate": 0.0001494, + "loss": 4.3735, "step": 251 }, { - "epoch": 0.21, - "learning_rate": 0.00014879999999999998, - "loss": 4.3568, + "epoch": 0.42, + "learning_rate": 0.00015, + "loss": 4.4111, "step": 252 }, { - "epoch": 0.21, - "learning_rate": 0.0001494, - "loss": 4.6498, + "epoch": 0.43, + "learning_rate": 0.00015059999999999997, + "loss": 4.5762, "step": 253 }, { - "epoch": 0.21, - "learning_rate": 0.00015, - "loss": 4.2444, + "epoch": 0.43, + "learning_rate": 0.0001512, + "loss": 4.3417, "step": 254 }, { - "epoch": 0.21, - "learning_rate": 0.00015059999999999997, - "loss": 4.4818, + "epoch": 0.43, + "learning_rate": 0.00015179999999999998, + "loss": 4.5584, "step": 255 }, { - "epoch": 0.22, - "learning_rate": 0.0001512, - "loss": 4.7729, + "epoch": 0.43, + "learning_rate": 0.0001524, + "loss": 4.5966, "step": 256 }, { - "epoch": 0.22, - "learning_rate": 0.00015179999999999998, - "loss": 4.2207, + "epoch": 0.43, + "learning_rate": 0.00015299999999999998, + "loss": 4.3418, "step": 257 }, { - "epoch": 0.22, - "learning_rate": 0.0001524, - "loss": 4.9552, + "epoch": 0.43, + "learning_rate": 0.0001536, + "loss": 4.3593, "step": 258 }, { - "epoch": 0.22, - "learning_rate": 0.00015299999999999998, - "loss": 4.5542, + "epoch": 0.44, + "learning_rate": 0.00015419999999999998, + "loss": 4.3958, "step": 259 }, { - "epoch": 0.22, - "learning_rate": 0.0001536, - "loss": 4.3937, + "epoch": 0.44, + "learning_rate": 0.0001548, + "loss": 4.558, "step": 260 }, { - "epoch": 0.22, - "learning_rate": 0.00015419999999999998, - "loss": 4.4997, + "epoch": 0.44, + "learning_rate": 0.00015539999999999998, + "loss": 4.4078, "step": 261 }, { - "epoch": 0.22, - "learning_rate": 0.0001548, - "loss": 4.4278, + "epoch": 0.44, + "learning_rate": 0.000156, + "loss": 4.3026, "step": 262 }, { - "epoch": 0.22, - "learning_rate": 0.00015539999999999998, - "loss": 4.6136, + "epoch": 0.44, + "learning_rate": 0.00015659999999999998, + "loss": 4.6461, "step": 263 }, { - "epoch": 0.22, - "learning_rate": 0.000156, - "loss": 4.8674, + "epoch": 0.44, + "learning_rate": 0.0001572, + "loss": 4.0745, "step": 264 }, { - "epoch": 0.22, - "learning_rate": 0.00015659999999999998, - "loss": 4.6685, + "epoch": 0.45, + "learning_rate": 0.0001578, + "loss": 4.3821, "step": 265 }, { - "epoch": 0.22, - "learning_rate": 0.0001572, - "loss": 5.0145, + "epoch": 0.45, + "learning_rate": 0.0001584, + "loss": 4.3248, "step": 266 }, { - "epoch": 0.22, - "learning_rate": 0.0001578, - "loss": 4.3919, + "epoch": 0.45, + "learning_rate": 0.000159, + "loss": 4.4178, "step": 267 }, { - "epoch": 0.23, - "learning_rate": 0.0001584, - "loss": 4.9681, + "epoch": 0.45, + "learning_rate": 0.0001596, + "loss": 4.3037, "step": 268 }, { - "epoch": 0.23, - "learning_rate": 0.000159, - "loss": 4.5067, + "epoch": 0.45, + "learning_rate": 0.0001602, + "loss": 4.4218, "step": 269 }, { - "epoch": 0.23, - "learning_rate": 0.0001596, - "loss": 4.2741, + "epoch": 0.45, + "learning_rate": 0.0001608, + "loss": 4.3435, "step": 270 }, { - "epoch": 0.23, - "learning_rate": 0.0001602, - "loss": 4.6239, + "epoch": 0.46, + "learning_rate": 0.0001614, + "loss": 4.2526, "step": 271 }, { - "epoch": 0.23, - "learning_rate": 0.0001608, - "loss": 4.2923, + "epoch": 0.46, + "learning_rate": 0.000162, + "loss": 4.5418, "step": 272 }, { - "epoch": 0.23, - "learning_rate": 0.0001614, - "loss": 4.4861, + "epoch": 0.46, + "learning_rate": 0.0001626, + "loss": 4.4274, "step": 273 }, { - "epoch": 0.23, - "learning_rate": 0.000162, - "loss": 4.523, + "epoch": 0.46, + "learning_rate": 0.0001632, + "loss": 4.3375, "step": 274 }, { - "epoch": 0.23, - "learning_rate": 0.0001626, - "loss": 4.4596, + "epoch": 0.46, + "learning_rate": 0.0001638, + "loss": 4.4932, "step": 275 }, { - "epoch": 0.23, - "learning_rate": 0.0001632, - "loss": 4.259, + "epoch": 0.46, + "learning_rate": 0.0001644, + "loss": 4.3847, "step": 276 }, { - "epoch": 0.23, - "learning_rate": 0.0001638, - "loss": 4.9241, + "epoch": 0.47, + "learning_rate": 0.000165, + "loss": 4.5048, "step": 277 }, { - "epoch": 0.23, - "learning_rate": 0.0001644, - "loss": 4.671, + "epoch": 0.47, + "learning_rate": 0.0001656, + "loss": 4.4597, "step": 278 }, { - "epoch": 0.23, - "learning_rate": 0.000165, - "loss": 4.841, + "epoch": 0.47, + "learning_rate": 0.0001662, + "loss": 4.4021, "step": 279 }, { - "epoch": 0.24, - "learning_rate": 0.0001656, - "loss": 4.6318, + "epoch": 0.47, + "learning_rate": 0.0001668, + "loss": 4.4614, "step": 280 }, { - "epoch": 0.24, - "learning_rate": 0.0001662, - "loss": 4.5362, + "epoch": 0.47, + "learning_rate": 0.0001674, + "loss": 4.5178, "step": 281 }, { - "epoch": 0.24, - "learning_rate": 0.0001668, - "loss": 4.4569, + "epoch": 0.47, + "learning_rate": 0.000168, + "loss": 4.4294, "step": 282 }, { - "epoch": 0.24, - "learning_rate": 0.0001674, - "loss": 4.7043, + "epoch": 0.48, + "learning_rate": 0.0001686, + "loss": 4.3767, "step": 283 }, { - "epoch": 0.24, - "learning_rate": 0.000168, - "loss": 4.3861, + "epoch": 0.48, + "learning_rate": 0.00016919999999999997, + "loss": 4.3318, "step": 284 }, { - "epoch": 0.24, - "learning_rate": 0.0001686, - "loss": 4.5491, + "epoch": 0.48, + "learning_rate": 0.00016979999999999998, + "loss": 4.594, "step": 285 }, { - "epoch": 0.24, - "learning_rate": 0.00016919999999999997, - "loss": 4.297, + "epoch": 0.48, + "learning_rate": 0.00017039999999999997, + "loss": 4.2769, "step": 286 }, { - "epoch": 0.24, - "learning_rate": 0.00016979999999999998, - "loss": 4.7139, + "epoch": 0.48, + "learning_rate": 0.00017099999999999998, + "loss": 4.8188, "step": 287 }, { - "epoch": 0.24, - "learning_rate": 0.00017039999999999997, - "loss": 4.3837, + "epoch": 0.48, + "learning_rate": 0.00017159999999999997, + "loss": 4.1727, "step": 288 }, { - "epoch": 0.24, - "learning_rate": 0.00017099999999999998, - "loss": 4.8364, + "epoch": 0.49, + "learning_rate": 0.00017219999999999998, + "loss": 4.4774, "step": 289 }, { - "epoch": 0.24, - "learning_rate": 0.00017159999999999997, - "loss": 4.851, + "epoch": 0.49, + "learning_rate": 0.00017279999999999997, + "loss": 4.6156, "step": 290 }, { - "epoch": 0.24, - "learning_rate": 0.00017219999999999998, - "loss": 5.3491, + "epoch": 0.49, + "learning_rate": 0.00017339999999999996, + "loss": 4.5715, "step": 291 }, { - "epoch": 0.25, - "learning_rate": 0.00017279999999999997, - "loss": 5.1197, + "epoch": 0.49, + "learning_rate": 0.00017399999999999997, + "loss": 4.3004, "step": 292 }, { - "epoch": 0.25, - "learning_rate": 0.00017339999999999996, - "loss": 4.6506, + "epoch": 0.49, + "learning_rate": 0.00017459999999999996, + "loss": 4.457, "step": 293 }, { - "epoch": 0.25, - "learning_rate": 0.00017399999999999997, - "loss": 4.8303, + "epoch": 0.49, + "learning_rate": 0.00017519999999999998, + "loss": 4.5484, "step": 294 }, { - "epoch": 0.25, - "learning_rate": 0.00017459999999999996, - "loss": 4.8478, + "epoch": 0.5, + "learning_rate": 0.00017579999999999996, + "loss": 4.6633, "step": 295 }, { - "epoch": 0.25, - "learning_rate": 0.00017519999999999998, - "loss": 6.4638, + "epoch": 0.5, + "learning_rate": 0.00017639999999999998, + "loss": 4.4057, "step": 296 }, { - "epoch": 0.25, - "learning_rate": 0.00017579999999999996, - "loss": 5.701, + "epoch": 0.5, + "learning_rate": 0.00017699999999999997, + "loss": 5.0306, "step": 297 }, { - "epoch": 0.25, - "learning_rate": 0.00017639999999999998, - "loss": 4.922, + "epoch": 0.5, + "learning_rate": 0.00017759999999999998, + "loss": 4.8782, "step": 298 }, { - "epoch": 0.25, - "learning_rate": 0.00017699999999999997, - "loss": 6.5197, + "epoch": 0.5, + "learning_rate": 0.00017819999999999997, + "loss": 4.7462, "step": 299 }, { - "epoch": 0.25, - "learning_rate": 0.00017759999999999998, - "loss": 5.6574, + "epoch": 0.5, + "learning_rate": 0.00017879999999999998, + "loss": 5.1954, "step": 300 }, { - "epoch": 0.25, - "learning_rate": 0.00017819999999999997, - "loss": 4.6945, + "epoch": 0.51, + "learning_rate": 0.00017939999999999997, + "loss": 4.601, "step": 301 }, { - "epoch": 0.25, - "learning_rate": 0.00017879999999999998, - "loss": 4.6706, + "epoch": 0.51, + "learning_rate": 0.00017999999999999998, + "loss": 4.5284, "step": 302 }, { - "epoch": 0.25, - "learning_rate": 0.00017939999999999997, - "loss": 5.0316, + "epoch": 0.51, + "learning_rate": 0.00018059999999999997, + "loss": 4.6252, "step": 303 }, { - "epoch": 0.26, - "learning_rate": 0.00017999999999999998, - "loss": 4.4729, + "epoch": 0.51, + "learning_rate": 0.00018119999999999999, + "loss": 4.521, "step": 304 }, { - "epoch": 0.26, - "learning_rate": 0.00018059999999999997, - "loss": 4.4609, + "epoch": 0.51, + "learning_rate": 0.00018179999999999997, + "loss": 4.5921, "step": 305 }, { - "epoch": 0.26, - "learning_rate": 0.00018119999999999999, - "loss": 4.6328, + "epoch": 0.51, + "learning_rate": 0.0001824, + "loss": 4.5414, "step": 306 }, { - "epoch": 0.26, - "learning_rate": 0.00018179999999999997, - "loss": 4.6626, + "epoch": 0.52, + "learning_rate": 0.00018299999999999998, + "loss": 4.5384, "step": 307 }, { - "epoch": 0.26, - "learning_rate": 0.0001824, - "loss": 4.3457, - "step": 308 + "epoch": 0.52, + "learning_rate": 0.0001836, + "loss": 4.7024, + "step": 308 }, { - "epoch": 0.26, - "learning_rate": 0.00018299999999999998, - "loss": 4.2517, + "epoch": 0.52, + "learning_rate": 0.00018419999999999998, + "loss": 4.5066, "step": 309 }, { - "epoch": 0.26, - "learning_rate": 0.0001836, - "loss": 5.1203, + "epoch": 0.52, + "learning_rate": 0.0001848, + "loss": 4.4578, "step": 310 }, { - "epoch": 0.26, - "learning_rate": 0.00018419999999999998, - "loss": 4.8707, + "epoch": 0.52, + "learning_rate": 0.00018539999999999998, + "loss": 4.2657, "step": 311 }, { - "epoch": 0.26, - "learning_rate": 0.0001848, - "loss": 4.4209, + "epoch": 0.52, + "learning_rate": 0.000186, + "loss": 4.5986, "step": 312 }, { - "epoch": 0.26, - "learning_rate": 0.00018539999999999998, - "loss": 4.4336, + "epoch": 0.53, + "learning_rate": 0.00018659999999999998, + "loss": 4.5689, "step": 313 }, { - "epoch": 0.26, - "learning_rate": 0.000186, - "loss": 4.6532, + "epoch": 0.53, + "learning_rate": 0.0001872, + "loss": 4.624, "step": 314 }, { - "epoch": 0.26, - "learning_rate": 0.00018659999999999998, - "loss": 4.2587, + "epoch": 0.53, + "learning_rate": 0.00018779999999999998, + "loss": 4.3777, "step": 315 }, { - "epoch": 0.27, - "learning_rate": 0.0001872, - "loss": 4.5758, + "epoch": 0.53, + "learning_rate": 0.00018839999999999997, + "loss": 4.6881, "step": 316 }, { - "epoch": 0.27, - "learning_rate": 0.00018779999999999998, - "loss": 4.7729, + "epoch": 0.53, + "learning_rate": 0.00018899999999999999, + "loss": 4.4102, "step": 317 }, { - "epoch": 0.27, - "learning_rate": 0.00018839999999999997, - "loss": 4.5811, + "epoch": 0.53, + "learning_rate": 0.00018959999999999997, + "loss": 4.8632, "step": 318 }, { - "epoch": 0.27, - "learning_rate": 0.00018899999999999999, - "loss": 4.648, + "epoch": 0.54, + "learning_rate": 0.0001902, + "loss": 4.5565, "step": 319 }, { - "epoch": 0.27, - "learning_rate": 0.00018959999999999997, - "loss": 4.4726, + "epoch": 0.54, + "learning_rate": 0.00019079999999999998, + "loss": 4.6379, "step": 320 }, { - "epoch": 0.27, - "learning_rate": 0.0001902, - "loss": 4.3312, + "epoch": 0.54, + "learning_rate": 0.0001914, + "loss": 4.2919, "step": 321 }, { - "epoch": 0.27, - "learning_rate": 0.00019079999999999998, - "loss": 4.325, + "epoch": 0.54, + "learning_rate": 0.00019199999999999998, + "loss": 4.4188, "step": 322 }, { - "epoch": 0.27, - "learning_rate": 0.0001914, - "loss": 4.4029, + "epoch": 0.54, + "learning_rate": 0.0001926, + "loss": 4.6448, "step": 323 }, { - "epoch": 0.27, - "learning_rate": 0.00019199999999999998, - "loss": 5.1371, + "epoch": 0.54, + "learning_rate": 0.00019319999999999998, + "loss": 4.4343, "step": 324 }, { - "epoch": 0.27, - "learning_rate": 0.0001926, - "loss": 4.5978, + "epoch": 0.55, + "learning_rate": 0.0001938, + "loss": 4.8242, "step": 325 }, { - "epoch": 0.27, - "learning_rate": 0.00019319999999999998, - "loss": 5.219, + "epoch": 0.55, + "learning_rate": 0.00019439999999999998, + "loss": 4.3897, "step": 326 }, { - "epoch": 0.27, - "learning_rate": 0.0001938, - "loss": 4.8118, + "epoch": 0.55, + "learning_rate": 0.000195, + "loss": 4.7531, "step": 327 }, { - "epoch": 0.28, - "learning_rate": 0.00019439999999999998, - "loss": 4.7945, + "epoch": 0.55, + "learning_rate": 0.00019559999999999998, + "loss": 4.5004, "step": 328 }, { - "epoch": 0.28, - "learning_rate": 0.000195, - "loss": 4.4908, + "epoch": 0.55, + "learning_rate": 0.0001962, + "loss": 4.3962, "step": 329 }, { - "epoch": 0.28, - "learning_rate": 0.00019559999999999998, - "loss": 4.9624, + "epoch": 0.55, + "learning_rate": 0.00019679999999999999, + "loss": 4.4023, "step": 330 }, { - "epoch": 0.28, - "learning_rate": 0.0001962, - "loss": 4.496, + "epoch": 0.56, + "learning_rate": 0.0001974, + "loss": 4.5184, "step": 331 }, { - "epoch": 0.28, - "learning_rate": 0.00019679999999999999, - "loss": 4.7033, + "epoch": 0.56, + "learning_rate": 0.000198, + "loss": 4.3142, "step": 332 }, { - "epoch": 0.28, - "learning_rate": 0.0001974, - "loss": 4.6788, + "epoch": 0.56, + "learning_rate": 0.0001986, + "loss": 4.5157, "step": 333 }, { - "epoch": 0.28, - "learning_rate": 0.000198, - "loss": 4.5631, + "epoch": 0.56, + "learning_rate": 0.0001992, + "loss": 4.5152, "step": 334 }, { - "epoch": 0.28, - "learning_rate": 0.0001986, - "loss": 4.702, + "epoch": 0.56, + "learning_rate": 0.0001998, + "loss": 4.4877, "step": 335 }, { - "epoch": 0.28, - "learning_rate": 0.0001992, - "loss": 5.5658, + "epoch": 0.56, + "learning_rate": 0.0002004, + "loss": 4.44, "step": 336 }, { - "epoch": 0.28, - "learning_rate": 0.0001998, - "loss": 4.7049, + "epoch": 0.57, + "learning_rate": 0.000201, + "loss": 4.6629, "step": 337 }, { - "epoch": 0.28, - "learning_rate": 0.0002004, - "loss": 4.7927, + "epoch": 0.57, + "learning_rate": 0.0002016, + "loss": 4.4032, "step": 338 }, { - "epoch": 0.28, - "learning_rate": 0.000201, - "loss": 4.7044, + "epoch": 0.57, + "learning_rate": 0.0002022, + "loss": 4.4355, "step": 339 }, { - "epoch": 0.29, - "learning_rate": 0.0002016, - "loss": 4.6731, + "epoch": 0.57, + "learning_rate": 0.0002028, + "loss": 4.5324, "step": 340 }, { - "epoch": 0.29, - "learning_rate": 0.0002022, - "loss": 4.8644, + "epoch": 0.57, + "learning_rate": 0.00020339999999999998, + "loss": 4.4673, "step": 341 }, { - "epoch": 0.29, - "learning_rate": 0.0002028, - "loss": 5.1974, + "epoch": 0.58, + "learning_rate": 0.000204, + "loss": 4.8003, "step": 342 }, { - "epoch": 0.29, - "learning_rate": 0.00020339999999999998, - "loss": 4.5775, + "epoch": 0.58, + "learning_rate": 0.00020459999999999999, + "loss": 4.6395, "step": 343 }, { - "epoch": 0.29, - "learning_rate": 0.000204, - "loss": 5.4593, + "epoch": 0.58, + "learning_rate": 0.0002052, + "loss": 4.6675, "step": 344 }, { - "epoch": 0.29, - "learning_rate": 0.00020459999999999999, - "loss": 4.8787, + "epoch": 0.58, + "learning_rate": 0.0002058, + "loss": 4.8611, "step": 345 }, { - "epoch": 0.29, - "learning_rate": 0.0002052, - "loss": 4.8025, + "epoch": 0.58, + "learning_rate": 0.00020639999999999998, + "loss": 4.363, "step": 346 }, { - "epoch": 0.29, - "learning_rate": 0.0002058, - "loss": 5.2866, + "epoch": 0.58, + "learning_rate": 0.00020699999999999996, + "loss": 4.3723, "step": 347 }, { - "epoch": 0.29, - "learning_rate": 0.00020639999999999998, - "loss": 5.9515, + "epoch": 0.59, + "learning_rate": 0.00020759999999999998, + "loss": 4.7011, "step": 348 }, { - "epoch": 0.29, - "learning_rate": 0.00020699999999999996, - "loss": 5.1083, + "epoch": 0.59, + "learning_rate": 0.00020819999999999996, + "loss": 4.5094, "step": 349 }, { - "epoch": 0.29, - "learning_rate": 0.00020759999999999998, - "loss": 4.6994, + "epoch": 0.59, + "learning_rate": 0.00020879999999999998, + "loss": 4.6712, "step": 350 }, { - "epoch": 0.3, - "learning_rate": 0.00020819999999999996, - "loss": 5.3215, + "epoch": 0.59, + "learning_rate": 0.00020939999999999997, + "loss": 4.629, "step": 351 }, { - "epoch": 0.3, - "learning_rate": 0.00020879999999999998, - "loss": 4.5359, + "epoch": 0.59, + "learning_rate": 0.00020999999999999998, + "loss": 4.5946, "step": 352 }, { - "epoch": 0.3, - "learning_rate": 0.00020939999999999997, - "loss": 5.0289, + "epoch": 0.59, + "learning_rate": 0.00021059999999999997, + "loss": 4.517, "step": 353 }, { - "epoch": 0.3, - "learning_rate": 0.00020999999999999998, - "loss": 4.9806, + "epoch": 0.6, + "learning_rate": 0.00021119999999999996, + "loss": 4.4386, "step": 354 }, { - "epoch": 0.3, - "learning_rate": 0.00021059999999999997, - "loss": 4.5323, + "epoch": 0.6, + "learning_rate": 0.00021179999999999997, + "loss": 4.5275, "step": 355 }, { - "epoch": 0.3, - "learning_rate": 0.00021119999999999996, - "loss": 5.1149, + "epoch": 0.6, + "learning_rate": 0.00021239999999999996, + "loss": 4.3692, "step": 356 }, { - "epoch": 0.3, - "learning_rate": 0.00021179999999999997, - "loss": 4.779, + "epoch": 0.6, + "learning_rate": 0.00021299999999999997, + "loss": 4.5466, "step": 357 }, { - "epoch": 0.3, - "learning_rate": 0.00021239999999999996, - "loss": 4.3883, + "epoch": 0.6, + "learning_rate": 0.00021359999999999996, + "loss": 4.9401, "step": 358 }, { - "epoch": 0.3, - "learning_rate": 0.00021299999999999997, - "loss": 4.7019, + "epoch": 0.6, + "learning_rate": 0.00021419999999999998, + "loss": 4.3218, "step": 359 }, { - "epoch": 0.3, - "learning_rate": 0.00021359999999999996, - "loss": 4.7448, + "epoch": 0.61, + "learning_rate": 0.00021479999999999996, + "loss": 4.6933, "step": 360 }, { - "epoch": 0.3, - "learning_rate": 0.00021419999999999998, - "loss": 4.5466, + "epoch": 0.61, + "learning_rate": 0.00021539999999999998, + "loss": 4.6379, "step": 361 }, { - "epoch": 0.3, - "learning_rate": 0.00021479999999999996, - "loss": 4.8753, + "epoch": 0.61, + "learning_rate": 0.00021599999999999996, + "loss": 4.377, "step": 362 }, { - "epoch": 0.31, - "learning_rate": 0.00021539999999999998, - "loss": 4.4886, + "epoch": 0.61, + "learning_rate": 0.00021659999999999998, + "loss": 4.898, "step": 363 }, { - "epoch": 0.31, - "learning_rate": 0.00021599999999999996, - "loss": 4.5213, + "epoch": 0.61, + "learning_rate": 0.00021719999999999997, + "loss": 4.4985, "step": 364 }, { - "epoch": 0.31, - "learning_rate": 0.00021659999999999998, - "loss": 5.0087, + "epoch": 0.61, + "learning_rate": 0.00021779999999999998, + "loss": 4.5904, "step": 365 }, { - "epoch": 0.31, - "learning_rate": 0.00021719999999999997, - "loss": 6.115, + "epoch": 0.62, + "learning_rate": 0.00021839999999999997, + "loss": 4.4725, "step": 366 }, { - "epoch": 0.31, - "learning_rate": 0.00021779999999999998, - "loss": 4.7595, + "epoch": 0.62, + "learning_rate": 0.00021899999999999998, + "loss": 4.4717, "step": 367 }, { - "epoch": 0.31, - "learning_rate": 0.00021839999999999997, - "loss": 4.8554, + "epoch": 0.62, + "learning_rate": 0.00021959999999999997, + "loss": 4.3563, "step": 368 }, { - "epoch": 0.31, - "learning_rate": 0.00021899999999999998, - "loss": 4.6204, + "epoch": 0.62, + "learning_rate": 0.00022019999999999999, + "loss": 4.4058, "step": 369 }, { - "epoch": 0.31, - "learning_rate": 0.00021959999999999997, - "loss": 4.7214, + "epoch": 0.62, + "learning_rate": 0.00022079999999999997, + "loss": 4.7998, "step": 370 }, { - "epoch": 0.31, - "learning_rate": 0.00022019999999999999, - "loss": 5.0148, + "epoch": 0.62, + "learning_rate": 0.0002214, + "loss": 4.5314, "step": 371 }, { - "epoch": 0.31, - "learning_rate": 0.00022079999999999997, - "loss": 4.7485, + "epoch": 0.63, + "learning_rate": 0.00022199999999999998, + "loss": 4.398, "step": 372 }, { - "epoch": 0.31, - "learning_rate": 0.0002214, - "loss": 4.7558, + "epoch": 0.63, + "learning_rate": 0.0002226, + "loss": 4.4327, "step": 373 }, { - "epoch": 0.31, - "learning_rate": 0.00022199999999999998, - "loss": 4.5435, + "epoch": 0.63, + "learning_rate": 0.00022319999999999998, + "loss": 4.7163, "step": 374 }, { - "epoch": 0.32, - "learning_rate": 0.0002226, - "loss": 4.7954, + "epoch": 0.63, + "learning_rate": 0.0002238, + "loss": 4.7412, "step": 375 }, { - "epoch": 0.32, - "learning_rate": 0.00022319999999999998, - "loss": 4.7268, + "epoch": 0.63, + "learning_rate": 0.00022439999999999998, + "loss": 4.6339, "step": 376 }, { - "epoch": 0.32, - "learning_rate": 0.0002238, - "loss": 4.6724, + "epoch": 0.63, + "learning_rate": 0.000225, + "loss": 4.5151, "step": 377 }, { - "epoch": 0.32, - "learning_rate": 0.00022439999999999998, - "loss": 4.4617, + "epoch": 0.64, + "learning_rate": 0.00022559999999999998, + "loss": 4.7596, "step": 378 }, { - "epoch": 0.32, - "learning_rate": 0.000225, - "loss": 4.7142, + "epoch": 0.64, + "learning_rate": 0.00022619999999999997, + "loss": 4.3111, "step": 379 }, { - "epoch": 0.32, - "learning_rate": 0.00022559999999999998, - "loss": 4.6315, + "epoch": 0.64, + "learning_rate": 0.00022679999999999998, + "loss": 4.5822, "step": 380 }, { - "epoch": 0.32, - "learning_rate": 0.00022619999999999997, - "loss": 4.3959, + "epoch": 0.64, + "learning_rate": 0.00022739999999999997, + "loss": 4.573, "step": 381 }, { - "epoch": 0.32, - "learning_rate": 0.00022679999999999998, - "loss": 4.5737, + "epoch": 0.64, + "learning_rate": 0.00022799999999999999, + "loss": 4.3831, "step": 382 }, { - "epoch": 0.32, - "learning_rate": 0.00022739999999999997, - "loss": 4.6512, + "epoch": 0.64, + "learning_rate": 0.00022859999999999997, + "loss": 4.5107, "step": 383 }, { - "epoch": 0.32, - "learning_rate": 0.00022799999999999999, - "loss": 4.6704, + "epoch": 0.65, + "learning_rate": 0.0002292, + "loss": 4.4021, "step": 384 }, { - "epoch": 0.32, - "learning_rate": 0.00022859999999999997, - "loss": 4.8355, + "epoch": 0.65, + "learning_rate": 0.00022979999999999997, + "loss": 4.7495, "step": 385 }, { - "epoch": 0.32, - "learning_rate": 0.0002292, - "loss": 4.6387, + "epoch": 0.65, + "learning_rate": 0.0002304, + "loss": 4.8891, "step": 386 }, { - "epoch": 0.33, - "learning_rate": 0.00022979999999999997, - "loss": 4.8218, + "epoch": 0.65, + "learning_rate": 0.00023099999999999998, + "loss": 5.0482, "step": 387 }, { - "epoch": 0.33, - "learning_rate": 0.0002304, - "loss": 4.5275, + "epoch": 0.65, + "learning_rate": 0.0002316, + "loss": 4.5507, "step": 388 }, { - "epoch": 0.33, - "learning_rate": 0.00023099999999999998, - "loss": 4.7885, + "epoch": 0.65, + "learning_rate": 0.00023219999999999998, + "loss": 4.5954, "step": 389 }, { - "epoch": 0.33, - "learning_rate": 0.0002316, - "loss": 4.6711, + "epoch": 0.66, + "learning_rate": 0.0002328, + "loss": 4.4654, "step": 390 }, { - "epoch": 0.33, - "learning_rate": 0.00023219999999999998, - "loss": 4.6913, + "epoch": 0.66, + "learning_rate": 0.00023339999999999998, + "loss": 4.5147, "step": 391 }, { - "epoch": 0.33, - "learning_rate": 0.0002328, - "loss": 5.5985, + "epoch": 0.66, + "learning_rate": 0.000234, + "loss": 4.6645, "step": 392 }, { - "epoch": 0.33, - "learning_rate": 0.00023339999999999998, - "loss": 4.501, + "epoch": 0.66, + "learning_rate": 0.00023459999999999998, + "loss": 4.6974, "step": 393 }, { - "epoch": 0.33, - "learning_rate": 0.000234, - "loss": 5.082, + "epoch": 0.66, + "learning_rate": 0.0002352, + "loss": 4.4298, "step": 394 }, { - "epoch": 0.33, - "learning_rate": 0.00023459999999999998, - "loss": 4.4409, + "epoch": 0.66, + "learning_rate": 0.00023579999999999999, + "loss": 4.8421, "step": 395 }, { - "epoch": 0.33, - "learning_rate": 0.0002352, - "loss": 5.9987, + "epoch": 0.67, + "learning_rate": 0.0002364, + "loss": 4.6812, "step": 396 }, { - "epoch": 0.33, - "learning_rate": 0.00023579999999999999, - "loss": 5.7972, + "epoch": 0.67, + "learning_rate": 0.000237, + "loss": 5.1311, "step": 397 }, { - "epoch": 0.33, - "learning_rate": 0.0002364, - "loss": 5.1902, + "epoch": 0.67, + "learning_rate": 0.0002376, + "loss": 5.5849, "step": 398 }, { - "epoch": 0.34, - "learning_rate": 0.000237, - "loss": 5.4818, + "epoch": 0.67, + "learning_rate": 0.0002382, + "loss": 4.9822, "step": 399 }, { - "epoch": 0.34, - "learning_rate": 0.0002376, - "loss": 4.9471, + "epoch": 0.67, + "learning_rate": 0.0002388, + "loss": 5.4755, "step": 400 }, { - "epoch": 0.34, - "learning_rate": 0.0002382, - "loss": 5.6634, + "epoch": 0.67, + "learning_rate": 0.0002394, + "loss": 5.0482, "step": 401 }, { - "epoch": 0.34, - "learning_rate": 0.0002388, - "loss": 5.0046, + "epoch": 0.68, + "learning_rate": 0.00023999999999999998, + "loss": 4.7492, "step": 402 }, { - "epoch": 0.34, - "learning_rate": 0.0002394, - "loss": 4.6863, + "epoch": 0.68, + "learning_rate": 0.0002406, + "loss": 4.5452, "step": 403 }, { - "epoch": 0.34, - "learning_rate": 0.00023999999999999998, - "loss": 4.6132, + "epoch": 0.68, + "learning_rate": 0.00024119999999999998, + "loss": 4.8564, "step": 404 }, { - "epoch": 0.34, - "learning_rate": 0.0002406, - "loss": 4.8202, + "epoch": 0.68, + "learning_rate": 0.0002418, + "loss": 4.568, "step": 405 }, { - "epoch": 0.34, - "learning_rate": 0.00024119999999999998, - "loss": 4.604, + "epoch": 0.68, + "learning_rate": 0.00024239999999999998, + "loss": 4.5792, "step": 406 }, { - "epoch": 0.34, - "learning_rate": 0.0002418, - "loss": 4.4995, + "epoch": 0.68, + "learning_rate": 0.000243, + "loss": 4.8879, "step": 407 }, { - "epoch": 0.34, - "learning_rate": 0.00024239999999999998, - "loss": 5.1402, + "epoch": 0.69, + "learning_rate": 0.00024359999999999999, + "loss": 4.5071, "step": 408 }, { - "epoch": 0.34, - "learning_rate": 0.000243, - "loss": 4.8212, + "epoch": 0.69, + "learning_rate": 0.00024419999999999997, + "loss": 4.9798, "step": 409 }, { - "epoch": 0.34, - "learning_rate": 0.00024359999999999999, - "loss": 4.763, + "epoch": 0.69, + "learning_rate": 0.0002448, + "loss": 4.5812, "step": 410 }, { - "epoch": 0.35, - "learning_rate": 0.00024419999999999997, - "loss": 5.0245, + "epoch": 0.69, + "learning_rate": 0.00024539999999999995, + "loss": 4.4708, "step": 411 }, { - "epoch": 0.35, - "learning_rate": 0.0002448, - "loss": 4.5873, + "epoch": 0.69, + "learning_rate": 0.00024599999999999996, + "loss": 4.7339, "step": 412 }, { - "epoch": 0.35, - "learning_rate": 0.00024539999999999995, - "loss": 5.1268, + "epoch": 0.69, + "learning_rate": 0.0002466, + "loss": 4.467, "step": 413 }, { - "epoch": 0.35, - "learning_rate": 0.00024599999999999996, - "loss": 4.3666, + "epoch": 0.7, + "learning_rate": 0.0002472, + "loss": 4.7394, "step": 414 }, { - "epoch": 0.35, - "learning_rate": 0.0002466, - "loss": 4.9047, + "epoch": 0.7, + "learning_rate": 0.00024779999999999995, + "loss": 4.5679, "step": 415 }, { - "epoch": 0.35, - "learning_rate": 0.0002472, - "loss": 4.7815, + "epoch": 0.7, + "learning_rate": 0.00024839999999999997, + "loss": 4.5506, "step": 416 }, { - "epoch": 0.35, - "learning_rate": 0.00024779999999999995, - "loss": 4.2988, + "epoch": 0.7, + "learning_rate": 0.000249, + "loss": 4.5689, "step": 417 }, { - "epoch": 0.35, - "learning_rate": 0.00024839999999999997, - "loss": 4.8659, + "epoch": 0.7, + "learning_rate": 0.00024959999999999994, + "loss": 4.1599, "step": 418 }, { - "epoch": 0.35, - "learning_rate": 0.000249, - "loss": 4.5983, + "epoch": 0.7, + "learning_rate": 0.00025019999999999996, + "loss": 4.8572, "step": 419 }, { - "epoch": 0.35, - "learning_rate": 0.00024959999999999994, - "loss": 4.5694, + "epoch": 0.71, + "learning_rate": 0.00025079999999999997, + "loss": 4.722, "step": 420 }, { - "epoch": 0.35, - "learning_rate": 0.00025019999999999996, - "loss": 5.0518, + "epoch": 0.71, + "learning_rate": 0.0002514, + "loss": 4.2483, "step": 421 }, { - "epoch": 0.35, - "learning_rate": 0.00025079999999999997, - "loss": 4.6937, + "epoch": 0.71, + "learning_rate": 0.00025199999999999995, + "loss": 4.3537, "step": 422 }, { - "epoch": 0.36, - "learning_rate": 0.0002514, - "loss": 4.7808, + "epoch": 0.71, + "learning_rate": 0.00025259999999999996, + "loss": 4.6733, "step": 423 }, { - "epoch": 0.36, - "learning_rate": 0.00025199999999999995, - "loss": 4.4995, + "epoch": 0.71, + "learning_rate": 0.0002532, + "loss": 4.4947, "step": 424 }, { - "epoch": 0.36, - "learning_rate": 0.00025259999999999996, - "loss": 5.3755, + "epoch": 0.71, + "learning_rate": 0.0002538, + "loss": 4.6257, "step": 425 }, { - "epoch": 0.36, - "learning_rate": 0.0002532, - "loss": 4.5319, + "epoch": 0.72, + "learning_rate": 0.00025439999999999995, + "loss": 4.5752, "step": 426 }, { - "epoch": 0.36, - "learning_rate": 0.0002538, - "loss": 4.3798, + "epoch": 0.72, + "learning_rate": 0.00025499999999999996, + "loss": 4.286, "step": 427 }, { - "epoch": 0.36, - "learning_rate": 0.00025439999999999995, - "loss": 4.449, + "epoch": 0.72, + "learning_rate": 0.0002556, + "loss": 4.2402, "step": 428 }, { - "epoch": 0.36, - "learning_rate": 0.00025499999999999996, - "loss": 5.896, + "epoch": 0.72, + "learning_rate": 0.0002562, + "loss": 4.8508, "step": 429 }, { - "epoch": 0.36, - "learning_rate": 0.0002556, - "loss": 5.1909, + "epoch": 0.72, + "learning_rate": 0.00025679999999999995, + "loss": 4.4226, "step": 430 }, { - "epoch": 0.36, - "learning_rate": 0.0002562, - "loss": 4.8339, + "epoch": 0.72, + "learning_rate": 0.00025739999999999997, + "loss": 4.8964, "step": 431 }, { - "epoch": 0.36, - "learning_rate": 0.00025679999999999995, - "loss": 4.6261, + "epoch": 0.73, + "learning_rate": 0.000258, + "loss": 4.3863, "step": 432 }, { - "epoch": 0.36, - "learning_rate": 0.00025739999999999997, - "loss": 4.5436, + "epoch": 0.73, + "learning_rate": 0.0002586, + "loss": 4.5225, "step": 433 }, { - "epoch": 0.36, - "learning_rate": 0.000258, - "loss": 4.7635, + "epoch": 0.73, + "learning_rate": 0.00025919999999999996, + "loss": 4.6005, "step": 434 }, { - "epoch": 0.37, - "learning_rate": 0.0002586, - "loss": 4.8065, + "epoch": 0.73, + "learning_rate": 0.00025979999999999997, + "loss": 4.6333, "step": 435 }, { - "epoch": 0.37, - "learning_rate": 0.00025919999999999996, - "loss": 4.8411, + "epoch": 0.73, + "learning_rate": 0.0002604, + "loss": 4.4055, "step": 436 }, { - "epoch": 0.37, - "learning_rate": 0.00025979999999999997, - "loss": 4.4299, + "epoch": 0.73, + "learning_rate": 0.000261, + "loss": 4.5249, "step": 437 }, { - "epoch": 0.37, - "learning_rate": 0.0002604, - "loss": 4.4418, + "epoch": 0.74, + "learning_rate": 0.00026159999999999996, + "loss": 4.528, "step": 438 }, { - "epoch": 0.37, - "learning_rate": 0.000261, - "loss": 4.808, + "epoch": 0.74, + "learning_rate": 0.0002622, + "loss": 4.4824, "step": 439 }, { - "epoch": 0.37, - "learning_rate": 0.00026159999999999996, - "loss": 4.6162, + "epoch": 0.74, + "learning_rate": 0.0002628, + "loss": 4.9304, "step": 440 }, { - "epoch": 0.37, - "learning_rate": 0.0002622, - "loss": 5.7287, + "epoch": 0.74, + "learning_rate": 0.00026339999999999995, + "loss": 4.9839, "step": 441 }, { - "epoch": 0.37, - "learning_rate": 0.0002628, - "loss": 4.6771, + "epoch": 0.74, + "learning_rate": 0.00026399999999999997, + "loss": 4.8169, "step": 442 }, { - "epoch": 0.37, - "learning_rate": 0.00026339999999999995, - "loss": 4.529, + "epoch": 0.74, + "learning_rate": 0.0002646, + "loss": 4.9251, "step": 443 }, { - "epoch": 0.37, - "learning_rate": 0.00026399999999999997, - "loss": 4.6037, + "epoch": 0.75, + "learning_rate": 0.0002652, + "loss": 4.7229, "step": 444 }, { - "epoch": 0.37, - "learning_rate": 0.0002646, - "loss": 4.5055, + "epoch": 0.75, + "learning_rate": 0.00026579999999999996, + "loss": 4.453, "step": 445 }, { - "epoch": 0.37, - "learning_rate": 0.0002652, - "loss": 5.2715, + "epoch": 0.75, + "learning_rate": 0.00026639999999999997, + "loss": 4.8808, "step": 446 }, { - "epoch": 0.38, - "learning_rate": 0.00026579999999999996, - "loss": 5.7413, + "epoch": 0.75, + "learning_rate": 0.000267, + "loss": 4.5888, "step": 447 }, { - "epoch": 0.38, - "learning_rate": 0.00026639999999999997, - "loss": 5.028, + "epoch": 0.75, + "learning_rate": 0.0002676, + "loss": 4.671, "step": 448 }, { - "epoch": 0.38, - "learning_rate": 0.000267, - "loss": 4.8087, + "epoch": 0.75, + "learning_rate": 0.00026819999999999996, + "loss": 5.2919, "step": 449 }, { - "epoch": 0.38, - "learning_rate": 0.0002676, - "loss": 4.8552, + "epoch": 0.76, + "learning_rate": 0.0002688, + "loss": 4.4541, "step": 450 }, { - "epoch": 0.38, - "learning_rate": 0.00026819999999999996, - "loss": 5.0898, + "epoch": 0.76, + "learning_rate": 0.0002694, + "loss": 4.5015, "step": 451 }, { - "epoch": 0.38, - "learning_rate": 0.0002688, - "loss": 4.875, + "epoch": 0.76, + "learning_rate": 0.00027, + "loss": 4.6347, "step": 452 }, { - "epoch": 0.38, - "learning_rate": 0.0002694, - "loss": 4.9811, + "epoch": 0.76, + "learning_rate": 0.00027059999999999996, + "loss": 4.4529, "step": 453 }, { - "epoch": 0.38, - "learning_rate": 0.00027, - "loss": 4.7576, + "epoch": 0.76, + "learning_rate": 0.0002712, + "loss": 4.7072, "step": 454 }, { - "epoch": 0.38, - "learning_rate": 0.00027059999999999996, - "loss": 5.1758, + "epoch": 0.77, + "learning_rate": 0.0002718, + "loss": 4.6178, "step": 455 }, { - "epoch": 0.38, - "learning_rate": 0.0002712, - "loss": 5.1518, + "epoch": 0.77, + "learning_rate": 0.0002724, + "loss": 4.7541, "step": 456 }, { - "epoch": 0.38, - "learning_rate": 0.0002718, - "loss": 4.7805, + "epoch": 0.77, + "learning_rate": 0.00027299999999999997, + "loss": 4.8252, "step": 457 }, { - "epoch": 0.39, - "learning_rate": 0.0002724, - "loss": 5.4872, + "epoch": 0.77, + "learning_rate": 0.0002736, + "loss": 4.7205, "step": 458 }, { - "epoch": 0.39, - "learning_rate": 0.00027299999999999997, - "loss": 4.6417, + "epoch": 0.77, + "learning_rate": 0.0002742, + "loss": 4.7718, "step": 459 }, { - "epoch": 0.39, - "learning_rate": 0.0002736, - "loss": 4.643, + "epoch": 0.77, + "learning_rate": 0.0002748, + "loss": 4.5784, "step": 460 }, { - "epoch": 0.39, - "learning_rate": 0.0002742, - "loss": 4.8394, + "epoch": 0.78, + "learning_rate": 0.00027539999999999997, + "loss": 4.7469, "step": 461 }, { - "epoch": 0.39, - "learning_rate": 0.0002748, - "loss": 4.617, + "epoch": 0.78, + "learning_rate": 0.000276, + "loss": 4.4427, "step": 462 }, { - "epoch": 0.39, - "learning_rate": 0.00027539999999999997, - "loss": 4.5665, + "epoch": 0.78, + "learning_rate": 0.0002766, + "loss": 4.5062, "step": 463 }, { - "epoch": 0.39, - "learning_rate": 0.000276, - "loss": 4.9569, + "epoch": 0.78, + "learning_rate": 0.0002772, + "loss": 4.6286, "step": 464 }, { - "epoch": 0.39, - "learning_rate": 0.0002766, - "loss": 4.7609, + "epoch": 0.78, + "learning_rate": 0.0002778, + "loss": 4.8309, "step": 465 }, { - "epoch": 0.39, - "learning_rate": 0.0002772, - "loss": 4.5635, + "epoch": 0.78, + "learning_rate": 0.0002784, + "loss": 4.5841, "step": 466 }, { - "epoch": 0.39, - "learning_rate": 0.0002778, - "loss": 4.691, + "epoch": 0.79, + "learning_rate": 0.000279, + "loss": 4.581, "step": 467 }, { - "epoch": 0.39, - "learning_rate": 0.0002784, - "loss": 4.7105, + "epoch": 0.79, + "learning_rate": 0.00027959999999999997, + "loss": 4.5023, "step": 468 }, { - "epoch": 0.39, - "learning_rate": 0.000279, - "loss": 4.5443, + "epoch": 0.79, + "learning_rate": 0.0002802, + "loss": 4.5518, "step": 469 }, { - "epoch": 0.4, - "learning_rate": 0.00027959999999999997, - "loss": 4.9041, + "epoch": 0.79, + "learning_rate": 0.0002808, + "loss": 4.3897, "step": 470 }, { - "epoch": 0.4, - "learning_rate": 0.0002802, - "loss": 4.6443, + "epoch": 0.79, + "learning_rate": 0.00028139999999999996, + "loss": 4.4474, "step": 471 }, { - "epoch": 0.4, - "learning_rate": 0.0002808, - "loss": 4.8053, + "epoch": 0.79, + "learning_rate": 0.00028199999999999997, + "loss": 4.4435, "step": 472 }, { - "epoch": 0.4, - "learning_rate": 0.00028139999999999996, - "loss": 4.5651, + "epoch": 0.8, + "learning_rate": 0.0002826, + "loss": 4.5926, "step": 473 }, { - "epoch": 0.4, - "learning_rate": 0.00028199999999999997, - "loss": 4.7821, + "epoch": 0.8, + "learning_rate": 0.00028319999999999994, + "loss": 4.5936, "step": 474 }, { - "epoch": 0.4, - "learning_rate": 0.0002826, - "loss": 5.1903, + "epoch": 0.8, + "learning_rate": 0.00028379999999999996, + "loss": 4.5352, "step": 475 }, { - "epoch": 0.4, - "learning_rate": 0.00028319999999999994, - "loss": 5.1694, + "epoch": 0.8, + "learning_rate": 0.0002844, + "loss": 4.7309, "step": 476 }, { - "epoch": 0.4, - "learning_rate": 0.00028379999999999996, - "loss": 4.3456, + "epoch": 0.8, + "learning_rate": 0.000285, + "loss": 4.6586, "step": 477 }, { - "epoch": 0.4, - "learning_rate": 0.0002844, - "loss": 4.8221, + "epoch": 0.8, + "learning_rate": 0.00028559999999999995, + "loss": 4.6348, "step": 478 }, { - "epoch": 0.4, - "learning_rate": 0.000285, - "loss": 4.6403, + "epoch": 0.81, + "learning_rate": 0.00028619999999999996, + "loss": 4.3622, "step": 479 }, { - "epoch": 0.4, - "learning_rate": 0.00028559999999999995, - "loss": 4.5352, + "epoch": 0.81, + "learning_rate": 0.0002868, + "loss": 4.3913, "step": 480 }, { - "epoch": 0.4, - "learning_rate": 0.00028619999999999996, - "loss": 4.5099, + "epoch": 0.81, + "learning_rate": 0.00028739999999999994, + "loss": 4.8266, "step": 481 }, { - "epoch": 0.41, - "learning_rate": 0.0002868, - "loss": 4.5161, + "epoch": 0.81, + "learning_rate": 0.00028799999999999995, + "loss": 4.8246, "step": 482 }, { - "epoch": 0.41, - "learning_rate": 0.00028739999999999994, - "loss": 5.6125, + "epoch": 0.81, + "learning_rate": 0.00028859999999999997, + "loss": 4.9268, "step": 483 }, { - "epoch": 0.41, - "learning_rate": 0.00028799999999999995, - "loss": 4.8112, + "epoch": 0.81, + "learning_rate": 0.0002892, + "loss": 4.881, "step": 484 }, { - "epoch": 0.41, - "learning_rate": 0.00028859999999999997, - "loss": 4.8629, + "epoch": 0.82, + "learning_rate": 0.00028979999999999994, + "loss": 4.6537, "step": 485 }, { - "epoch": 0.41, - "learning_rate": 0.0002892, - "loss": 4.764, + "epoch": 0.82, + "learning_rate": 0.00029039999999999996, + "loss": 4.5566, "step": 486 }, { - "epoch": 0.41, - "learning_rate": 0.00028979999999999994, - "loss": 4.4091, + "epoch": 0.82, + "learning_rate": 0.00029099999999999997, + "loss": 4.5838, "step": 487 }, { - "epoch": 0.41, - "learning_rate": 0.00029039999999999996, - "loss": 4.6947, + "epoch": 0.82, + "learning_rate": 0.0002916, + "loss": 4.5166, "step": 488 }, { - "epoch": 0.41, - "learning_rate": 0.00029099999999999997, - "loss": 5.0368, + "epoch": 0.82, + "learning_rate": 0.00029219999999999995, + "loss": 4.6484, "step": 489 }, { - "epoch": 0.41, - "learning_rate": 0.0002916, - "loss": 4.567, + "epoch": 0.82, + "learning_rate": 0.00029279999999999996, + "loss": 4.3926, "step": 490 }, { - "epoch": 0.41, - "learning_rate": 0.00029219999999999995, - "loss": 5.1349, + "epoch": 0.83, + "learning_rate": 0.0002934, + "loss": 4.9063, "step": 491 }, - { - "epoch": 0.41, - "learning_rate": 0.00029279999999999996, - "loss": 4.7064, - "step": 492 - }, - { - "epoch": 0.41, - "learning_rate": 0.0002934, - "loss": 5.4823, - "step": 493 - }, - { - "epoch": 0.42, - "learning_rate": 0.000294, - "loss": 4.9341, - "step": 494 - }, - { - "epoch": 0.42, - "learning_rate": 0.00029459999999999995, - "loss": 5.279, - "step": 495 - }, - { - "epoch": 0.42, - "learning_rate": 0.00029519999999999997, - "loss": 4.8398, - "step": 496 - }, - { - "epoch": 0.42, - "learning_rate": 0.0002958, - "loss": 4.7063, - "step": 497 - }, - { - "epoch": 0.42, - "learning_rate": 0.0002964, - "loss": 4.7881, - "step": 498 - }, - { - "epoch": 0.42, - "learning_rate": 0.00029699999999999996, - "loss": 4.6494, - "step": 499 - }, - { - "epoch": 0.42, - "learning_rate": 0.00029759999999999997, - "loss": 4.833, - "step": 500 - }, - { - "epoch": 0.42, - "eval_loss": 5.194062232971191, - "eval_runtime": 747.3505, - "eval_samples_per_second": 3.535, - "eval_steps_per_second": 0.296, - "eval_wer": 1.9524395081316939, - "step": 500 - }, - { - "epoch": 0.42, - "learning_rate": 0.0002982, - "loss": 4.9335, - "step": 501 - }, - { - "epoch": 0.42, - "learning_rate": 0.0002988, - "loss": 4.9661, - "step": 502 - }, - { - "epoch": 0.42, - "learning_rate": 0.00029939999999999996, - "loss": 4.9686, - "step": 503 - }, - { - "epoch": 0.42, - "learning_rate": 0.0003, - "loss": 4.7973, - "step": 504 - }, - { - "epoch": 0.42, - "learning_rate": 0.0002995645863570392, - "loss": 4.7865, - "step": 505 - }, - { - "epoch": 0.43, - "learning_rate": 0.0002991291727140783, - "loss": 5.0596, - "step": 506 - }, - { - "epoch": 0.43, - "learning_rate": 0.00029869375907111753, - "loss": 4.8275, - "step": 507 - }, - { - "epoch": 0.43, - "learning_rate": 0.00029825834542815676, - "loss": 4.6202, - "step": 508 - }, - { - "epoch": 0.43, - "learning_rate": 0.00029782293178519587, - "loss": 4.6348, - "step": 509 - }, - { - "epoch": 0.43, - "learning_rate": 0.0002973875181422351, - "loss": 5.3447, - "step": 510 - }, - { - "epoch": 0.43, - "learning_rate": 0.00029695210449927426, - "loss": 5.0884, - "step": 511 - }, - { - "epoch": 0.43, - "learning_rate": 0.0002965166908563135, - "loss": 4.7513, - "step": 512 - }, - { - "epoch": 0.43, - "learning_rate": 0.00029608127721335266, - "loss": 4.877, - "step": 513 - }, - { - "epoch": 0.43, - "learning_rate": 0.0002956458635703918, - "loss": 4.4649, - "step": 514 - }, - { - "epoch": 0.43, - "learning_rate": 0.00029521044992743105, - "loss": 4.5183, - "step": 515 - }, - { - "epoch": 0.43, - "learning_rate": 0.0002947750362844702, - "loss": 5.2663, - "step": 516 - }, - { - "epoch": 0.43, - "learning_rate": 0.0002943396226415094, - "loss": 5.9906, - "step": 517 - }, - { - "epoch": 0.44, - "learning_rate": 0.0002939042089985486, - "loss": 4.666, - "step": 518 - }, - { - "epoch": 0.44, - "learning_rate": 0.0002934687953555878, - "loss": 5.5156, - "step": 519 - }, - { - "epoch": 0.44, - "learning_rate": 0.00029303338171262695, - "loss": 4.5457, - "step": 520 - }, - { - "epoch": 0.44, - "learning_rate": 0.00029259796806966617, - "loss": 5.0324, - "step": 521 - }, - { - "epoch": 0.44, - "learning_rate": 0.00029216255442670534, - "loss": 4.7233, - "step": 522 - }, - { - "epoch": 0.44, - "learning_rate": 0.0002917271407837445, - "loss": 5.6246, - "step": 523 - }, - { - "epoch": 0.44, - "learning_rate": 0.00029129172714078373, - "loss": 4.8244, - "step": 524 - }, - { - "epoch": 0.44, - "learning_rate": 0.0002908563134978229, - "loss": 4.9059, - "step": 525 - }, - { - "epoch": 0.44, - "learning_rate": 0.00029042089985486207, - "loss": 4.4507, - "step": 526 - }, - { - "epoch": 0.44, - "learning_rate": 0.0002899854862119013, - "loss": 5.0691, - "step": 527 - }, - { - "epoch": 0.44, - "learning_rate": 0.00028955007256894046, - "loss": 5.1142, - "step": 528 - }, - { - "epoch": 0.44, - "learning_rate": 0.0002891146589259796, - "loss": 4.7393, - "step": 529 - }, - { - "epoch": 0.45, - "learning_rate": 0.00028867924528301885, - "loss": 4.8084, - "step": 530 - }, - { - "epoch": 0.45, - "learning_rate": 0.000288243831640058, - "loss": 4.7766, - "step": 531 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002878084179970972, - "loss": 5.0283, - "step": 532 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002873730043541364, - "loss": 4.6411, - "step": 533 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002869375907111756, - "loss": 4.8474, - "step": 534 - }, - { - "epoch": 0.45, - "learning_rate": 0.00028650217706821475, - "loss": 4.5725, - "step": 535 - }, - { - "epoch": 0.45, - "learning_rate": 0.00028606676342525397, - "loss": 4.5082, - "step": 536 - }, - { - "epoch": 0.45, - "learning_rate": 0.00028563134978229314, - "loss": 4.8596, - "step": 537 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002851959361393323, - "loss": 5.065, - "step": 538 - }, - { - "epoch": 0.45, - "learning_rate": 0.00028476052249637153, - "loss": 4.7098, - "step": 539 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002843251088534107, - "loss": 5.1196, - "step": 540 - }, - { - "epoch": 0.45, - "learning_rate": 0.0002838896952104499, - "loss": 4.8682, - "step": 541 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002834542815674891, - "loss": 5.5616, - "step": 542 - }, - { - "epoch": 0.46, - "learning_rate": 0.00028301886792452826, - "loss": 4.684, - "step": 543 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002825834542815675, - "loss": 5.0173, - "step": 544 - }, - { - "epoch": 0.46, - "learning_rate": 0.00028214804063860665, - "loss": 5.1818, - "step": 545 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002817126269956458, - "loss": 4.6626, - "step": 546 - }, - { - "epoch": 0.46, - "learning_rate": 0.00028127721335268504, - "loss": 4.9806, - "step": 547 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002808417997097242, - "loss": 5.3829, - "step": 548 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002804063860667634, - "loss": 4.9341, - "step": 549 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002799709724238026, - "loss": 6.3752, - "step": 550 - }, - { - "epoch": 0.46, - "learning_rate": 0.0002795355587808418, - "loss": 4.9254, - "step": 551 - }, - { - "epoch": 0.46, - "learning_rate": 0.00027910014513788094, - "loss": 4.783, - "step": 552 - }, - { - "epoch": 0.46, - "learning_rate": 0.00027866473149492017, - "loss": 5.1896, - "step": 553 - }, - { - "epoch": 0.47, - "learning_rate": 0.00027822931785195933, - "loss": 4.8299, - "step": 554 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002777939042089985, - "loss": 4.9769, - "step": 555 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002773584905660377, - "loss": 4.8809, - "step": 556 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002769230769230769, - "loss": 4.5533, - "step": 557 - }, - { - "epoch": 0.47, - "learning_rate": 0.00027648766328011606, - "loss": 4.6023, - "step": 558 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002760522496371553, - "loss": 4.9475, - "step": 559 - }, - { - "epoch": 0.47, - "learning_rate": 0.00027561683599419446, - "loss": 4.9687, - "step": 560 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002751814223512336, - "loss": 4.69, - "step": 561 - }, - { - "epoch": 0.47, - "learning_rate": 0.00027474600870827285, - "loss": 4.7413, - "step": 562 - }, - { - "epoch": 0.47, - "learning_rate": 0.000274310595065312, - "loss": 4.9996, - "step": 563 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002738751814223512, - "loss": 4.3674, - "step": 564 - }, - { - "epoch": 0.47, - "learning_rate": 0.0002734397677793904, - "loss": 5.0086, - "step": 565 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002730043541364296, - "loss": 4.6329, - "step": 566 - }, - { - "epoch": 0.48, - "learning_rate": 0.00027256894049346875, - "loss": 4.7313, - "step": 567 - }, - { - "epoch": 0.48, - "learning_rate": 0.00027213352685050797, - "loss": 4.7225, - "step": 568 - }, - { - "epoch": 0.48, - "learning_rate": 0.00027169811320754714, - "loss": 4.6538, - "step": 569 - }, - { - "epoch": 0.48, - "learning_rate": 0.00027126269956458636, - "loss": 4.5962, - "step": 570 - }, - { - "epoch": 0.48, - "learning_rate": 0.00027082728592162553, - "loss": 4.6124, - "step": 571 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002703918722786647, - "loss": 5.0632, - "step": 572 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002699564586357039, - "loss": 4.5618, - "step": 573 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002695210449927431, - "loss": 4.836, - "step": 574 - }, - { - "epoch": 0.48, - "learning_rate": 0.00026908563134978226, - "loss": 4.9042, - "step": 575 - }, - { - "epoch": 0.48, - "learning_rate": 0.0002686502177068215, - "loss": 4.8837, - "step": 576 - }, - { - "epoch": 0.49, - "learning_rate": 0.00026821480406386065, - "loss": 5.2048, - "step": 577 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002677793904208998, - "loss": 4.6342, - "step": 578 - }, - { - "epoch": 0.49, - "learning_rate": 0.00026734397677793904, - "loss": 4.6749, - "step": 579 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002669085631349782, - "loss": 4.7553, - "step": 580 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002664731494920174, - "loss": 4.8569, - "step": 581 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002660377358490566, - "loss": 4.7583, - "step": 582 - }, - { - "epoch": 0.49, - "learning_rate": 0.00026560232220609577, - "loss": 4.7735, - "step": 583 - }, - { - "epoch": 0.49, - "learning_rate": 0.00026516690856313494, - "loss": 4.561, - "step": 584 - }, - { - "epoch": 0.49, - "learning_rate": 0.00026473149492017416, - "loss": 5.3574, - "step": 585 - }, - { - "epoch": 0.49, - "learning_rate": 0.00026429608127721333, - "loss": 4.7699, - "step": 586 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002638606676342525, - "loss": 4.9943, - "step": 587 - }, - { - "epoch": 0.49, - "learning_rate": 0.0002634252539912917, - "loss": 4.5543, - "step": 588 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002629898403483309, - "loss": 4.8902, - "step": 589 - }, - { - "epoch": 0.5, - "learning_rate": 0.00026255442670537006, - "loss": 4.8949, - "step": 590 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002621190130624093, - "loss": 4.6346, - "step": 591 - }, - { - "epoch": 0.5, - "learning_rate": 0.00026168359941944845, - "loss": 4.4538, - "step": 592 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002612481857764876, - "loss": 5.4038, - "step": 593 - }, - { - "epoch": 0.5, - "learning_rate": 0.00026081277213352684, - "loss": 4.6285, - "step": 594 - }, - { - "epoch": 0.5, - "learning_rate": 0.000260377358490566, - "loss": 4.6779, - "step": 595 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002599419448476052, - "loss": 4.552, - "step": 596 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002595065312046444, - "loss": 4.6429, - "step": 597 - }, - { - "epoch": 0.5, - "learning_rate": 0.0002590711175616836, - "loss": 5.7606, - "step": 598 - }, - { - "epoch": 0.5, - "learning_rate": 0.00025863570391872274, - "loss": 6.3581, - "step": 599 - }, - { - "epoch": 0.5, - "learning_rate": 0.00025820029027576197, - "loss": 6.8859, - "step": 600 - }, - { - "epoch": 0.51, - "learning_rate": 0.00025776487663280113, - "loss": 4.6724, - "step": 601 - }, - { - "epoch": 0.51, - "learning_rate": 0.00025732946298984036, - "loss": 4.9873, - "step": 602 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002568940493468795, - "loss": 5.0404, - "step": 603 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002564586357039187, - "loss": 4.778, - "step": 604 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002560232220609579, - "loss": 4.7111, - "step": 605 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002555878084179971, - "loss": 4.8009, - "step": 606 - }, - { - "epoch": 0.51, - "learning_rate": 0.00025515239477503626, - "loss": 4.6809, - "step": 607 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002547169811320755, - "loss": 4.8737, - "step": 608 - }, - { - "epoch": 0.51, - "learning_rate": 0.00025428156748911465, - "loss": 4.7996, - "step": 609 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002538461538461538, - "loss": 4.9332, - "step": 610 - }, - { - "epoch": 0.51, - "learning_rate": 0.00025341074020319304, - "loss": 4.8498, - "step": 611 - }, - { - "epoch": 0.51, - "learning_rate": 0.0002529753265602322, - "loss": 4.7182, - "step": 612 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002525399129172714, - "loss": 6.0865, - "step": 613 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002521044992743106, - "loss": 4.4399, - "step": 614 - }, - { - "epoch": 0.52, - "learning_rate": 0.00025166908563134977, - "loss": 4.6551, - "step": 615 - }, - { - "epoch": 0.52, - "learning_rate": 0.00025123367198838894, - "loss": 4.9617, - "step": 616 - }, - { - "epoch": 0.52, - "learning_rate": 0.00025079825834542816, - "loss": 4.8646, - "step": 617 - }, - { - "epoch": 0.52, - "learning_rate": 0.00025036284470246733, - "loss": 5.0519, - "step": 618 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002499274310595065, - "loss": 4.8651, - "step": 619 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002494920174165457, - "loss": 4.9591, - "step": 620 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002490566037735849, - "loss": 4.3819, - "step": 621 - }, - { - "epoch": 0.52, - "learning_rate": 0.00024862119013062406, - "loss": 4.329, - "step": 622 - }, - { - "epoch": 0.52, - "learning_rate": 0.0002481857764876633, - "loss": 5.9491, - "step": 623 - }, - { - "epoch": 0.52, - "learning_rate": 0.00024775036284470245, - "loss": 5.1706, - "step": 624 - }, - { - "epoch": 0.53, - "learning_rate": 0.0002473149492017416, - "loss": 5.0565, - "step": 625 - }, - { - "epoch": 0.53, - "learning_rate": 0.00024687953555878084, - "loss": 5.0496, - "step": 626 - }, - { - "epoch": 0.53, - "learning_rate": 0.00024644412191582, - "loss": 4.7394, - "step": 627 - }, - { - "epoch": 0.53, - "learning_rate": 0.0002460087082728592, - "loss": 4.4442, - "step": 628 - }, - { - "epoch": 0.53, - "learning_rate": 0.0002455732946298984, - "loss": 4.8163, - "step": 629 - }, - { - "epoch": 0.53, - "learning_rate": 0.00024513788098693757, - "loss": 4.612, - "step": 630 - }, - { - "epoch": 0.53, - "learning_rate": 0.0002447024673439768, - "loss": 4.9465, - "step": 631 - }, - { - "epoch": 0.53, - "learning_rate": 0.00024426705370101596, - "loss": 4.5676, - "step": 632 - }, - { - "epoch": 0.53, - "learning_rate": 0.00024383164005805513, - "loss": 4.994, - "step": 633 - }, - { - "epoch": 0.53, - "learning_rate": 0.00024339622641509433, - "loss": 4.6503, - "step": 634 - }, - { - "epoch": 0.53, - "learning_rate": 0.0002429608127721335, - "loss": 4.8686, - "step": 635 - }, - { - "epoch": 0.53, - "learning_rate": 0.0002425253991291727, - "loss": 4.9115, - "step": 636 - }, - { - "epoch": 0.54, - "learning_rate": 0.0002420899854862119, - "loss": 4.5344, - "step": 637 - }, - { - "epoch": 0.54, - "learning_rate": 0.00024165457184325106, - "loss": 4.5513, - "step": 638 - }, - { - "epoch": 0.54, - "learning_rate": 0.00024121915820029025, - "loss": 4.5369, - "step": 639 - }, - { - "epoch": 0.54, - "learning_rate": 0.00024078374455732945, - "loss": 5.0302, - "step": 640 - }, - { - "epoch": 0.54, - "learning_rate": 0.00024034833091436862, - "loss": 4.6188, - "step": 641 - }, - { - "epoch": 0.54, - "learning_rate": 0.0002399129172714078, - "loss": 5.3999, - "step": 642 - }, - { - "epoch": 0.54, - "learning_rate": 0.000239477503628447, - "loss": 5.2607, - "step": 643 - }, - { - "epoch": 0.54, - "learning_rate": 0.0002390420899854862, - "loss": 4.7053, - "step": 644 - }, - { - "epoch": 0.54, - "learning_rate": 0.00023860667634252537, - "loss": 4.7037, - "step": 645 - }, - { - "epoch": 0.54, - "learning_rate": 0.00023817126269956457, - "loss": 4.5346, - "step": 646 - }, - { - "epoch": 0.54, - "learning_rate": 0.00023773584905660377, - "loss": 4.3213, - "step": 647 - }, - { - "epoch": 0.54, - "learning_rate": 0.00023730043541364293, - "loss": 5.9242, - "step": 648 - }, - { - "epoch": 0.55, - "learning_rate": 0.00023686502177068213, - "loss": 5.0213, - "step": 649 - }, - { - "epoch": 0.55, - "learning_rate": 0.00023642960812772133, - "loss": 5.1531, - "step": 650 - }, - { - "epoch": 0.55, - "learning_rate": 0.0002359941944847605, - "loss": 4.8283, - "step": 651 - }, - { - "epoch": 0.55, - "learning_rate": 0.0002355587808417997, - "loss": 4.6261, - "step": 652 - }, - { - "epoch": 0.55, - "learning_rate": 0.00023512336719883889, - "loss": 4.6689, - "step": 653 - }, - { - "epoch": 0.55, - "learning_rate": 0.00023468795355587805, - "loss": 5.0152, - "step": 654 - }, - { - "epoch": 0.55, - "learning_rate": 0.00023425253991291725, - "loss": 4.5758, - "step": 655 - }, - { - "epoch": 0.55, - "learning_rate": 0.00023381712626995645, - "loss": 4.5943, - "step": 656 - }, - { - "epoch": 0.55, - "learning_rate": 0.00023338171262699562, - "loss": 4.6639, - "step": 657 - }, - { - "epoch": 0.55, - "learning_rate": 0.0002329462989840348, - "loss": 4.8844, - "step": 658 - }, - { - "epoch": 0.55, - "learning_rate": 0.000232510885341074, - "loss": 4.9718, - "step": 659 - }, - { - "epoch": 0.55, - "learning_rate": 0.0002320754716981132, - "loss": 5.0065, - "step": 660 - }, - { - "epoch": 0.56, - "learning_rate": 0.00023164005805515237, - "loss": 4.713, - "step": 661 - }, - { - "epoch": 0.56, - "learning_rate": 0.00023120464441219157, - "loss": 4.4193, - "step": 662 - }, - { - "epoch": 0.56, - "learning_rate": 0.00023076923076923076, - "loss": 4.6658, - "step": 663 - }, - { - "epoch": 0.56, - "learning_rate": 0.00023033381712626993, - "loss": 4.5546, - "step": 664 - }, - { - "epoch": 0.56, - "learning_rate": 0.00022989840348330913, - "loss": 4.9011, - "step": 665 - }, - { - "epoch": 0.56, - "learning_rate": 0.00022946298984034832, - "loss": 4.826, - "step": 666 - }, - { - "epoch": 0.56, - "learning_rate": 0.0002290275761973875, - "loss": 4.7478, - "step": 667 - }, - { - "epoch": 0.56, - "learning_rate": 0.0002285921625544267, - "loss": 4.6145, - "step": 668 - }, - { - "epoch": 0.56, - "learning_rate": 0.00022815674891146588, - "loss": 5.2825, - "step": 669 - }, - { - "epoch": 0.56, - "learning_rate": 0.00022772133526850505, - "loss": 4.8788, - "step": 670 - }, - { - "epoch": 0.56, - "learning_rate": 0.00022728592162554425, - "loss": 4.7793, - "step": 671 - }, - { - "epoch": 0.56, - "learning_rate": 0.00022685050798258345, - "loss": 5.0238, - "step": 672 - }, - { - "epoch": 0.57, - "learning_rate": 0.00022641509433962264, - "loss": 4.7159, - "step": 673 - }, - { - "epoch": 0.57, - "learning_rate": 0.0002259796806966618, - "loss": 4.7365, - "step": 674 - }, - { - "epoch": 0.57, - "learning_rate": 0.000225544267053701, - "loss": 4.5953, - "step": 675 - }, - { - "epoch": 0.57, - "learning_rate": 0.0002251088534107402, - "loss": 4.844, - "step": 676 - }, - { - "epoch": 0.57, - "learning_rate": 0.00022467343976777937, - "loss": 4.8889, - "step": 677 - }, - { - "epoch": 0.57, - "learning_rate": 0.00022423802612481857, - "loss": 4.7826, - "step": 678 - }, - { - "epoch": 0.57, - "learning_rate": 0.00022380261248185776, - "loss": 4.8066, - "step": 679 - }, - { - "epoch": 0.57, - "learning_rate": 0.00022336719883889693, - "loss": 4.8016, - "step": 680 - }, - { - "epoch": 0.57, - "learning_rate": 0.00022293178519593613, - "loss": 5.1085, - "step": 681 - }, - { - "epoch": 0.57, - "learning_rate": 0.00022249637155297532, - "loss": 4.293, - "step": 682 - }, - { - "epoch": 0.57, - "learning_rate": 0.0002220609579100145, - "loss": 4.7588, - "step": 683 - }, - { - "epoch": 0.58, - "learning_rate": 0.0002216255442670537, - "loss": 4.5345, - "step": 684 - }, - { - "epoch": 0.58, - "learning_rate": 0.00022119013062409288, - "loss": 4.9384, - "step": 685 - }, - { - "epoch": 0.58, - "learning_rate": 0.00022075471698113205, - "loss": 5.1311, - "step": 686 - }, - { - "epoch": 0.58, - "learning_rate": 0.00022031930333817125, - "loss": 4.5077, - "step": 687 - }, - { - "epoch": 0.58, - "learning_rate": 0.00021988388969521044, - "loss": 4.6148, - "step": 688 - }, - { - "epoch": 0.58, - "learning_rate": 0.00021944847605224964, - "loss": 4.7591, - "step": 689 - }, - { - "epoch": 0.58, - "learning_rate": 0.0002190130624092888, - "loss": 5.0202, - "step": 690 - }, - { - "epoch": 0.58, - "learning_rate": 0.000218577648766328, - "loss": 4.8682, - "step": 691 - }, - { - "epoch": 0.58, - "learning_rate": 0.0002181422351233672, - "loss": 4.6642, - "step": 692 - }, - { - "epoch": 0.58, - "learning_rate": 0.00021770682148040637, - "loss": 4.8987, - "step": 693 - }, - { - "epoch": 0.58, - "learning_rate": 0.00021727140783744556, - "loss": 5.5827, - "step": 694 - }, - { - "epoch": 0.58, - "learning_rate": 0.00021683599419448476, - "loss": 4.66, - "step": 695 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021640058055152393, - "loss": 4.6692, - "step": 696 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021596516690856313, - "loss": 4.7535, - "step": 697 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021552975326560232, - "loss": 5.7668, - "step": 698 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021509433962264146, - "loss": 4.7838, - "step": 699 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021465892597968069, - "loss": 5.2674, - "step": 700 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021422351233671988, - "loss": 5.1239, - "step": 701 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021378809869375908, - "loss": 4.4763, - "step": 702 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021335268505079825, - "loss": 4.6522, - "step": 703 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021291727140783744, - "loss": 4.6981, - "step": 704 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021248185776487664, - "loss": 4.4445, - "step": 705 - }, - { - "epoch": 0.59, - "learning_rate": 0.00021204644412191578, - "loss": 4.6095, - "step": 706 - }, - { - "epoch": 0.59, - "learning_rate": 0.000211611030478955, - "loss": 4.5051, - "step": 707 - }, - { - "epoch": 0.6, - "learning_rate": 0.0002111756168359942, - "loss": 4.9519, - "step": 708 - }, - { - "epoch": 0.6, - "learning_rate": 0.00021074020319303334, - "loss": 4.5414, - "step": 709 - }, - { - "epoch": 0.6, - "learning_rate": 0.00021030478955007256, - "loss": 4.7941, - "step": 710 - }, - { - "epoch": 0.6, - "learning_rate": 0.00020986937590711176, - "loss": 4.3142, - "step": 711 - }, - { - "epoch": 0.6, - "learning_rate": 0.0002094339622641509, - "loss": 4.7656, - "step": 712 - }, - { - "epoch": 0.6, - "learning_rate": 0.0002089985486211901, - "loss": 4.8588, - "step": 713 - }, - { - "epoch": 0.6, - "learning_rate": 0.00020856313497822932, - "loss": 4.4229, - "step": 714 - }, - { - "epoch": 0.6, - "learning_rate": 0.00020812772133526846, - "loss": 4.9272, - "step": 715 - }, - { - "epoch": 0.6, - "learning_rate": 0.00020769230769230766, - "loss": 4.5207, - "step": 716 - }, - { - "epoch": 0.6, - "learning_rate": 0.00020725689404934685, - "loss": 4.4965, - "step": 717 - }, - { - "epoch": 0.6, - "learning_rate": 0.00020682148040638608, - "loss": 4.8384, - "step": 718 - }, - { - "epoch": 0.6, - "learning_rate": 0.00020638606676342522, - "loss": 4.5704, - "step": 719 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020595065312046441, - "loss": 4.487, - "step": 720 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020551523947750364, - "loss": 4.4206, - "step": 721 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020507982583454278, - "loss": 4.3743, - "step": 722 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020464441219158197, - "loss": 5.2636, - "step": 723 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020420899854862117, - "loss": 4.4234, - "step": 724 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020377358490566034, - "loss": 4.6968, - "step": 725 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020333817126269953, - "loss": 5.4371, - "step": 726 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020290275761973873, - "loss": 4.4697, - "step": 727 - }, - { - "epoch": 0.61, - "learning_rate": 0.0002024673439767779, - "loss": 4.819, - "step": 728 - }, - { - "epoch": 0.61, - "learning_rate": 0.0002020319303338171, - "loss": 4.3262, - "step": 729 - }, - { - "epoch": 0.61, - "learning_rate": 0.0002015965166908563, - "loss": 4.698, - "step": 730 - }, - { - "epoch": 0.61, - "learning_rate": 0.00020116110304789546, - "loss": 4.8936, - "step": 731 - }, - { - "epoch": 0.62, - "learning_rate": 0.00020072568940493466, - "loss": 5.0859, - "step": 732 - }, - { - "epoch": 0.62, - "learning_rate": 0.00020029027576197385, - "loss": 4.4665, - "step": 733 - }, - { - "epoch": 0.62, - "learning_rate": 0.00019985486211901305, - "loss": 4.5697, - "step": 734 - }, - { - "epoch": 0.62, - "learning_rate": 0.00019941944847605222, - "loss": 5.1158, - "step": 735 - }, - { - "epoch": 0.62, - "learning_rate": 0.0001989840348330914, - "loss": 4.5841, - "step": 736 - }, - { - "epoch": 0.62, - "learning_rate": 0.0001985486211901306, - "loss": 4.6718, - "step": 737 - }, - { - "epoch": 0.62, - "learning_rate": 0.00019811320754716978, - "loss": 4.9392, - "step": 738 - }, - { - "epoch": 0.62, - "learning_rate": 0.00019767779390420897, - "loss": 4.9908, - "step": 739 - }, - { - "epoch": 0.62, - "learning_rate": 0.00019724238026124817, - "loss": 4.5999, - "step": 740 - }, - { - "epoch": 0.62, - "learning_rate": 0.00019680696661828734, - "loss": 4.8487, - "step": 741 - }, - { - "epoch": 0.62, - "learning_rate": 0.00019637155297532653, - "loss": 5.2399, - "step": 742 - }, - { - "epoch": 0.62, - "learning_rate": 0.00019593613933236573, - "loss": 4.5827, - "step": 743 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001955007256894049, - "loss": 4.6606, - "step": 744 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001950653120464441, - "loss": 5.3387, - "step": 745 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001946298984034833, - "loss": 5.545, - "step": 746 - }, - { - "epoch": 0.63, - "learning_rate": 0.00019419448476052249, - "loss": 5.0344, - "step": 747 - }, - { - "epoch": 0.63, - "learning_rate": 0.00019375907111756165, - "loss": 5.9759, - "step": 748 - }, - { - "epoch": 0.63, - "learning_rate": 0.00019332365747460085, - "loss": 4.5979, - "step": 749 - }, - { - "epoch": 0.63, - "learning_rate": 0.00019288824383164005, - "loss": 4.7675, - "step": 750 - }, - { - "epoch": 0.63, - "learning_rate": 0.00019245283018867922, - "loss": 4.6892, - "step": 751 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001920174165457184, - "loss": 4.8701, - "step": 752 - }, - { - "epoch": 0.63, - "learning_rate": 0.0001915820029027576, - "loss": 4.6816, - "step": 753 - }, - { - "epoch": 0.63, - "learning_rate": 0.00019114658925979678, - "loss": 4.3436, - "step": 754 - }, - { - "epoch": 0.63, - "learning_rate": 0.00019071117561683597, - "loss": 4.7221, - "step": 755 - }, - { - "epoch": 0.64, - "learning_rate": 0.00019027576197387517, - "loss": 4.4863, - "step": 756 - }, - { - "epoch": 0.64, - "learning_rate": 0.00018984034833091434, - "loss": 4.4107, - "step": 757 - }, - { - "epoch": 0.64, - "learning_rate": 0.00018940493468795353, - "loss": 4.9736, - "step": 758 - }, - { - "epoch": 0.64, - "learning_rate": 0.00018896952104499273, - "loss": 4.3843, - "step": 759 - }, - { - "epoch": 0.64, - "learning_rate": 0.0001885341074020319, - "loss": 4.7796, - "step": 760 - }, - { - "epoch": 0.64, - "learning_rate": 0.0001880986937590711, - "loss": 4.3747, - "step": 761 - }, - { - "epoch": 0.64, - "learning_rate": 0.0001876632801161103, - "loss": 4.4827, - "step": 762 - }, - { - "epoch": 0.64, - "learning_rate": 0.00018722786647314948, - "loss": 4.5165, - "step": 763 - }, - { - "epoch": 0.64, - "learning_rate": 0.00018679245283018865, - "loss": 4.4969, - "step": 764 - }, - { - "epoch": 0.64, - "learning_rate": 0.00018635703918722785, - "loss": 4.7664, - "step": 765 - }, - { - "epoch": 0.64, - "learning_rate": 0.00018592162554426704, - "loss": 4.4828, - "step": 766 - }, - { - "epoch": 0.64, - "learning_rate": 0.00018548621190130621, - "loss": 4.7631, - "step": 767 - }, - { - "epoch": 0.65, - "learning_rate": 0.0001850507982583454, - "loss": 4.2864, - "step": 768 - }, - { - "epoch": 0.65, - "learning_rate": 0.0001846153846153846, - "loss": 4.6127, - "step": 769 - }, - { - "epoch": 0.65, - "learning_rate": 0.00018417997097242377, - "loss": 4.6952, - "step": 770 - }, - { - "epoch": 0.65, - "learning_rate": 0.00018374455732946297, - "loss": 4.8167, - "step": 771 - }, - { - "epoch": 0.65, - "learning_rate": 0.00018330914368650217, - "loss": 4.6103, - "step": 772 - }, - { - "epoch": 0.65, - "learning_rate": 0.00018287373004354133, - "loss": 4.5139, - "step": 773 - }, - { - "epoch": 0.65, - "learning_rate": 0.00018243831640058053, - "loss": 4.5599, - "step": 774 - }, - { - "epoch": 0.65, - "learning_rate": 0.00018200290275761973, - "loss": 4.8973, - "step": 775 - }, - { - "epoch": 0.65, - "learning_rate": 0.00018156748911465892, - "loss": 4.4198, - "step": 776 - }, - { - "epoch": 0.65, - "learning_rate": 0.0001811320754716981, - "loss": 4.5592, - "step": 777 - }, - { - "epoch": 0.65, - "learning_rate": 0.0001806966618287373, - "loss": 4.8103, - "step": 778 - }, - { - "epoch": 0.65, - "learning_rate": 0.00018026124818577648, - "loss": 4.4989, - "step": 779 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017982583454281565, - "loss": 4.7375, - "step": 780 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017939042089985485, - "loss": 4.5348, - "step": 781 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017895500725689404, - "loss": 4.7231, - "step": 782 - }, - { - "epoch": 0.66, - "learning_rate": 0.0001785195936139332, - "loss": 4.6088, - "step": 783 - }, - { - "epoch": 0.66, - "learning_rate": 0.0001780841799709724, - "loss": 4.8997, - "step": 784 - }, - { - "epoch": 0.66, - "learning_rate": 0.0001776487663280116, - "loss": 4.9077, - "step": 785 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017721335268505077, - "loss": 4.7907, - "step": 786 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017677793904208997, - "loss": 4.8741, - "step": 787 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017634252539912916, - "loss": 4.9389, - "step": 788 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017590711175616833, - "loss": 4.4176, - "step": 789 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017547169811320753, - "loss": 4.3583, - "step": 790 - }, - { - "epoch": 0.66, - "learning_rate": 0.00017503628447024673, - "loss": 4.8908, - "step": 791 - }, - { - "epoch": 0.67, - "learning_rate": 0.00017460087082728592, - "loss": 4.8744, - "step": 792 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001741654571843251, - "loss": 5.2567, - "step": 793 - }, - { - "epoch": 0.67, - "learning_rate": 0.00017373004354136429, - "loss": 4.4392, - "step": 794 - }, - { - "epoch": 0.67, - "learning_rate": 0.00017329462989840348, - "loss": 4.7618, - "step": 795 - }, - { - "epoch": 0.67, - "learning_rate": 0.00017285921625544265, - "loss": 4.2707, - "step": 796 - }, - { - "epoch": 0.67, - "learning_rate": 0.00017242380261248185, - "loss": 4.4303, - "step": 797 - }, - { - "epoch": 0.67, - "learning_rate": 0.00017198838896952104, - "loss": 4.8627, - "step": 798 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001715529753265602, - "loss": 5.733, - "step": 799 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001711175616835994, - "loss": 4.6037, - "step": 800 - }, - { - "epoch": 0.67, - "learning_rate": 0.0001706821480406386, - "loss": 5.4275, - "step": 801 - }, - { - "epoch": 0.67, - "learning_rate": 0.00017024673439767777, - "loss": 4.9265, - "step": 802 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016981132075471697, - "loss": 4.5584, - "step": 803 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016937590711175616, - "loss": 4.7239, - "step": 804 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016894049346879536, - "loss": 4.3661, - "step": 805 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016850507982583453, - "loss": 4.5606, - "step": 806 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016806966618287372, - "loss": 4.9325, - "step": 807 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016763425253991292, - "loss": 4.6627, - "step": 808 - }, - { - "epoch": 0.68, - "learning_rate": 0.0001671988388969521, - "loss": 4.933, - "step": 809 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016676342525399128, - "loss": 5.0272, - "step": 810 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016632801161103048, - "loss": 4.7472, - "step": 811 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016589259796806965, - "loss": 4.8959, - "step": 812 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016545718432510884, - "loss": 4.4994, - "step": 813 - }, - { - "epoch": 0.68, - "learning_rate": 0.00016502177068214804, - "loss": 5.1709, - "step": 814 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001645863570391872, - "loss": 4.6819, - "step": 815 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001641509433962264, - "loss": 4.7547, - "step": 816 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001637155297532656, - "loss": 4.5043, - "step": 817 - }, - { - "epoch": 0.69, - "learning_rate": 0.00016328011611030477, - "loss": 4.2965, - "step": 818 - }, - { - "epoch": 0.69, - "learning_rate": 0.00016284470246734397, - "loss": 4.3956, - "step": 819 - }, - { - "epoch": 0.69, - "learning_rate": 0.00016240928882438316, - "loss": 4.6692, - "step": 820 - }, - { - "epoch": 0.69, - "learning_rate": 0.00016197387518142236, - "loss": 4.7531, - "step": 821 - }, - { - "epoch": 0.69, - "learning_rate": 0.00016153846153846153, - "loss": 4.4669, - "step": 822 - }, - { - "epoch": 0.69, - "learning_rate": 0.00016110304789550072, - "loss": 4.6101, - "step": 823 - }, - { - "epoch": 0.69, - "learning_rate": 0.00016066763425253992, - "loss": 5.0766, - "step": 824 - }, - { - "epoch": 0.69, - "learning_rate": 0.0001602322206095791, - "loss": 4.725, - "step": 825 - }, - { - "epoch": 0.69, - "learning_rate": 0.00015979680696661828, - "loss": 4.9691, - "step": 826 - }, - { - "epoch": 0.7, - "learning_rate": 0.00015936139332365748, - "loss": 4.5378, - "step": 827 - }, - { - "epoch": 0.7, - "learning_rate": 0.00015892597968069665, - "loss": 4.6755, - "step": 828 - }, - { - "epoch": 0.7, - "learning_rate": 0.00015849056603773584, - "loss": 4.8338, - "step": 829 - }, - { - "epoch": 0.7, - "learning_rate": 0.00015805515239477504, - "loss": 4.8006, - "step": 830 - }, - { - "epoch": 0.7, - "learning_rate": 0.0001576197387518142, - "loss": 4.6393, - "step": 831 - }, - { - "epoch": 0.7, - "learning_rate": 0.0001571843251088534, - "loss": 4.6717, - "step": 832 - }, - { - "epoch": 0.7, - "learning_rate": 0.0001567489114658926, - "loss": 4.3868, - "step": 833 - }, - { - "epoch": 0.7, - "learning_rate": 0.0001563134978229318, - "loss": 4.5879, - "step": 834 - }, - { - "epoch": 0.7, - "learning_rate": 0.00015587808417997096, - "loss": 4.4318, - "step": 835 - }, - { - "epoch": 0.7, - "learning_rate": 0.00015544267053701016, - "loss": 4.6803, - "step": 836 - }, - { - "epoch": 0.7, - "learning_rate": 0.00015500725689404936, - "loss": 4.5354, - "step": 837 - }, - { - "epoch": 0.7, - "learning_rate": 0.00015457184325108852, - "loss": 4.3185, - "step": 838 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015413642960812772, - "loss": 4.5848, - "step": 839 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015370101596516692, - "loss": 4.4021, - "step": 840 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015326560232220606, - "loss": 4.5211, - "step": 841 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015283018867924528, - "loss": 5.1432, - "step": 842 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015239477503628448, - "loss": 5.2663, - "step": 843 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015195936139332362, - "loss": 5.1663, - "step": 844 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015152394775036284, - "loss": 4.4686, - "step": 845 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015108853410740204, - "loss": 4.3942, - "step": 846 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015065312046444118, - "loss": 4.5995, - "step": 847 - }, - { - "epoch": 0.71, - "learning_rate": 0.00015021770682148038, - "loss": 4.3086, - "step": 848 - }, - { - "epoch": 0.71, - "learning_rate": 0.0001497822931785196, - "loss": 4.7466, - "step": 849 - }, - { - "epoch": 0.71, - "learning_rate": 0.00014934687953555877, - "loss": 4.8963, - "step": 850 - }, - { - "epoch": 0.72, - "learning_rate": 0.00014891146589259794, - "loss": 5.0766, - "step": 851 - }, - { - "epoch": 0.72, - "learning_rate": 0.00014847605224963713, - "loss": 4.5171, - "step": 852 - }, - { - "epoch": 0.72, - "learning_rate": 0.00014804063860667633, - "loss": 4.7791, - "step": 853 - }, - { - "epoch": 0.72, - "learning_rate": 0.00014760522496371552, - "loss": 4.4509, - "step": 854 - }, - { - "epoch": 0.72, - "learning_rate": 0.0001471698113207547, - "loss": 4.5231, - "step": 855 - }, - { - "epoch": 0.72, - "learning_rate": 0.0001467343976777939, - "loss": 4.6349, - "step": 856 - }, - { - "epoch": 0.72, - "learning_rate": 0.00014629898403483308, - "loss": 4.4544, - "step": 857 - }, - { - "epoch": 0.72, - "learning_rate": 0.00014586357039187225, - "loss": 4.3388, - "step": 858 - }, - { - "epoch": 0.72, - "learning_rate": 0.00014542815674891145, - "loss": 4.5616, - "step": 859 - }, - { - "epoch": 0.72, - "learning_rate": 0.00014499274310595064, - "loss": 4.6136, - "step": 860 - }, - { - "epoch": 0.72, - "learning_rate": 0.0001445573294629898, - "loss": 4.1963, - "step": 861 - }, - { - "epoch": 0.72, - "learning_rate": 0.000144121915820029, - "loss": 4.5643, - "step": 862 - }, - { - "epoch": 0.73, - "learning_rate": 0.0001436865021770682, - "loss": 4.4824, - "step": 863 - }, - { - "epoch": 0.73, - "learning_rate": 0.00014325108853410737, - "loss": 4.5173, - "step": 864 - }, - { - "epoch": 0.73, - "learning_rate": 0.00014281567489114657, - "loss": 4.4166, - "step": 865 - }, - { - "epoch": 0.73, - "learning_rate": 0.00014238026124818577, - "loss": 4.4031, - "step": 866 - }, - { - "epoch": 0.73, - "learning_rate": 0.00014194484760522496, - "loss": 4.2849, - "step": 867 - }, - { - "epoch": 0.73, - "learning_rate": 0.00014150943396226413, - "loss": 4.4519, - "step": 868 - }, - { - "epoch": 0.73, - "learning_rate": 0.00014107402031930333, - "loss": 4.8794, - "step": 869 - }, - { - "epoch": 0.73, - "learning_rate": 0.00014063860667634252, - "loss": 4.1686, - "step": 870 - }, - { - "epoch": 0.73, - "learning_rate": 0.0001402031930333817, - "loss": 4.1047, - "step": 871 - }, - { - "epoch": 0.73, - "learning_rate": 0.0001397677793904209, - "loss": 4.4521, - "step": 872 - }, - { - "epoch": 0.73, - "learning_rate": 0.00013933236574746008, - "loss": 4.7197, - "step": 873 - }, - { - "epoch": 0.73, - "learning_rate": 0.00013889695210449925, - "loss": 4.439, - "step": 874 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013846153846153845, - "loss": 4.6244, - "step": 875 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013802612481857764, - "loss": 4.2425, - "step": 876 - }, - { - "epoch": 0.74, - "learning_rate": 0.0001375907111756168, - "loss": 4.508, - "step": 877 - }, - { - "epoch": 0.74, - "learning_rate": 0.000137155297532656, - "loss": 4.7327, - "step": 878 - }, - { - "epoch": 0.74, - "learning_rate": 0.0001367198838896952, - "loss": 4.3981, - "step": 879 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013628447024673437, - "loss": 4.5901, - "step": 880 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013584905660377357, - "loss": 4.3026, - "step": 881 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013541364296081276, - "loss": 4.745, - "step": 882 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013497822931785196, - "loss": 4.4287, - "step": 883 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013454281567489113, - "loss": 4.421, - "step": 884 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013410740203193032, - "loss": 4.5041, - "step": 885 - }, - { - "epoch": 0.74, - "learning_rate": 0.00013367198838896952, - "loss": 4.484, - "step": 886 - }, - { - "epoch": 0.75, - "learning_rate": 0.0001332365747460087, - "loss": 4.6316, - "step": 887 - }, - { - "epoch": 0.75, - "learning_rate": 0.00013280116110304789, - "loss": 4.5471, - "step": 888 - }, - { - "epoch": 0.75, - "learning_rate": 0.00013236574746008708, - "loss": 4.901, - "step": 889 - }, - { - "epoch": 0.75, - "learning_rate": 0.00013193033381712625, - "loss": 4.8479, - "step": 890 - }, - { - "epoch": 0.75, - "learning_rate": 0.00013149492017416545, - "loss": 5.0569, - "step": 891 - }, - { - "epoch": 0.75, - "learning_rate": 0.00013105950653120464, - "loss": 4.5231, - "step": 892 - }, - { - "epoch": 0.75, - "learning_rate": 0.0001306240928882438, - "loss": 4.5042, - "step": 893 - }, - { - "epoch": 0.75, - "learning_rate": 0.000130188679245283, - "loss": 4.9721, - "step": 894 - }, - { - "epoch": 0.75, - "learning_rate": 0.0001297532656023222, - "loss": 4.4415, - "step": 895 - }, - { - "epoch": 0.75, - "learning_rate": 0.00012931785195936137, - "loss": 4.7974, - "step": 896 - }, - { - "epoch": 0.75, - "learning_rate": 0.00012888243831640057, - "loss": 4.4361, - "step": 897 - }, - { - "epoch": 0.75, - "learning_rate": 0.00012844702467343976, - "loss": 4.6191, - "step": 898 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012801161103047896, - "loss": 4.5298, - "step": 899 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012757619738751813, - "loss": 4.6714, - "step": 900 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012714078374455732, - "loss": 4.4432, - "step": 901 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012670537010159652, - "loss": 4.5082, - "step": 902 - }, - { - "epoch": 0.76, - "learning_rate": 0.0001262699564586357, - "loss": 4.5454, - "step": 903 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012583454281567488, - "loss": 4.9197, - "step": 904 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012539912917271408, - "loss": 4.4218, - "step": 905 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012496371552975325, - "loss": 4.8506, - "step": 906 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012452830188679244, - "loss": 4.5414, - "step": 907 - }, - { - "epoch": 0.76, - "learning_rate": 0.00012409288824383164, - "loss": 5.0392, - "step": 908 - }, - { - "epoch": 0.76, - "learning_rate": 0.0001236574746008708, - "loss": 4.4776, - "step": 909 - }, - { - "epoch": 0.77, - "learning_rate": 0.00012322206095791, - "loss": 4.8287, - "step": 910 - }, - { - "epoch": 0.77, - "learning_rate": 0.0001227866473149492, - "loss": 4.3984, - "step": 911 - }, - { - "epoch": 0.77, - "learning_rate": 0.0001223512336719884, - "loss": 4.3385, - "step": 912 - }, - { - "epoch": 0.77, - "learning_rate": 0.00012191582002902757, - "loss": 4.3623, - "step": 913 - }, - { - "epoch": 0.77, - "learning_rate": 0.00012148040638606675, - "loss": 4.3756, - "step": 914 - }, - { - "epoch": 0.77, - "learning_rate": 0.00012104499274310594, - "loss": 4.7014, - "step": 915 - }, - { - "epoch": 0.77, - "learning_rate": 0.00012060957910014513, - "loss": 4.699, - "step": 916 - }, - { - "epoch": 0.77, - "learning_rate": 0.00012017416545718431, - "loss": 4.4153, - "step": 917 - }, - { - "epoch": 0.77, - "learning_rate": 0.0001197387518142235, - "loss": 4.4669, - "step": 918 - }, - { - "epoch": 0.77, - "learning_rate": 0.00011930333817126269, - "loss": 4.7014, - "step": 919 - }, - { - "epoch": 0.77, - "learning_rate": 0.00011886792452830188, - "loss": 4.6546, - "step": 920 - }, - { - "epoch": 0.77, - "learning_rate": 0.00011843251088534106, - "loss": 4.3518, - "step": 921 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011799709724238025, - "loss": 4.7564, - "step": 922 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011756168359941944, - "loss": 4.3889, - "step": 923 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011712626995645863, - "loss": 4.7127, - "step": 924 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011669085631349781, - "loss": 4.354, - "step": 925 - }, - { - "epoch": 0.78, - "learning_rate": 0.000116255442670537, - "loss": 4.3408, - "step": 926 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011582002902757619, - "loss": 4.6165, - "step": 927 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011538461538461538, - "loss": 4.4945, - "step": 928 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011494920174165456, - "loss": 4.3871, - "step": 929 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011451378809869375, - "loss": 4.2537, - "step": 930 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011407837445573294, - "loss": 4.7368, - "step": 931 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011364296081277212, - "loss": 4.7466, - "step": 932 - }, - { - "epoch": 0.78, - "learning_rate": 0.00011320754716981132, - "loss": 4.4367, - "step": 933 - }, - { - "epoch": 0.79, - "learning_rate": 0.0001127721335268505, - "loss": 4.5984, - "step": 934 - }, - { - "epoch": 0.79, - "learning_rate": 0.00011233671988388969, - "loss": 4.6279, - "step": 935 - }, - { - "epoch": 0.79, - "learning_rate": 0.00011190130624092888, - "loss": 4.4899, - "step": 936 - }, - { - "epoch": 0.79, - "learning_rate": 0.00011146589259796806, - "loss": 4.5431, - "step": 937 - }, - { - "epoch": 0.79, - "learning_rate": 0.00011103047895500725, - "loss": 4.5151, - "step": 938 - }, - { - "epoch": 0.79, - "learning_rate": 0.00011059506531204644, - "loss": 4.4568, - "step": 939 - }, - { - "epoch": 0.79, - "learning_rate": 0.00011015965166908562, - "loss": 4.644, - "step": 940 - }, - { - "epoch": 0.79, - "learning_rate": 0.00010972423802612482, - "loss": 4.4773, - "step": 941 - }, - { - "epoch": 0.79, - "learning_rate": 0.000109288824383164, - "loss": 4.5586, - "step": 942 - }, - { - "epoch": 0.79, - "learning_rate": 0.00010885341074020318, - "loss": 4.1504, - "step": 943 - }, - { - "epoch": 0.79, - "learning_rate": 0.00010841799709724238, - "loss": 4.3842, - "step": 944 - }, - { - "epoch": 0.79, - "learning_rate": 0.00010798258345428156, - "loss": 4.5864, - "step": 945 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010754716981132073, - "loss": 4.3185, - "step": 946 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010711175616835994, - "loss": 4.4693, - "step": 947 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010667634252539912, - "loss": 4.3239, - "step": 948 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010624092888243832, - "loss": 4.4073, - "step": 949 - }, - { - "epoch": 0.8, - "learning_rate": 0.0001058055152394775, - "loss": 4.4021, - "step": 950 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010537010159651667, - "loss": 4.2928, - "step": 951 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010493468795355588, - "loss": 4.7388, - "step": 952 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010449927431059505, - "loss": 4.2615, - "step": 953 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010406386066763423, - "loss": 4.1396, - "step": 954 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010362844702467343, - "loss": 4.36, - "step": 955 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010319303338171261, - "loss": 4.6646, - "step": 956 - }, - { - "epoch": 0.8, - "learning_rate": 0.00010275761973875182, - "loss": 4.6852, - "step": 957 - }, - { - "epoch": 0.81, - "learning_rate": 0.00010232220609579099, - "loss": 4.4552, - "step": 958 - }, - { - "epoch": 0.81, - "learning_rate": 0.00010188679245283017, - "loss": 4.7696, - "step": 959 - }, - { - "epoch": 0.81, - "learning_rate": 0.00010145137880986937, - "loss": 4.4537, - "step": 960 - }, - { - "epoch": 0.81, - "learning_rate": 0.00010101596516690855, - "loss": 4.3338, - "step": 961 - }, - { - "epoch": 0.81, - "learning_rate": 0.00010058055152394773, - "loss": 4.4076, - "step": 962 - }, - { - "epoch": 0.81, - "learning_rate": 0.00010014513788098693, - "loss": 4.159, - "step": 963 - }, - { - "epoch": 0.81, - "learning_rate": 9.970972423802611e-05, - "loss": 4.9478, - "step": 964 - }, - { - "epoch": 0.81, - "learning_rate": 9.92743105950653e-05, - "loss": 4.5281, - "step": 965 - }, - { - "epoch": 0.81, - "learning_rate": 9.883889695210449e-05, - "loss": 4.223, - "step": 966 - }, - { - "epoch": 0.81, - "learning_rate": 9.840348330914367e-05, - "loss": 4.5062, - "step": 967 - }, - { - "epoch": 0.81, - "learning_rate": 9.796806966618286e-05, - "loss": 4.4958, - "step": 968 - }, - { - "epoch": 0.81, - "learning_rate": 9.753265602322205e-05, - "loss": 4.06, - "step": 969 - }, - { - "epoch": 0.82, - "learning_rate": 9.709724238026124e-05, - "loss": 4.3861, - "step": 970 - }, - { - "epoch": 0.82, - "learning_rate": 9.666182873730043e-05, - "loss": 4.263, - "step": 971 - }, - { - "epoch": 0.82, - "learning_rate": 9.622641509433961e-05, - "loss": 4.2316, - "step": 972 - }, - { - "epoch": 0.82, - "learning_rate": 9.57910014513788e-05, - "loss": 4.6462, - "step": 973 - }, - { - "epoch": 0.82, - "learning_rate": 9.535558780841799e-05, - "loss": 4.3962, - "step": 974 - }, - { - "epoch": 0.82, - "learning_rate": 9.492017416545717e-05, - "loss": 4.3858, - "step": 975 - }, - { - "epoch": 0.82, - "learning_rate": 9.448476052249636e-05, - "loss": 4.4299, - "step": 976 - }, - { - "epoch": 0.82, - "learning_rate": 9.404934687953555e-05, - "loss": 4.3167, - "step": 977 - }, - { - "epoch": 0.82, - "learning_rate": 9.361393323657474e-05, - "loss": 4.3328, - "step": 978 - }, - { - "epoch": 0.82, - "learning_rate": 9.317851959361392e-05, - "loss": 4.2086, - "step": 979 - }, - { - "epoch": 0.82, - "learning_rate": 9.274310595065311e-05, - "loss": 4.6029, - "step": 980 - }, - { - "epoch": 0.82, - "learning_rate": 9.23076923076923e-05, - "loss": 4.8252, - "step": 981 - }, - { - "epoch": 0.83, - "learning_rate": 9.187227866473149e-05, - "loss": 4.7713, - "step": 982 - }, - { - "epoch": 0.83, - "learning_rate": 9.143686502177067e-05, - "loss": 4.2025, - "step": 983 - }, - { - "epoch": 0.83, - "learning_rate": 9.100145137880986e-05, - "loss": 4.5659, - "step": 984 - }, - { - "epoch": 0.83, - "learning_rate": 9.056603773584905e-05, - "loss": 4.2216, - "step": 985 - }, - { - "epoch": 0.83, - "learning_rate": 9.013062409288824e-05, - "loss": 4.3045, - "step": 986 - }, { "epoch": 0.83, - "learning_rate": 8.969521044992742e-05, - "loss": 4.4816, - "step": 987 - }, - { - "epoch": 0.83, - "learning_rate": 8.92597968069666e-05, - "loss": 4.3277, - "step": 988 - }, - { - "epoch": 0.83, - "learning_rate": 8.88243831640058e-05, - "loss": 4.37, - "step": 989 + "learning_rate": 0.000294, + "loss": 4.5744, + "step": 492 }, { "epoch": 0.83, - "learning_rate": 8.838896952104498e-05, - "loss": 4.2816, - "step": 990 + "learning_rate": 0.00029459999999999995, + "loss": 4.2873, + "step": 493 }, { "epoch": 0.83, - "learning_rate": 8.795355587808417e-05, - "loss": 4.5968, - "step": 991 + "learning_rate": 0.00029519999999999997, + "loss": 4.6052, + "step": 494 }, { "epoch": 0.83, - "learning_rate": 8.751814223512336e-05, - "loss": 4.5067, - "step": 992 + "learning_rate": 0.0002958, + "loss": 4.7127, + "step": 495 }, { "epoch": 0.83, - "learning_rate": 8.708272859216254e-05, - "loss": 4.1224, - "step": 993 - }, - { - "epoch": 0.84, - "learning_rate": 8.664731494920174e-05, - "loss": 4.3888, - "step": 994 - }, - { - "epoch": 0.84, - "learning_rate": 8.621190130624092e-05, - "loss": 4.4085, - "step": 995 - }, - { - "epoch": 0.84, - "learning_rate": 8.57764876632801e-05, - "loss": 4.9815, - "step": 996 - }, - { - "epoch": 0.84, - "learning_rate": 8.53410740203193e-05, - "loss": 4.5786, - "step": 997 - }, - { - "epoch": 0.84, - "learning_rate": 8.490566037735848e-05, - "loss": 4.4196, - "step": 998 - }, - { - "epoch": 0.84, - "learning_rate": 8.447024673439768e-05, - "loss": 4.5136, - "step": 999 + "learning_rate": 0.0002964, + "loss": 4.8242, + "step": 496 }, { "epoch": 0.84, - "learning_rate": 8.403483309143686e-05, - "loss": 4.5883, - "step": 1000 + "learning_rate": 0.00029699999999999996, + "loss": 4.4981, + "step": 497 }, { "epoch": 0.84, - "eval_loss": 4.580376625061035, - "eval_runtime": 767.4671, - "eval_samples_per_second": 3.442, - "eval_steps_per_second": 0.288, - "eval_wer": 1.9531535105117017, - "step": 1000 + "learning_rate": 0.00029759999999999997, + "loss": 4.5186, + "step": 498 }, { "epoch": 0.84, - "learning_rate": 8.359941944847604e-05, - "loss": 4.4247, - "step": 1001 + "learning_rate": 0.0002982, + "loss": 5.0847, + "step": 499 }, { "epoch": 0.84, - "learning_rate": 8.316400580551524e-05, - "loss": 4.3317, - "step": 1002 + "learning_rate": 0.0002988, + "loss": 4.7686, + "step": 500 }, { "epoch": 0.84, - "learning_rate": 8.272859216255442e-05, - "loss": 4.3181, - "step": 1003 + "eval_loss": 4.900937080383301, + "eval_runtime": 769.6154, + "eval_samples_per_second": 3.433, + "eval_steps_per_second": 0.287, + "eval_wer": 1.9500793335977786, + "step": 500 }, { "epoch": 0.84, - "learning_rate": 8.22931785195936e-05, - "loss": 4.6889, - "step": 1004 + "learning_rate": 0.00029939999999999996, + "loss": 4.7283, + "step": 501 }, { "epoch": 0.84, - "learning_rate": 8.18577648766328e-05, - "loss": 4.4723, - "step": 1005 - }, - { - "epoch": 0.85, - "learning_rate": 8.142235123367198e-05, - "loss": 4.2809, - "step": 1006 - }, - { - "epoch": 0.85, - "learning_rate": 8.098693759071118e-05, - "loss": 4.3144, - "step": 1007 - }, - { - "epoch": 0.85, - "learning_rate": 8.055152394775036e-05, - "loss": 4.2816, - "step": 1008 - }, - { - "epoch": 0.85, - "learning_rate": 8.011611030478954e-05, - "loss": 4.317, - "step": 1009 - }, - { - "epoch": 0.85, - "learning_rate": 7.968069666182874e-05, - "loss": 4.316, - "step": 1010 - }, - { - "epoch": 0.85, - "learning_rate": 7.924528301886792e-05, - "loss": 4.2896, - "step": 1011 + "learning_rate": 0.0003, + "loss": 4.4173, + "step": 502 }, { "epoch": 0.85, - "learning_rate": 7.88098693759071e-05, - "loss": 4.3558, - "step": 1012 + "learning_rate": 0.00029680851063829784, + "loss": 4.5777, + "step": 503 }, { "epoch": 0.85, - "learning_rate": 7.83744557329463e-05, - "loss": 4.2193, - "step": 1013 + "learning_rate": 0.0002936170212765957, + "loss": 4.8135, + "step": 504 }, { "epoch": 0.85, - "learning_rate": 7.793904208998548e-05, - "loss": 4.2468, - "step": 1014 + "learning_rate": 0.0002904255319148936, + "loss": 4.7635, + "step": 505 }, { "epoch": 0.85, - "learning_rate": 7.750362844702468e-05, - "loss": 4.342, - "step": 1015 + "learning_rate": 0.0002872340425531915, + "loss": 4.7398, + "step": 506 }, { "epoch": 0.85, - "learning_rate": 7.706821480406386e-05, - "loss": 4.5863, - "step": 1016 + "learning_rate": 0.00028404255319148934, + "loss": 4.7005, + "step": 507 }, { "epoch": 0.85, - "learning_rate": 7.663280116110303e-05, - "loss": 4.2557, - "step": 1017 - }, - { - "epoch": 0.86, - "learning_rate": 7.619738751814224e-05, - "loss": 4.4134, - "step": 1018 - }, - { - "epoch": 0.86, - "learning_rate": 7.576197387518142e-05, - "loss": 4.5682, - "step": 1019 - }, - { - "epoch": 0.86, - "learning_rate": 7.532656023222059e-05, - "loss": 4.3113, - "step": 1020 - }, - { - "epoch": 0.86, - "learning_rate": 7.48911465892598e-05, - "loss": 4.5479, - "step": 1021 - }, - { - "epoch": 0.86, - "learning_rate": 7.445573294629897e-05, - "loss": 4.4953, - "step": 1022 + "learning_rate": 0.0002808510638297872, + "loss": 4.7717, + "step": 508 }, { "epoch": 0.86, - "learning_rate": 7.402031930333816e-05, - "loss": 4.8161, - "step": 1023 + "learning_rate": 0.00027765957446808506, + "loss": 4.5064, + "step": 509 }, { "epoch": 0.86, - "learning_rate": 7.358490566037735e-05, - "loss": 4.127, - "step": 1024 + "learning_rate": 0.000274468085106383, + "loss": 4.806, + "step": 510 }, { "epoch": 0.86, - "learning_rate": 7.314949201741654e-05, - "loss": 4.2629, - "step": 1025 + "learning_rate": 0.00027127659574468084, + "loss": 4.6994, + "step": 511 }, { "epoch": 0.86, - "learning_rate": 7.271407837445572e-05, - "loss": 4.204, - "step": 1026 + "learning_rate": 0.0002680851063829787, + "loss": 4.5931, + "step": 512 }, { "epoch": 0.86, - "learning_rate": 7.22786647314949e-05, - "loss": 4.3665, - "step": 1027 + "learning_rate": 0.00026489361702127656, + "loss": 4.4888, + "step": 513 }, { "epoch": 0.86, - "learning_rate": 7.18432510885341e-05, - "loss": 4.2541, - "step": 1028 - }, - { - "epoch": 0.87, - "learning_rate": 7.140783744557328e-05, - "loss": 4.4344, - "step": 1029 - }, - { - "epoch": 0.87, - "learning_rate": 7.097242380261248e-05, - "loss": 4.192, - "step": 1030 - }, - { - "epoch": 0.87, - "learning_rate": 7.053701015965166e-05, - "loss": 4.418, - "step": 1031 - }, - { - "epoch": 0.87, - "learning_rate": 7.010159651669085e-05, - "loss": 4.4248, - "step": 1032 - }, - { - "epoch": 0.87, - "learning_rate": 6.966618287373004e-05, - "loss": 4.8127, - "step": 1033 - }, - { - "epoch": 0.87, - "learning_rate": 6.923076923076922e-05, - "loss": 4.0747, - "step": 1034 + "learning_rate": 0.0002617021276595745, + "loss": 4.4274, + "step": 514 }, { "epoch": 0.87, - "learning_rate": 6.87953555878084e-05, - "loss": 4.1781, - "step": 1035 + "learning_rate": 0.0002585106382978723, + "loss": 4.5197, + "step": 515 }, { "epoch": 0.87, - "learning_rate": 6.83599419448476e-05, - "loss": 4.1634, - "step": 1036 + "learning_rate": 0.0002553191489361702, + "loss": 4.4446, + "step": 516 }, { "epoch": 0.87, - "learning_rate": 6.792452830188678e-05, - "loss": 4.5865, - "step": 1037 + "learning_rate": 0.00025212765957446806, + "loss": 4.7041, + "step": 517 }, { "epoch": 0.87, - "learning_rate": 6.748911465892598e-05, - "loss": 4.5169, - "step": 1038 + "learning_rate": 0.0002489361702127659, + "loss": 4.5883, + "step": 518 }, { "epoch": 0.87, - "learning_rate": 6.705370101596516e-05, - "loss": 4.3504, - "step": 1039 + "learning_rate": 0.00024574468085106384, + "loss": 4.5084, + "step": 519 }, { "epoch": 0.87, - "learning_rate": 6.661828737300434e-05, - "loss": 4.2245, - "step": 1040 - }, - { - "epoch": 0.88, - "learning_rate": 6.618287373004354e-05, - "loss": 4.4095, - "step": 1041 - }, - { - "epoch": 0.88, - "learning_rate": 6.574746008708272e-05, - "loss": 4.3194, - "step": 1042 - }, - { - "epoch": 0.88, - "learning_rate": 6.53120464441219e-05, - "loss": 4.3948, - "step": 1043 - }, - { - "epoch": 0.88, - "learning_rate": 6.48766328011611e-05, - "loss": 4.3705, - "step": 1044 - }, - { - "epoch": 0.88, - "learning_rate": 6.444121915820028e-05, - "loss": 4.7656, - "step": 1045 - }, - { - "epoch": 0.88, - "learning_rate": 6.400580551523948e-05, - "loss": 4.2077, - "step": 1046 + "learning_rate": 0.00024255319148936167, + "loss": 4.5682, + "step": 520 }, { "epoch": 0.88, - "learning_rate": 6.357039187227866e-05, - "loss": 4.268, - "step": 1047 + "learning_rate": 0.00023936170212765956, + "loss": 4.7571, + "step": 521 }, { "epoch": 0.88, - "learning_rate": 6.313497822931784e-05, - "loss": 4.4685, - "step": 1048 + "learning_rate": 0.00023617021276595742, + "loss": 4.9358, + "step": 522 }, { "epoch": 0.88, - "learning_rate": 6.269956458635704e-05, - "loss": 4.9569, - "step": 1049 + "learning_rate": 0.00023297872340425529, + "loss": 4.4314, + "step": 523 }, { "epoch": 0.88, - "learning_rate": 6.226415094339622e-05, - "loss": 4.5919, - "step": 1050 + "learning_rate": 0.00022978723404255317, + "loss": 4.1877, + "step": 524 }, { "epoch": 0.88, - "learning_rate": 6.18287373004354e-05, - "loss": 4.4872, - "step": 1051 + "learning_rate": 0.00022659574468085106, + "loss": 4.7371, + "step": 525 }, { "epoch": 0.88, - "learning_rate": 6.13933236574746e-05, - "loss": 4.219, - "step": 1052 - }, - { - "epoch": 0.89, - "learning_rate": 6.095791001451378e-05, - "loss": 4.2008, - "step": 1053 - }, - { - "epoch": 0.89, - "learning_rate": 6.052249637155297e-05, - "loss": 4.6235, - "step": 1054 - }, - { - "epoch": 0.89, - "learning_rate": 6.0087082728592154e-05, - "loss": 4.5256, - "step": 1055 - }, - { - "epoch": 0.89, - "learning_rate": 5.965166908563134e-05, - "loss": 4.4614, - "step": 1056 - }, - { - "epoch": 0.89, - "learning_rate": 5.921625544267053e-05, - "loss": 4.4202, - "step": 1057 - }, - { - "epoch": 0.89, - "learning_rate": 5.878084179970972e-05, - "loss": 4.1262, - "step": 1058 - }, - { - "epoch": 0.89, - "learning_rate": 5.8345428156748904e-05, - "loss": 4.4001, - "step": 1059 + "learning_rate": 0.0002234042553191489, + "loss": 4.6611, + "step": 526 }, { "epoch": 0.89, - "learning_rate": 5.791001451378809e-05, - "loss": 4.4339, - "step": 1060 + "learning_rate": 0.00022021276595744679, + "loss": 4.6147, + "step": 527 }, { "epoch": 0.89, - "learning_rate": 5.747460087082728e-05, - "loss": 4.4371, - "step": 1061 + "learning_rate": 0.00021702127659574468, + "loss": 4.4807, + "step": 528 }, { "epoch": 0.89, - "learning_rate": 5.703918722786647e-05, - "loss": 4.2648, - "step": 1062 + "learning_rate": 0.00021382978723404254, + "loss": 4.8237, + "step": 529 }, { "epoch": 0.89, - "learning_rate": 5.660377358490566e-05, - "loss": 4.6402, - "step": 1063 + "learning_rate": 0.0002106382978723404, + "loss": 4.3209, + "step": 530 }, { "epoch": 0.89, - "learning_rate": 5.616835994194484e-05, - "loss": 4.252, - "step": 1064 - }, - { - "epoch": 0.9, - "learning_rate": 5.573294629898403e-05, - "loss": 4.1555, - "step": 1065 - }, - { - "epoch": 0.9, - "learning_rate": 5.529753265602322e-05, - "loss": 4.2044, - "step": 1066 - }, - { - "epoch": 0.9, - "learning_rate": 5.486211901306241e-05, - "loss": 4.278, - "step": 1067 - }, - { - "epoch": 0.9, - "learning_rate": 5.442670537010159e-05, - "loss": 4.2826, - "step": 1068 - }, - { - "epoch": 0.9, - "learning_rate": 5.399129172714078e-05, - "loss": 4.2409, - "step": 1069 + "learning_rate": 0.0002074468085106383, + "loss": 4.4169, + "step": 531 }, { - "epoch": 0.9, - "learning_rate": 5.355587808417997e-05, - "loss": 4.4354, - "step": 1070 + "epoch": 0.89, + "learning_rate": 0.00020425531914893615, + "loss": 4.6349, + "step": 532 }, { "epoch": 0.9, - "learning_rate": 5.312046444121916e-05, - "loss": 4.2282, - "step": 1071 + "learning_rate": 0.00020106382978723404, + "loss": 4.6437, + "step": 533 }, { "epoch": 0.9, - "learning_rate": 5.2685050798258335e-05, - "loss": 4.5258, - "step": 1072 + "learning_rate": 0.00019787234042553187, + "loss": 4.4318, + "step": 534 }, { "epoch": 0.9, - "learning_rate": 5.2249637155297524e-05, - "loss": 3.9681, - "step": 1073 + "learning_rate": 0.00019468085106382976, + "loss": 4.3929, + "step": 535 }, { "epoch": 0.9, - "learning_rate": 5.181422351233671e-05, - "loss": 4.4695, - "step": 1074 + "learning_rate": 0.00019148936170212765, + "loss": 4.5963, + "step": 536 }, { "epoch": 0.9, - "learning_rate": 5.137880986937591e-05, - "loss": 4.4431, - "step": 1075 + "learning_rate": 0.0001882978723404255, + "loss": 4.474, + "step": 537 }, { "epoch": 0.9, - "learning_rate": 5.0943396226415085e-05, - "loss": 4.6197, - "step": 1076 - }, - { - "epoch": 0.91, - "learning_rate": 5.0507982583454274e-05, - "loss": 4.5793, - "step": 1077 - }, - { - "epoch": 0.91, - "learning_rate": 5.007256894049346e-05, - "loss": 4.4406, - "step": 1078 - }, - { - "epoch": 0.91, - "learning_rate": 4.963715529753265e-05, - "loss": 4.2787, - "step": 1079 - }, - { - "epoch": 0.91, - "learning_rate": 4.9201741654571834e-05, - "loss": 4.106, - "step": 1080 - }, - { - "epoch": 0.91, - "learning_rate": 4.8766328011611024e-05, - "loss": 4.434, - "step": 1081 - }, - { - "epoch": 0.91, - "learning_rate": 4.833091436865021e-05, - "loss": 4.4248, - "step": 1082 + "learning_rate": 0.0001851063829787234, + "loss": 4.396, + "step": 538 }, { "epoch": 0.91, - "learning_rate": 4.78955007256894e-05, - "loss": 4.7218, - "step": 1083 + "learning_rate": 0.00018191489361702126, + "loss": 4.7928, + "step": 539 }, { "epoch": 0.91, - "learning_rate": 4.7460087082728584e-05, - "loss": 4.2686, - "step": 1084 + "learning_rate": 0.00017872340425531912, + "loss": 4.569, + "step": 540 }, { "epoch": 0.91, - "learning_rate": 4.702467343976777e-05, - "loss": 4.2947, - "step": 1085 + "learning_rate": 0.000175531914893617, + "loss": 4.4839, + "step": 541 }, { "epoch": 0.91, - "learning_rate": 4.658925979680696e-05, - "loss": 4.2582, - "step": 1086 + "learning_rate": 0.0001723404255319149, + "loss": 4.6734, + "step": 542 }, { "epoch": 0.91, - "learning_rate": 4.615384615384615e-05, - "loss": 4.2652, - "step": 1087 + "learning_rate": 0.00016914893617021274, + "loss": 4.405, + "step": 543 }, { "epoch": 0.91, - "learning_rate": 4.5718432510885334e-05, - "loss": 4.8073, - "step": 1088 - }, - { - "epoch": 0.92, - "learning_rate": 4.528301886792452e-05, - "loss": 4.2283, - "step": 1089 - }, - { - "epoch": 0.92, - "learning_rate": 4.484760522496371e-05, - "loss": 4.3468, - "step": 1090 - }, - { - "epoch": 0.92, - "learning_rate": 4.44121915820029e-05, - "loss": 4.2619, - "step": 1091 - }, - { - "epoch": 0.92, - "learning_rate": 4.397677793904208e-05, - "loss": 4.1893, - "step": 1092 - }, - { - "epoch": 0.92, - "learning_rate": 4.354136429608127e-05, - "loss": 4.2702, - "step": 1093 - }, - { - "epoch": 0.92, - "learning_rate": 4.310595065312046e-05, - "loss": 4.3322, - "step": 1094 + "learning_rate": 0.00016595744680851062, + "loss": 4.5196, + "step": 544 }, { "epoch": 0.92, - "learning_rate": 4.267053701015965e-05, - "loss": 4.3116, - "step": 1095 + "learning_rate": 0.00016276595744680849, + "loss": 4.4101, + "step": 545 }, { "epoch": 0.92, - "learning_rate": 4.223512336719884e-05, - "loss": 4.6978, - "step": 1096 + "learning_rate": 0.00015957446808510637, + "loss": 4.9518, + "step": 546 }, { "epoch": 0.92, - "learning_rate": 4.179970972423802e-05, - "loss": 4.4064, - "step": 1097 + "learning_rate": 0.00015638297872340426, + "loss": 4.4228, + "step": 547 }, { "epoch": 0.92, - "learning_rate": 4.136429608127721e-05, - "loss": 5.3322, - "step": 1098 + "learning_rate": 0.0001531914893617021, + "loss": 5.044, + "step": 548 }, { "epoch": 0.92, - "learning_rate": 4.09288824383164e-05, - "loss": 4.75, - "step": 1099 + "learning_rate": 0.00015, + "loss": 5.1572, + "step": 549 }, { "epoch": 0.92, - "learning_rate": 4.049346879535559e-05, - "loss": 4.291, - "step": 1100 - }, - { - "epoch": 0.93, - "learning_rate": 4.005805515239477e-05, - "loss": 4.9697, - "step": 1101 - }, - { - "epoch": 0.93, - "learning_rate": 3.962264150943396e-05, - "loss": 4.2999, - "step": 1102 - }, - { - "epoch": 0.93, - "learning_rate": 3.918722786647315e-05, - "loss": 4.236, - "step": 1103 - }, - { - "epoch": 0.93, - "learning_rate": 3.875181422351234e-05, - "loss": 4.1279, - "step": 1104 - }, - { - "epoch": 0.93, - "learning_rate": 3.8316400580551515e-05, - "loss": 4.295, - "step": 1105 - }, - { - "epoch": 0.93, - "learning_rate": 3.788098693759071e-05, - "loss": 4.0693, - "step": 1106 + "learning_rate": 0.00014680851063829785, + "loss": 4.4465, + "step": 550 }, { "epoch": 0.93, - "learning_rate": 3.74455732946299e-05, - "loss": 4.7203, - "step": 1107 + "learning_rate": 0.00014361702127659574, + "loss": 4.5287, + "step": 551 }, { "epoch": 0.93, - "learning_rate": 3.701015965166908e-05, - "loss": 4.3843, - "step": 1108 + "learning_rate": 0.0001404255319148936, + "loss": 4.616, + "step": 552 }, { "epoch": 0.93, - "learning_rate": 3.657474600870827e-05, - "loss": 4.3651, - "step": 1109 + "learning_rate": 0.0001372340425531915, + "loss": 4.5176, + "step": 553 }, { "epoch": 0.93, - "learning_rate": 3.613933236574745e-05, - "loss": 4.1765, - "step": 1110 + "learning_rate": 0.00013404255319148935, + "loss": 4.3372, + "step": 554 }, { "epoch": 0.93, - "learning_rate": 3.570391872278664e-05, - "loss": 4.3239, - "step": 1111 + "learning_rate": 0.00013085106382978724, + "loss": 4.4698, + "step": 555 }, { "epoch": 0.93, - "learning_rate": 3.526850507982583e-05, - "loss": 4.728, - "step": 1112 - }, - { - "epoch": 0.94, - "learning_rate": 3.483309143686502e-05, - "loss": 4.3438, - "step": 1113 - }, - { - "epoch": 0.94, - "learning_rate": 3.43976777939042e-05, - "loss": 3.9835, - "step": 1114 - }, - { - "epoch": 0.94, - "learning_rate": 3.396226415094339e-05, - "loss": 4.4246, - "step": 1115 - }, - { - "epoch": 0.94, - "learning_rate": 3.352685050798258e-05, - "loss": 4.4157, - "step": 1116 - }, - { - "epoch": 0.94, - "learning_rate": 3.309143686502177e-05, - "loss": 4.4774, - "step": 1117 - }, - { - "epoch": 0.94, - "learning_rate": 3.265602322206095e-05, - "loss": 4.1395, - "step": 1118 + "learning_rate": 0.0001276595744680851, + "loss": 4.4738, + "step": 556 }, { "epoch": 0.94, - "learning_rate": 3.222060957910014e-05, - "loss": 3.9259, - "step": 1119 + "learning_rate": 0.00012446808510638296, + "loss": 4.6455, + "step": 557 }, { "epoch": 0.94, - "learning_rate": 3.178519593613933e-05, - "loss": 4.4353, - "step": 1120 + "learning_rate": 0.00012127659574468084, + "loss": 4.5513, + "step": 558 }, { "epoch": 0.94, - "learning_rate": 3.134978229317852e-05, - "loss": 4.4326, - "step": 1121 + "learning_rate": 0.00011808510638297871, + "loss": 4.3848, + "step": 559 }, { "epoch": 0.94, - "learning_rate": 3.09143686502177e-05, - "loss": 4.5233, - "step": 1122 + "learning_rate": 0.00011489361702127659, + "loss": 4.3127, + "step": 560 }, { "epoch": 0.94, - "learning_rate": 3.047895500725689e-05, - "loss": 4.1491, - "step": 1123 + "learning_rate": 0.00011170212765957445, + "loss": 4.7444, + "step": 561 }, { "epoch": 0.94, - "learning_rate": 3.0043541364296077e-05, - "loss": 4.3489, - "step": 1124 - }, - { - "epoch": 0.95, - "learning_rate": 2.9608127721335266e-05, - "loss": 4.705, - "step": 1125 - }, - { - "epoch": 0.95, - "learning_rate": 2.9172714078374452e-05, - "loss": 4.2625, - "step": 1126 - }, - { - "epoch": 0.95, - "learning_rate": 2.873730043541364e-05, - "loss": 4.344, - "step": 1127 - }, - { - "epoch": 0.95, - "learning_rate": 2.830188679245283e-05, - "loss": 4.274, - "step": 1128 - }, - { - "epoch": 0.95, - "learning_rate": 2.7866473149492016e-05, - "loss": 4.3003, - "step": 1129 - }, - { - "epoch": 0.95, - "learning_rate": 2.7431059506531205e-05, - "loss": 4.5258, - "step": 1130 + "learning_rate": 0.00010851063829787234, + "loss": 4.4381, + "step": 562 }, { "epoch": 0.95, - "learning_rate": 2.699564586357039e-05, - "loss": 4.0453, - "step": 1131 + "learning_rate": 0.0001053191489361702, + "loss": 4.4416, + "step": 563 }, { "epoch": 0.95, - "learning_rate": 2.656023222060958e-05, - "loss": 4.1659, - "step": 1132 + "learning_rate": 0.00010212765957446807, + "loss": 4.8204, + "step": 564 }, { "epoch": 0.95, - "learning_rate": 2.6124818577648762e-05, - "loss": 4.3705, - "step": 1133 + "learning_rate": 9.893617021276594e-05, + "loss": 4.5756, + "step": 565 }, { "epoch": 0.95, - "learning_rate": 2.5689404934687955e-05, - "loss": 4.1862, - "step": 1134 + "learning_rate": 9.574468085106382e-05, + "loss": 4.5617, + "step": 566 }, { "epoch": 0.95, - "learning_rate": 2.5253991291727137e-05, - "loss": 4.496, - "step": 1135 - }, - { - "epoch": 0.96, - "learning_rate": 2.4818577648766326e-05, - "loss": 4.1733, - "step": 1136 - }, - { - "epoch": 0.96, - "learning_rate": 2.4383164005805512e-05, - "loss": 4.204, - "step": 1137 - }, - { - "epoch": 0.96, - "learning_rate": 2.39477503628447e-05, - "loss": 4.274, - "step": 1138 - }, - { - "epoch": 0.96, - "learning_rate": 2.3512336719883887e-05, - "loss": 4.1716, - "step": 1139 - }, - { - "epoch": 0.96, - "learning_rate": 2.3076923076923076e-05, - "loss": 4.2, - "step": 1140 - }, - { - "epoch": 0.96, - "learning_rate": 2.264150943396226e-05, - "loss": 4.4644, - "step": 1141 + "learning_rate": 9.25531914893617e-05, + "loss": 4.4047, + "step": 567 }, { "epoch": 0.96, - "learning_rate": 2.220609579100145e-05, - "loss": 4.2474, - "step": 1142 + "learning_rate": 8.936170212765956e-05, + "loss": 4.4505, + "step": 568 }, { "epoch": 0.96, - "learning_rate": 2.1770682148040636e-05, - "loss": 4.4789, - "step": 1143 + "learning_rate": 8.617021276595745e-05, + "loss": 4.356, + "step": 569 }, { "epoch": 0.96, - "learning_rate": 2.1335268505079825e-05, - "loss": 4.5033, - "step": 1144 + "learning_rate": 8.297872340425531e-05, + "loss": 4.4439, + "step": 570 }, { "epoch": 0.96, - "learning_rate": 2.089985486211901e-05, - "loss": 4.1848, - "step": 1145 + "learning_rate": 7.978723404255319e-05, + "loss": 4.4821, + "step": 571 }, { "epoch": 0.96, - "learning_rate": 2.04644412191582e-05, - "loss": 4.5938, - "step": 1146 + "learning_rate": 7.659574468085105e-05, + "loss": 4.3846, + "step": 572 }, { "epoch": 0.96, - "learning_rate": 2.0029027576197386e-05, - "loss": 5.8271, - "step": 1147 - }, - { - "epoch": 0.97, - "learning_rate": 1.9593613933236575e-05, - "loss": 4.3477, - "step": 1148 - }, - { - "epoch": 0.97, - "learning_rate": 1.9158200290275757e-05, - "loss": 4.6769, - "step": 1149 - }, - { - "epoch": 0.97, - "learning_rate": 1.872278664731495e-05, - "loss": 4.2925, - "step": 1150 - }, - { - "epoch": 0.97, - "learning_rate": 1.8287373004354136e-05, - "loss": 4.2285, - "step": 1151 - }, - { - "epoch": 0.97, - "learning_rate": 1.785195936139332e-05, - "loss": 4.0426, - "step": 1152 - }, - { - "epoch": 0.97, - "learning_rate": 1.741654571843251e-05, - "loss": 4.1833, - "step": 1153 + "learning_rate": 7.340425531914892e-05, + "loss": 4.4297, + "step": 573 }, { "epoch": 0.97, - "learning_rate": 1.6981132075471696e-05, - "loss": 4.0901, - "step": 1154 + "learning_rate": 7.02127659574468e-05, + "loss": 4.3423, + "step": 574 }, { "epoch": 0.97, - "learning_rate": 1.6545718432510885e-05, - "loss": 4.4645, - "step": 1155 + "learning_rate": 6.702127659574467e-05, + "loss": 4.3857, + "step": 575 }, { "epoch": 0.97, - "learning_rate": 1.611030478955007e-05, - "loss": 4.3001, - "step": 1156 + "learning_rate": 6.382978723404255e-05, + "loss": 4.3895, + "step": 576 }, { "epoch": 0.97, - "learning_rate": 1.567489114658926e-05, - "loss": 4.4954, - "step": 1157 + "learning_rate": 6.063829787234042e-05, + "loss": 4.5989, + "step": 577 }, { "epoch": 0.97, - "learning_rate": 1.5239477503628446e-05, - "loss": 4.2257, - "step": 1158 + "learning_rate": 5.7446808510638294e-05, + "loss": 4.2233, + "step": 578 }, { "epoch": 0.97, - "learning_rate": 1.4804063860667633e-05, - "loss": 4.0976, - "step": 1159 - }, - { - "epoch": 0.98, - "learning_rate": 1.436865021770682e-05, - "loss": 4.3462, - "step": 1160 - }, - { - "epoch": 0.98, - "learning_rate": 1.3933236574746008e-05, - "loss": 4.327, - "step": 1161 - }, - { - "epoch": 0.98, - "learning_rate": 1.3497822931785195e-05, - "loss": 4.5508, - "step": 1162 - }, - { - "epoch": 0.98, - "learning_rate": 1.3062409288824381e-05, - "loss": 4.4825, - "step": 1163 - }, - { - "epoch": 0.98, - "learning_rate": 1.2626995645863568e-05, - "loss": 4.1438, - "step": 1164 - }, - { - "epoch": 0.98, - "learning_rate": 1.2191582002902756e-05, - "loss": 4.3982, - "step": 1165 + "learning_rate": 5.425531914893617e-05, + "loss": 4.3258, + "step": 579 }, { "epoch": 0.98, - "learning_rate": 1.1756168359941943e-05, - "loss": 4.4645, - "step": 1166 + "learning_rate": 5.106382978723404e-05, + "loss": 4.4218, + "step": 580 }, { "epoch": 0.98, - "learning_rate": 1.132075471698113e-05, - "loss": 4.077, - "step": 1167 + "learning_rate": 4.787234042553191e-05, + "loss": 4.5631, + "step": 581 }, { "epoch": 0.98, - "learning_rate": 1.0885341074020318e-05, - "loss": 4.1337, - "step": 1168 + "learning_rate": 4.468085106382978e-05, + "loss": 4.3969, + "step": 582 }, { "epoch": 0.98, - "learning_rate": 1.0449927431059506e-05, - "loss": 4.2341, - "step": 1169 + "learning_rate": 4.1489361702127656e-05, + "loss": 4.203, + "step": 583 }, { "epoch": 0.98, - "learning_rate": 1.0014513788098693e-05, - "loss": 4.3717, - "step": 1170 + "learning_rate": 3.8297872340425525e-05, + "loss": 4.3834, + "step": 584 }, { "epoch": 0.98, - "learning_rate": 9.579100145137879e-06, - "loss": 4.2809, - "step": 1171 - }, - { - "epoch": 0.99, - "learning_rate": 9.143686502177068e-06, - "loss": 4.0888, - "step": 1172 - }, - { - "epoch": 0.99, - "learning_rate": 8.708272859216255e-06, - "loss": 4.2783, - "step": 1173 - }, - { - "epoch": 0.99, - "learning_rate": 8.272859216255443e-06, - "loss": 4.2921, - "step": 1174 - }, - { - "epoch": 0.99, - "learning_rate": 7.83744557329463e-06, - "loss": 4.0271, - "step": 1175 - }, - { - "epoch": 0.99, - "learning_rate": 7.4020319303338166e-06, - "loss": 4.2022, - "step": 1176 - }, - { - "epoch": 0.99, - "learning_rate": 6.966618287373004e-06, - "loss": 4.2548, - "step": 1177 + "learning_rate": 3.51063829787234e-05, + "loss": 4.4053, + "step": 585 }, { "epoch": 0.99, - "learning_rate": 6.5312046444121905e-06, - "loss": 4.6275, - "step": 1178 + "learning_rate": 3.1914893617021275e-05, + "loss": 4.3702, + "step": 586 }, { "epoch": 0.99, - "learning_rate": 6.095791001451378e-06, - "loss": 4.0099, - "step": 1179 + "learning_rate": 2.8723404255319147e-05, + "loss": 4.4538, + "step": 587 }, { "epoch": 0.99, - "learning_rate": 5.660377358490565e-06, - "loss": 4.3623, - "step": 1180 + "learning_rate": 2.553191489361702e-05, + "loss": 4.617, + "step": 588 }, { "epoch": 0.99, - "learning_rate": 5.224963715529753e-06, - "loss": 4.3703, - "step": 1181 + "learning_rate": 2.234042553191489e-05, + "loss": 4.4198, + "step": 589 }, { "epoch": 0.99, - "learning_rate": 4.789550072568939e-06, - "loss": 4.3294, - "step": 1182 + "learning_rate": 1.9148936170212762e-05, + "loss": 4.2778, + "step": 590 }, { "epoch": 0.99, - "learning_rate": 4.354136429608128e-06, - "loss": 4.2065, - "step": 1183 - }, - { - "epoch": 1.0, - "learning_rate": 3.918722786647315e-06, - "loss": 4.3322, - "step": 1184 - }, - { - "epoch": 1.0, - "learning_rate": 3.483309143686502e-06, - "loss": 4.2888, - "step": 1185 - }, - { - "epoch": 1.0, - "learning_rate": 3.047895500725689e-06, - "loss": 4.372, - "step": 1186 + "learning_rate": 1.5957446808510637e-05, + "loss": 4.5074, + "step": 591 }, { "epoch": 1.0, - "learning_rate": 2.6124818577648764e-06, - "loss": 4.1643, - "step": 1187 + "learning_rate": 1.276595744680851e-05, + "loss": 4.901, + "step": 592 }, { "epoch": 1.0, - "learning_rate": 2.177068214804064e-06, - "loss": 4.3376, - "step": 1188 + "learning_rate": 9.574468085106381e-06, + "loss": 4.704, + "step": 593 }, { "epoch": 1.0, - "learning_rate": 1.741654571843251e-06, - "loss": 4.4054, - "step": 1189 + "learning_rate": 6.382978723404255e-06, + "loss": 4.5343, + "step": 594 }, { "epoch": 1.0, - "step": 1189, + "step": 594, "total_flos": 0.0, - "train_loss": 4.616779040648618, - "train_runtime": 6389.1881, - "train_samples_per_second": 4.467, - "train_steps_per_second": 0.186 + "train_loss": 4.477476079857309, + "train_runtime": 5476.9762, + "train_samples_per_second": 5.211, + "train_steps_per_second": 0.108 } ], - "max_steps": 1189, + "max_steps": 594, "num_train_epochs": 1, "total_flos": 0.0, "trial_name": null,