| { | |
| "best_metric": 2.0949487686157227, | |
| "best_model_checkpoint": "output/checkpoint-2000", | |
| "epoch": 1.38650390625, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.7847, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 2.789, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.7653, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.7217, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2e-05, | |
| "loss": 2.7964, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.4e-05, | |
| "loss": 2.7433, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 2.6567, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 2.7926, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.6e-05, | |
| "loss": 2.7431, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4e-05, | |
| "loss": 2.7118, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 2.7353, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.8e-05, | |
| "loss": 2.7365, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 2.6999, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 2.6944, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6e-05, | |
| "loss": 2.6036, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 2.6631, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 2.6435, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.2e-05, | |
| "loss": 2.6806, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 7.6e-05, | |
| "loss": 2.5884, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8e-05, | |
| "loss": 2.6082, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.4e-05, | |
| "loss": 2.5936, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 2.5439, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 2.5566, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.6e-05, | |
| "loss": 2.5111, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001, | |
| "loss": 2.5215, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00010400000000000001, | |
| "loss": 2.4623, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00010800000000000001, | |
| "loss": 2.4546, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00011200000000000001, | |
| "loss": 2.5861, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000116, | |
| "loss": 2.5458, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00012, | |
| "loss": 2.473, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000124, | |
| "loss": 2.44, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00012800000000000002, | |
| "loss": 2.4321, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000132, | |
| "loss": 2.4441, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00013600000000000003, | |
| "loss": 2.4543, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00014, | |
| "loss": 2.4442, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000144, | |
| "loss": 2.3461, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000148, | |
| "loss": 2.2818, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000152, | |
| "loss": 2.3907, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00015600000000000002, | |
| "loss": 2.3692, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00016, | |
| "loss": 2.3582, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000164, | |
| "loss": 2.3354, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000168, | |
| "loss": 2.4426, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000172, | |
| "loss": 2.2715, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00017600000000000002, | |
| "loss": 2.3155, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00018, | |
| "loss": 2.243, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00018400000000000003, | |
| "loss": 2.3097, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000188, | |
| "loss": 2.2454, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000192, | |
| "loss": 2.3081, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000196, | |
| "loss": 2.3393, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002, | |
| "loss": 2.3029, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00020400000000000003, | |
| "loss": 2.2424, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00020800000000000001, | |
| "loss": 2.2761, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00021200000000000003, | |
| "loss": 2.355, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00021600000000000002, | |
| "loss": 2.3057, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00022000000000000003, | |
| "loss": 2.3098, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00022400000000000002, | |
| "loss": 2.2867, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00022799999999999999, | |
| "loss": 2.3558, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000232, | |
| "loss": 2.3114, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000236, | |
| "loss": 2.2818, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00024, | |
| "loss": 2.3194, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000244, | |
| "loss": 2.2405, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000248, | |
| "loss": 2.31, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000252, | |
| "loss": 2.3165, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00025600000000000004, | |
| "loss": 2.3839, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00026000000000000003, | |
| "loss": 2.3172, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000264, | |
| "loss": 2.2911, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000268, | |
| "loss": 2.2946, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00027200000000000005, | |
| "loss": 2.3343, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000276, | |
| "loss": 2.3077, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00028, | |
| "loss": 2.3141, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000284, | |
| "loss": 2.3114, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000288, | |
| "loss": 2.2935, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000292, | |
| "loss": 2.375, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000296, | |
| "loss": 2.3633, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00030000000000000003, | |
| "loss": 2.2862, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000304, | |
| "loss": 2.2846, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000308, | |
| "loss": 2.2553, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00031200000000000005, | |
| "loss": 2.2767, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00031600000000000004, | |
| "loss": 2.3474, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00032, | |
| "loss": 2.3817, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000324, | |
| "loss": 2.345, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000328, | |
| "loss": 2.3573, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000332, | |
| "loss": 2.3012, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000336, | |
| "loss": 2.2964, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00034, | |
| "loss": 2.3411, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000344, | |
| "loss": 2.2456, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000348, | |
| "loss": 2.2561, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00035200000000000005, | |
| "loss": 2.2142, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00035600000000000003, | |
| "loss": 2.2198, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00036, | |
| "loss": 2.2613, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000364, | |
| "loss": 2.3107, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00036800000000000005, | |
| "loss": 2.2593, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00037200000000000004, | |
| "loss": 2.2547, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000376, | |
| "loss": 2.3041, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00038, | |
| "loss": 2.3346, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000384, | |
| "loss": 2.338, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000388, | |
| "loss": 2.2237, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000392, | |
| "loss": 2.2255, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00039600000000000003, | |
| "loss": 2.2983, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004, | |
| "loss": 2.2828, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000404, | |
| "loss": 2.3934, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00040800000000000005, | |
| "loss": 2.3819, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00041200000000000004, | |
| "loss": 2.3157, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00041600000000000003, | |
| "loss": 2.2675, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00042, | |
| "loss": 2.2912, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00042400000000000006, | |
| "loss": 2.2761, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00042800000000000005, | |
| "loss": 2.3368, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00043200000000000004, | |
| "loss": 2.2968, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000436, | |
| "loss": 2.3208, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00044000000000000007, | |
| "loss": 2.2038, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00044400000000000006, | |
| "loss": 2.3589, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00044800000000000005, | |
| "loss": 2.3036, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000452, | |
| "loss": 2.2948, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00045599999999999997, | |
| "loss": 2.1675, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00045999999999999996, | |
| "loss": 2.3433, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000464, | |
| "loss": 2.2702, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000468, | |
| "loss": 2.3075, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000472, | |
| "loss": 2.1956, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000476, | |
| "loss": 2.234, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00048, | |
| "loss": 2.2742, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000484, | |
| "loss": 2.3237, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000488, | |
| "loss": 2.3122, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000492, | |
| "loss": 2.2578, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000496, | |
| "loss": 2.3176, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005, | |
| "loss": 2.2668, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000504, | |
| "loss": 2.3111, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000508, | |
| "loss": 2.2754, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005120000000000001, | |
| "loss": 2.2973, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005160000000000001, | |
| "loss": 2.3103, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0005200000000000001, | |
| "loss": 2.2599, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000524, | |
| "loss": 2.2084, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000528, | |
| "loss": 2.3057, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000532, | |
| "loss": 2.2949, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000536, | |
| "loss": 2.3372, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0005400000000000001, | |
| "loss": 2.3106, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0005440000000000001, | |
| "loss": 2.3088, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0005480000000000001, | |
| "loss": 2.4012, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000552, | |
| "loss": 2.2033, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000556, | |
| "loss": 2.2326, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00056, | |
| "loss": 2.2218, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000564, | |
| "loss": 2.2605, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000568, | |
| "loss": 2.2576, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000572, | |
| "loss": 2.2998, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000576, | |
| "loss": 2.3869, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00058, | |
| "loss": 2.2979, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000584, | |
| "loss": 2.2765, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000588, | |
| "loss": 2.3446, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000592, | |
| "loss": 2.2205, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0005960000000000001, | |
| "loss": 2.3164, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006000000000000001, | |
| "loss": 2.2707, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000604, | |
| "loss": 2.3147, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000608, | |
| "loss": 2.2527, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000612, | |
| "loss": 2.27, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000616, | |
| "loss": 2.2256, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00062, | |
| "loss": 2.2865, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0006240000000000001, | |
| "loss": 2.2749, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0006280000000000001, | |
| "loss": 2.3338, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0006320000000000001, | |
| "loss": 2.3202, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0006360000000000001, | |
| "loss": 2.271, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00064, | |
| "loss": 2.2443, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000644, | |
| "loss": 2.2537, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000648, | |
| "loss": 2.2206, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000652, | |
| "loss": 2.2687, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000656, | |
| "loss": 2.2745, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00066, | |
| "loss": 2.283, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000664, | |
| "loss": 2.2333, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000668, | |
| "loss": 2.2706, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000672, | |
| "loss": 2.2628, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0006760000000000001, | |
| "loss": 2.3193, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00068, | |
| "loss": 2.318, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000684, | |
| "loss": 2.2414, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000688, | |
| "loss": 2.3725, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000692, | |
| "loss": 2.3124, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000696, | |
| "loss": 2.2374, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007, | |
| "loss": 2.2959, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007040000000000001, | |
| "loss": 2.2835, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007080000000000001, | |
| "loss": 2.3138, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007120000000000001, | |
| "loss": 2.3125, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007160000000000001, | |
| "loss": 2.2884, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00072, | |
| "loss": 2.3444, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000724, | |
| "loss": 2.3208, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000728, | |
| "loss": 2.2983, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000732, | |
| "loss": 2.2733, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007360000000000001, | |
| "loss": 2.3119, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007400000000000001, | |
| "loss": 2.3803, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007440000000000001, | |
| "loss": 2.2776, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007480000000000001, | |
| "loss": 2.241, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000752, | |
| "loss": 2.2678, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0007559999999999999, | |
| "loss": 2.2228, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00076, | |
| "loss": 2.3228, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000764, | |
| "loss": 2.2546, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000768, | |
| "loss": 2.3122, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000772, | |
| "loss": 2.3318, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000776, | |
| "loss": 2.369, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00078, | |
| "loss": 2.2685, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000784, | |
| "loss": 2.275, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007880000000000001, | |
| "loss": 2.2667, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007920000000000001, | |
| "loss": 2.2844, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000796, | |
| "loss": 2.3124, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0008, | |
| "loss": 2.3202, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.1868629455566406, | |
| "eval_rouge1": 11.8686, | |
| "eval_rouge2": 3.9117, | |
| "eval_rougeL": 7.4858, | |
| "eval_rougeLsum": 7.4779, | |
| "eval_runtime": 8978.087, | |
| "eval_samples_per_second": 0.056, | |
| "eval_steps_per_second": 0.056, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007995555555555556, | |
| "loss": 2.2859, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007991111111111112, | |
| "loss": 2.2128, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007986666666666667, | |
| "loss": 2.254, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007982222222222223, | |
| "loss": 2.312, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007977777777777778, | |
| "loss": 2.3304, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007973333333333334, | |
| "loss": 2.3126, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007968888888888889, | |
| "loss": 2.3429, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007964444444444445, | |
| "loss": 2.3222, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000796, | |
| "loss": 2.2903, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0007955555555555556, | |
| "loss": 2.3405, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007951111111111112, | |
| "loss": 2.3691, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007946666666666667, | |
| "loss": 2.2961, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007942222222222222, | |
| "loss": 2.3318, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007937777777777778, | |
| "loss": 2.2794, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007933333333333334, | |
| "loss": 2.2903, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000792888888888889, | |
| "loss": 2.321, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007924444444444445, | |
| "loss": 2.2768, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007920000000000001, | |
| "loss": 2.2916, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007915555555555557, | |
| "loss": 2.226, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007911111111111111, | |
| "loss": 2.3054, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007906666666666666, | |
| "loss": 2.3064, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007902222222222222, | |
| "loss": 2.28, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007897777777777778, | |
| "loss": 2.3623, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007893333333333334, | |
| "loss": 2.3449, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000788888888888889, | |
| "loss": 2.2675, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007884444444444445, | |
| "loss": 2.38, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007880000000000001, | |
| "loss": 2.3044, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007875555555555556, | |
| "loss": 2.3083, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0007871111111111111, | |
| "loss": 2.1913, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007866666666666666, | |
| "loss": 2.2425, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007862222222222222, | |
| "loss": 2.3341, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007857777777777778, | |
| "loss": 2.2575, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007853333333333334, | |
| "loss": 2.3312, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000784888888888889, | |
| "loss": 2.2838, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007844444444444445, | |
| "loss": 2.2315, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000784, | |
| "loss": 2.3253, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007835555555555556, | |
| "loss": 2.2557, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007831111111111112, | |
| "loss": 2.3264, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007826666666666668, | |
| "loss": 2.3318, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007822222222222222, | |
| "loss": 2.3574, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007817777777777778, | |
| "loss": 2.2863, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007813333333333334, | |
| "loss": 2.3765, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007808888888888889, | |
| "loss": 2.2848, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007804444444444444, | |
| "loss": 2.3497, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00078, | |
| "loss": 2.2716, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007795555555555556, | |
| "loss": 2.3409, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007791111111111112, | |
| "loss": 2.3397, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007786666666666668, | |
| "loss": 2.232, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007782222222222222, | |
| "loss": 2.2983, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0007777777777777778, | |
| "loss": 2.3148, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007773333333333333, | |
| "loss": 2.2546, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007768888888888889, | |
| "loss": 2.2874, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007764444444444444, | |
| "loss": 2.3787, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000776, | |
| "loss": 2.3286, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007755555555555556, | |
| "loss": 2.3242, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007751111111111112, | |
| "loss": 2.2548, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007746666666666668, | |
| "loss": 2.2251, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007742222222222223, | |
| "loss": 2.291, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007737777777777779, | |
| "loss": 2.2951, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007733333333333333, | |
| "loss": 2.2871, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007728888888888889, | |
| "loss": 2.29, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007724444444444444, | |
| "loss": 2.2888, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000772, | |
| "loss": 2.3121, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007715555555555556, | |
| "loss": 2.2543, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007711111111111112, | |
| "loss": 2.2383, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007706666666666668, | |
| "loss": 2.3759, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007702222222222223, | |
| "loss": 2.2838, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007697777777777778, | |
| "loss": 2.283, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0007693333333333333, | |
| "loss": 2.3027, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007688888888888889, | |
| "loss": 2.4145, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007684444444444445, | |
| "loss": 2.305, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000768, | |
| "loss": 2.335, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007675555555555556, | |
| "loss": 2.3401, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007671111111111112, | |
| "loss": 2.3155, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007666666666666667, | |
| "loss": 2.3303, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007662222222222222, | |
| "loss": 2.2867, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007657777777777778, | |
| "loss": 2.39, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007653333333333334, | |
| "loss": 2.2372, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000764888888888889, | |
| "loss": 2.3359, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007644444444444445, | |
| "loss": 2.2477, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000764, | |
| "loss": 2.1907, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007635555555555556, | |
| "loss": 2.3424, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007631111111111111, | |
| "loss": 2.254, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007626666666666667, | |
| "loss": 2.2896, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007622222222222222, | |
| "loss": 2.2964, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007617777777777778, | |
| "loss": 2.2756, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007613333333333334, | |
| "loss": 2.3348, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000760888888888889, | |
| "loss": 2.3344, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0007604444444444446, | |
| "loss": 2.2948, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00076, | |
| "loss": 2.3696, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007595555555555555, | |
| "loss": 2.3261, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007591111111111111, | |
| "loss": 2.3931, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007586666666666667, | |
| "loss": 2.3551, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007582222222222223, | |
| "loss": 2.3412, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007577777777777778, | |
| "loss": 2.3517, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007573333333333334, | |
| "loss": 2.3178, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.000756888888888889, | |
| "loss": 2.3577, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007564444444444445, | |
| "loss": 2.2807, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007559999999999999, | |
| "loss": 2.3332, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007555555555555555, | |
| "loss": 2.3292, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007551111111111111, | |
| "loss": 2.2887, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007546666666666667, | |
| "loss": 2.3498, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007542222222222223, | |
| "loss": 2.3857, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007537777777777778, | |
| "loss": 2.3264, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007533333333333334, | |
| "loss": 2.2637, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007528888888888889, | |
| "loss": 2.3084, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007524444444444445, | |
| "loss": 2.3164, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.000752, | |
| "loss": 2.3356, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007515555555555555, | |
| "loss": 2.3407, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0007511111111111111, | |
| "loss": 2.3375, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007506666666666667, | |
| "loss": 2.3178, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007502222222222223, | |
| "loss": 2.379, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007497777777777778, | |
| "loss": 2.3506, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007493333333333334, | |
| "loss": 2.3232, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007488888888888889, | |
| "loss": 2.3266, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007484444444444445, | |
| "loss": 2.2864, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007480000000000001, | |
| "loss": 2.2662, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007475555555555556, | |
| "loss": 2.3208, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007471111111111112, | |
| "loss": 2.3149, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007466666666666667, | |
| "loss": 2.2972, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007462222222222223, | |
| "loss": 2.4095, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007457777777777778, | |
| "loss": 2.3448, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007453333333333333, | |
| "loss": 2.2673, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007448888888888889, | |
| "loss": 2.345, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007444444444444445, | |
| "loss": 2.2822, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007440000000000001, | |
| "loss": 2.2933, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007435555555555556, | |
| "loss": 2.2883, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007431111111111112, | |
| "loss": 2.2863, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0007426666666666668, | |
| "loss": 2.273, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007422222222222222, | |
| "loss": 2.281, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007417777777777777, | |
| "loss": 2.2877, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007413333333333333, | |
| "loss": 2.2801, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007408888888888889, | |
| "loss": 2.3044, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007404444444444445, | |
| "loss": 2.3417, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007400000000000001, | |
| "loss": 2.3744, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007395555555555556, | |
| "loss": 2.3285, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007391111111111112, | |
| "loss": 2.3032, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007386666666666667, | |
| "loss": 2.275, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007382222222222223, | |
| "loss": 2.3184, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007377777777777778, | |
| "loss": 2.4252, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007373333333333333, | |
| "loss": 2.3806, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007368888888888889, | |
| "loss": 2.4038, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007364444444444445, | |
| "loss": 2.4387, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007360000000000001, | |
| "loss": 2.2712, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007355555555555556, | |
| "loss": 2.3618, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007351111111111111, | |
| "loss": 2.3015, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007346666666666667, | |
| "loss": 2.2618, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007342222222222223, | |
| "loss": 2.3319, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007337777777777779, | |
| "loss": 2.2604, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0007333333333333333, | |
| "loss": 2.3124, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007328888888888889, | |
| "loss": 2.3053, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007324444444444445, | |
| "loss": 2.2509, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000732, | |
| "loss": 2.2795, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007315555555555555, | |
| "loss": 2.3321, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007311111111111111, | |
| "loss": 2.3181, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007306666666666667, | |
| "loss": 2.2793, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007302222222222223, | |
| "loss": 2.2477, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007297777777777779, | |
| "loss": 2.3584, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007293333333333334, | |
| "loss": 2.2824, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007288888888888889, | |
| "loss": 2.2678, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007284444444444445, | |
| "loss": 2.2886, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000728, | |
| "loss": 2.3444, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007275555555555556, | |
| "loss": 2.2967, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007271111111111111, | |
| "loss": 2.334, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007266666666666667, | |
| "loss": 2.2629, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007262222222222223, | |
| "loss": 2.3195, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007257777777777779, | |
| "loss": 2.2583, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007253333333333334, | |
| "loss": 2.2994, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000724888888888889, | |
| "loss": 2.307, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0007244444444444444, | |
| "loss": 2.3516, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000724, | |
| "loss": 2.3624, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007235555555555556, | |
| "loss": 2.3008, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007231111111111111, | |
| "loss": 2.3502, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007226666666666667, | |
| "loss": 2.3483, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007222222222222223, | |
| "loss": 2.3154, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007217777777777779, | |
| "loss": 2.2634, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007213333333333334, | |
| "loss": 2.2936, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007208888888888889, | |
| "loss": 2.3522, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007204444444444445, | |
| "loss": 2.2925, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00072, | |
| "loss": 2.2873, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007195555555555556, | |
| "loss": 2.3467, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007191111111111111, | |
| "loss": 2.3031, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007186666666666667, | |
| "loss": 2.3181, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007182222222222223, | |
| "loss": 2.3257, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007177777777777778, | |
| "loss": 2.2915, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007173333333333333, | |
| "loss": 2.3113, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007168888888888889, | |
| "loss": 2.3474, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007164444444444445, | |
| "loss": 2.226, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0007160000000000001, | |
| "loss": 2.2839, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007155555555555556, | |
| "loss": 2.2699, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007151111111111111, | |
| "loss": 2.2928, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007146666666666667, | |
| "loss": 2.2372, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007142222222222222, | |
| "loss": 2.2812, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007137777777777778, | |
| "loss": 2.265, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007133333333333334, | |
| "loss": 2.2571, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007128888888888889, | |
| "loss": 2.3748, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007124444444444445, | |
| "loss": 2.2414, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007120000000000001, | |
| "loss": 2.2718, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007115555555555557, | |
| "loss": 2.2649, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007111111111111111, | |
| "loss": 2.2645, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.1945419311523438, | |
| "eval_rouge1": 12.1435, | |
| "eval_rouge2": 4.0018, | |
| "eval_rougeL": 7.5854, | |
| "eval_rougeLsum": 7.5909, | |
| "eval_runtime": 9033.7247, | |
| "eval_samples_per_second": 0.055, | |
| "eval_steps_per_second": 0.055, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007106666666666666, | |
| "loss": 2.2846, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007102222222222222, | |
| "loss": 2.2636, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007097777777777778, | |
| "loss": 2.3398, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007093333333333334, | |
| "loss": 2.265, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007088888888888889, | |
| "loss": 2.2955, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007084444444444445, | |
| "loss": 2.3476, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007080000000000001, | |
| "loss": 2.3451, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007075555555555556, | |
| "loss": 2.308, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.000707111111111111, | |
| "loss": 2.2872, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0007066666666666666, | |
| "loss": 2.2483, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007062222222222222, | |
| "loss": 2.2962, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007057777777777778, | |
| "loss": 2.3136, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007053333333333334, | |
| "loss": 2.2673, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007048888888888889, | |
| "loss": 2.2162, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007044444444444445, | |
| "loss": 2.2537, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007040000000000001, | |
| "loss": 2.2983, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007035555555555556, | |
| "loss": 2.2838, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007031111111111112, | |
| "loss": 2.2714, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007026666666666667, | |
| "loss": 2.2799, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007022222222222222, | |
| "loss": 2.3028, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007017777777777778, | |
| "loss": 2.2599, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007013333333333334, | |
| "loss": 2.318, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007008888888888889, | |
| "loss": 2.2827, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007004444444444445, | |
| "loss": 2.3303, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0007, | |
| "loss": 2.3569, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0006995555555555556, | |
| "loss": 2.2757, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0006991111111111112, | |
| "loss": 2.3166, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0006986666666666667, | |
| "loss": 2.2164, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0006982222222222223, | |
| "loss": 2.259, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0006977777777777778, | |
| "loss": 2.3298, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006973333333333334, | |
| "loss": 2.2499, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006968888888888889, | |
| "loss": 2.3553, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006964444444444444, | |
| "loss": 2.2167, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.000696, | |
| "loss": 2.303, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006955555555555556, | |
| "loss": 2.3341, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006951111111111112, | |
| "loss": 2.3226, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006946666666666667, | |
| "loss": 2.3525, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006942222222222223, | |
| "loss": 2.3144, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006937777777777779, | |
| "loss": 2.3401, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006933333333333333, | |
| "loss": 2.2758, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006928888888888889, | |
| "loss": 2.2498, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006924444444444444, | |
| "loss": 2.2928, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.000692, | |
| "loss": 2.2825, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006915555555555556, | |
| "loss": 2.2283, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006911111111111112, | |
| "loss": 2.3107, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006906666666666667, | |
| "loss": 2.3323, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006902222222222223, | |
| "loss": 2.3402, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006897777777777778, | |
| "loss": 2.3507, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0006893333333333334, | |
| "loss": 2.3219, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000688888888888889, | |
| "loss": 2.3656, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006884444444444444, | |
| "loss": 2.2787, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000688, | |
| "loss": 2.3104, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006875555555555556, | |
| "loss": 2.2685, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006871111111111112, | |
| "loss": 2.293, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006866666666666667, | |
| "loss": 2.2669, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006862222222222222, | |
| "loss": 2.2529, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006857777777777778, | |
| "loss": 2.3113, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006853333333333334, | |
| "loss": 2.3299, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000684888888888889, | |
| "loss": 2.3506, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006844444444444444, | |
| "loss": 2.2797, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000684, | |
| "loss": 2.3021, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006835555555555556, | |
| "loss": 2.2821, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006831111111111111, | |
| "loss": 2.3057, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006826666666666667, | |
| "loss": 2.2947, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006822222222222222, | |
| "loss": 2.2819, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006817777777777778, | |
| "loss": 2.2521, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006813333333333334, | |
| "loss": 2.2611, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.000680888888888889, | |
| "loss": 2.2953, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0006804444444444445, | |
| "loss": 2.2593, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00068, | |
| "loss": 2.311, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006795555555555556, | |
| "loss": 2.3297, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006791111111111111, | |
| "loss": 2.3235, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006786666666666667, | |
| "loss": 2.2148, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006782222222222222, | |
| "loss": 2.2254, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006777777777777778, | |
| "loss": 2.2851, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006773333333333334, | |
| "loss": 2.3619, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.000676888888888889, | |
| "loss": 2.2525, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006764444444444445, | |
| "loss": 2.2553, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006760000000000001, | |
| "loss": 2.3169, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006755555555555555, | |
| "loss": 2.3435, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006751111111111111, | |
| "loss": 2.2737, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006746666666666667, | |
| "loss": 2.2905, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006742222222222222, | |
| "loss": 2.242, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006737777777777778, | |
| "loss": 2.2425, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006733333333333334, | |
| "loss": 2.3406, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.000672888888888889, | |
| "loss": 2.3018, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006724444444444445, | |
| "loss": 2.3139, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.000672, | |
| "loss": 2.3142, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006715555555555556, | |
| "loss": 2.3257, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0006711111111111111, | |
| "loss": 2.3662, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006706666666666667, | |
| "loss": 2.3062, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006702222222222222, | |
| "loss": 2.321, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006697777777777778, | |
| "loss": 2.3185, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006693333333333334, | |
| "loss": 2.3271, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006688888888888889, | |
| "loss": 2.3214, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006684444444444445, | |
| "loss": 2.2758, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000668, | |
| "loss": 2.2785, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006675555555555556, | |
| "loss": 2.3329, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006671111111111112, | |
| "loss": 2.2952, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006666666666666668, | |
| "loss": 2.3203, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006662222222222222, | |
| "loss": 2.3027, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006657777777777778, | |
| "loss": 2.2896, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006653333333333333, | |
| "loss": 2.2812, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006648888888888889, | |
| "loss": 2.2475, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006644444444444445, | |
| "loss": 2.2417, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000664, | |
| "loss": 2.3216, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006635555555555556, | |
| "loss": 2.2549, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006631111111111112, | |
| "loss": 2.2115, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0006626666666666668, | |
| "loss": 2.268, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006622222222222222, | |
| "loss": 2.2714, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006617777777777777, | |
| "loss": 2.3372, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006613333333333333, | |
| "loss": 2.2497, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006608888888888889, | |
| "loss": 2.2294, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006604444444444445, | |
| "loss": 2.3672, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00066, | |
| "loss": 2.2738, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006595555555555556, | |
| "loss": 2.2095, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006591111111111112, | |
| "loss": 2.2842, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006586666666666667, | |
| "loss": 2.2827, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006582222222222223, | |
| "loss": 2.2626, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006577777777777777, | |
| "loss": 2.2834, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006573333333333333, | |
| "loss": 2.2697, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006568888888888889, | |
| "loss": 2.2912, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006564444444444445, | |
| "loss": 2.1689, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000656, | |
| "loss": 2.1933, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006555555555555556, | |
| "loss": 2.2468, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006551111111111112, | |
| "loss": 2.2942, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006546666666666667, | |
| "loss": 2.3006, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006542222222222223, | |
| "loss": 2.3068, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0006537777777777778, | |
| "loss": 2.2492, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006533333333333333, | |
| "loss": 2.2692, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006528888888888889, | |
| "loss": 2.3024, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006524444444444445, | |
| "loss": 2.2746, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000652, | |
| "loss": 2.3316, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006515555555555556, | |
| "loss": 2.2942, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006511111111111111, | |
| "loss": 2.3328, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006506666666666667, | |
| "loss": 2.2542, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006502222222222223, | |
| "loss": 2.2412, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006497777777777778, | |
| "loss": 2.3459, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006493333333333334, | |
| "loss": 2.3569, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000648888888888889, | |
| "loss": 2.3715, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006484444444444445, | |
| "loss": 2.2707, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000648, | |
| "loss": 2.306, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006475555555555555, | |
| "loss": 2.3753, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006471111111111111, | |
| "loss": 2.2649, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006466666666666667, | |
| "loss": 2.3003, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006462222222222223, | |
| "loss": 2.3531, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006457777777777778, | |
| "loss": 2.3344, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0006453333333333334, | |
| "loss": 2.3191, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000644888888888889, | |
| "loss": 2.2959, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006444444444444444, | |
| "loss": 2.27, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000644, | |
| "loss": 2.2969, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006435555555555555, | |
| "loss": 2.289, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006431111111111111, | |
| "loss": 2.3174, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006426666666666667, | |
| "loss": 2.2707, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006422222222222223, | |
| "loss": 2.3502, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006417777777777778, | |
| "loss": 2.3211, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006413333333333334, | |
| "loss": 2.3761, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006408888888888889, | |
| "loss": 2.3187, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006404444444444445, | |
| "loss": 2.261, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00064, | |
| "loss": 2.3179, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006395555555555555, | |
| "loss": 2.2901, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006391111111111111, | |
| "loss": 2.3096, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006386666666666667, | |
| "loss": 2.2477, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006382222222222223, | |
| "loss": 2.3581, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006377777777777778, | |
| "loss": 2.3848, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006373333333333333, | |
| "loss": 2.2378, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006368888888888889, | |
| "loss": 2.2877, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006364444444444445, | |
| "loss": 2.2103, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006360000000000001, | |
| "loss": 2.2478, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0006355555555555555, | |
| "loss": 2.3312, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006351111111111111, | |
| "loss": 2.2237, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006346666666666667, | |
| "loss": 2.2685, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006342222222222223, | |
| "loss": 2.2737, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006337777777777778, | |
| "loss": 2.256, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006333333333333333, | |
| "loss": 2.2408, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006328888888888889, | |
| "loss": 2.2518, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006324444444444445, | |
| "loss": 2.3739, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006320000000000001, | |
| "loss": 2.2377, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006315555555555556, | |
| "loss": 2.3264, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006311111111111112, | |
| "loss": 2.2435, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006306666666666667, | |
| "loss": 2.3089, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006302222222222222, | |
| "loss": 2.3225, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006297777777777778, | |
| "loss": 2.3002, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006293333333333333, | |
| "loss": 2.3349, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006288888888888889, | |
| "loss": 2.3076, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006284444444444445, | |
| "loss": 2.2994, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006280000000000001, | |
| "loss": 2.2747, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006275555555555557, | |
| "loss": 2.3033, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006271111111111112, | |
| "loss": 2.2344, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0006266666666666666, | |
| "loss": 2.2424, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006262222222222222, | |
| "loss": 2.26, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006257777777777778, | |
| "loss": 2.2474, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006253333333333333, | |
| "loss": 2.2715, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006248888888888889, | |
| "loss": 2.2424, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006244444444444445, | |
| "loss": 2.2571, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006240000000000001, | |
| "loss": 2.284, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006235555555555556, | |
| "loss": 2.3298, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006231111111111111, | |
| "loss": 2.2766, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006226666666666667, | |
| "loss": 2.368, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006222222222222223, | |
| "loss": 2.3322, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.1741747856140137, | |
| "eval_rouge1": 12.0956, | |
| "eval_rouge2": 3.9621, | |
| "eval_rougeL": 7.843, | |
| "eval_rougeLsum": 7.8272, | |
| "eval_runtime": 9059.7927, | |
| "eval_samples_per_second": 0.055, | |
| "eval_steps_per_second": 0.055, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006217777777777778, | |
| "loss": 2.2629, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006213333333333333, | |
| "loss": 2.2242, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006208888888888889, | |
| "loss": 2.2483, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006204444444444445, | |
| "loss": 2.2495, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00062, | |
| "loss": 2.313, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006195555555555556, | |
| "loss": 2.2913, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006191111111111111, | |
| "loss": 2.3303, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006186666666666667, | |
| "loss": 2.2479, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006182222222222223, | |
| "loss": 2.2751, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0006177777777777779, | |
| "loss": 2.2311, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006173333333333333, | |
| "loss": 2.2294, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006168888888888889, | |
| "loss": 2.2305, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006164444444444444, | |
| "loss": 2.277, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000616, | |
| "loss": 2.347, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006155555555555556, | |
| "loss": 2.2763, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006151111111111111, | |
| "loss": 2.3201, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006146666666666667, | |
| "loss": 2.2555, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006142222222222223, | |
| "loss": 2.3702, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006137777777777779, | |
| "loss": 2.2319, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006133333333333334, | |
| "loss": 2.2326, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006128888888888888, | |
| "loss": 2.3561, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006124444444444444, | |
| "loss": 2.2702, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.000612, | |
| "loss": 2.2726, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006115555555555556, | |
| "loss": 2.3036, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006111111111111111, | |
| "loss": 2.316, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006106666666666667, | |
| "loss": 2.2936, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006102222222222223, | |
| "loss": 2.3333, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006097777777777779, | |
| "loss": 2.3695, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0006093333333333334, | |
| "loss": 2.1621, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006088888888888888, | |
| "loss": 2.288, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006084444444444444, | |
| "loss": 2.275, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.000608, | |
| "loss": 2.2326, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006075555555555556, | |
| "loss": 2.3058, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006071111111111111, | |
| "loss": 2.3021, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006066666666666667, | |
| "loss": 2.2709, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006062222222222223, | |
| "loss": 2.2401, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006057777777777778, | |
| "loss": 2.2679, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006053333333333334, | |
| "loss": 2.2649, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006048888888888889, | |
| "loss": 2.2832, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006044444444444445, | |
| "loss": 2.2549, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.000604, | |
| "loss": 2.2033, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006035555555555556, | |
| "loss": 2.3478, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006031111111111112, | |
| "loss": 2.2811, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006026666666666667, | |
| "loss": 2.2695, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006022222222222222, | |
| "loss": 2.3959, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006017777777777778, | |
| "loss": 2.2135, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006013333333333334, | |
| "loss": 2.2501, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006008888888888889, | |
| "loss": 2.2801, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0006004444444444445, | |
| "loss": 2.2968, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0006000000000000001, | |
| "loss": 2.2578, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005995555555555556, | |
| "loss": 2.1764, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005991111111111111, | |
| "loss": 2.3049, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005986666666666666, | |
| "loss": 2.2893, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005982222222222222, | |
| "loss": 2.2749, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005977777777777778, | |
| "loss": 2.3037, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005973333333333334, | |
| "loss": 2.2434, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005968888888888889, | |
| "loss": 2.3198, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005964444444444445, | |
| "loss": 2.2131, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005960000000000001, | |
| "loss": 2.2619, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005955555555555556, | |
| "loss": 2.2657, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005951111111111111, | |
| "loss": 2.2787, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005946666666666666, | |
| "loss": 2.3317, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005942222222222222, | |
| "loss": 2.3017, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005937777777777778, | |
| "loss": 2.2839, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005933333333333334, | |
| "loss": 2.2235, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000592888888888889, | |
| "loss": 2.191, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005924444444444445, | |
| "loss": 2.2485, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.000592, | |
| "loss": 2.2507, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0005915555555555556, | |
| "loss": 2.2717, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005911111111111112, | |
| "loss": 2.1896, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005906666666666666, | |
| "loss": 2.3243, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005902222222222222, | |
| "loss": 2.1593, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005897777777777778, | |
| "loss": 2.2796, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005893333333333334, | |
| "loss": 2.2769, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005888888888888889, | |
| "loss": 2.3809, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005884444444444445, | |
| "loss": 2.2652, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.000588, | |
| "loss": 2.2152, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005875555555555556, | |
| "loss": 2.18, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005871111111111112, | |
| "loss": 2.2659, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005866666666666667, | |
| "loss": 2.2558, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005862222222222222, | |
| "loss": 2.2737, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005857777777777778, | |
| "loss": 2.2537, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005853333333333334, | |
| "loss": 2.2587, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005848888888888889, | |
| "loss": 2.2707, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005844444444444444, | |
| "loss": 2.2646, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.000584, | |
| "loss": 2.2837, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005835555555555556, | |
| "loss": 2.2998, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005831111111111112, | |
| "loss": 2.2286, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005826666666666668, | |
| "loss": 2.2666, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0005822222222222223, | |
| "loss": 2.2959, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005817777777777778, | |
| "loss": 2.3227, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005813333333333333, | |
| "loss": 2.3145, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005808888888888889, | |
| "loss": 2.2154, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005804444444444444, | |
| "loss": 2.2541, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00058, | |
| "loss": 2.2052, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005795555555555556, | |
| "loss": 2.236, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005791111111111112, | |
| "loss": 2.2407, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005786666666666668, | |
| "loss": 2.2687, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005782222222222223, | |
| "loss": 2.2907, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005777777777777778, | |
| "loss": 2.2697, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005773333333333333, | |
| "loss": 2.2348, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005768888888888889, | |
| "loss": 2.2328, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005764444444444444, | |
| "loss": 2.2347, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.000576, | |
| "loss": 2.2297, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005755555555555556, | |
| "loss": 2.2458, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005751111111111112, | |
| "loss": 2.2093, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005746666666666667, | |
| "loss": 2.227, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005742222222222222, | |
| "loss": 2.3247, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005737777777777778, | |
| "loss": 2.2955, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0005733333333333334, | |
| "loss": 2.2264, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000572888888888889, | |
| "loss": 2.2543, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005724444444444444, | |
| "loss": 2.3499, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000572, | |
| "loss": 2.2414, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005715555555555556, | |
| "loss": 2.2703, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005711111111111111, | |
| "loss": 2.3098, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005706666666666667, | |
| "loss": 2.2562, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005702222222222222, | |
| "loss": 2.2553, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005697777777777778, | |
| "loss": 2.2428, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005693333333333334, | |
| "loss": 2.2984, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000568888888888889, | |
| "loss": 2.2823, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005684444444444446, | |
| "loss": 2.3694, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000568, | |
| "loss": 2.2332, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005675555555555555, | |
| "loss": 2.2421, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005671111111111111, | |
| "loss": 2.251, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005666666666666667, | |
| "loss": 2.1915, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005662222222222222, | |
| "loss": 2.2584, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005657777777777778, | |
| "loss": 2.2784, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005653333333333334, | |
| "loss": 2.2584, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000564888888888889, | |
| "loss": 2.2316, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0005644444444444445, | |
| "loss": 2.2502, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000564, | |
| "loss": 2.2364, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005635555555555555, | |
| "loss": 2.2743, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005631111111111111, | |
| "loss": 2.2735, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005626666666666667, | |
| "loss": 2.2353, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005622222222222222, | |
| "loss": 2.2888, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005617777777777778, | |
| "loss": 2.1532, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005613333333333334, | |
| "loss": 2.2746, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000560888888888889, | |
| "loss": 2.2409, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005604444444444445, | |
| "loss": 2.2122, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00056, | |
| "loss": 2.2223, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005595555555555555, | |
| "loss": 2.228, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005591111111111111, | |
| "loss": 2.2854, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005586666666666667, | |
| "loss": 2.2992, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005582222222222223, | |
| "loss": 2.2765, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005577777777777778, | |
| "loss": 2.2645, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005573333333333334, | |
| "loss": 2.2738, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005568888888888889, | |
| "loss": 2.2665, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0005564444444444445, | |
| "loss": 2.1959, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.000556, | |
| "loss": 2.3094, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005555555555555556, | |
| "loss": 2.3302, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005551111111111111, | |
| "loss": 2.2088, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005546666666666667, | |
| "loss": 2.2774, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005542222222222223, | |
| "loss": 2.2279, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005537777777777778, | |
| "loss": 2.2091, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005533333333333333, | |
| "loss": 2.3221, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005528888888888889, | |
| "loss": 2.3202, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005524444444444445, | |
| "loss": 2.2552, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.000552, | |
| "loss": 2.2166, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005515555555555556, | |
| "loss": 2.286, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005511111111111112, | |
| "loss": 2.2642, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005506666666666668, | |
| "loss": 2.3546, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005502222222222222, | |
| "loss": 2.2763, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005497777777777777, | |
| "loss": 2.2846, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005493333333333333, | |
| "loss": 2.3388, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005488888888888889, | |
| "loss": 2.1978, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005484444444444445, | |
| "loss": 2.2363, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005480000000000001, | |
| "loss": 2.1851, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005475555555555556, | |
| "loss": 2.2744, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0005471111111111112, | |
| "loss": 2.3183, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005466666666666667, | |
| "loss": 2.1745, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005462222222222222, | |
| "loss": 2.2923, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005457777777777777, | |
| "loss": 2.3098, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005453333333333333, | |
| "loss": 2.2963, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005448888888888889, | |
| "loss": 2.2082, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005444444444444445, | |
| "loss": 2.222, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005440000000000001, | |
| "loss": 2.2265, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005435555555555556, | |
| "loss": 2.2619, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005431111111111111, | |
| "loss": 2.2613, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005426666666666667, | |
| "loss": 2.3053, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005422222222222223, | |
| "loss": 2.203, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005417777777777777, | |
| "loss": 2.2989, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005413333333333333, | |
| "loss": 2.2543, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005408888888888889, | |
| "loss": 2.2812, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005404444444444445, | |
| "loss": 2.2044, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005400000000000001, | |
| "loss": 2.2387, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005395555555555556, | |
| "loss": 2.2621, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005391111111111111, | |
| "loss": 2.2329, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005386666666666667, | |
| "loss": 2.2585, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0005382222222222223, | |
| "loss": 2.2237, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005377777777777779, | |
| "loss": 2.2546, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005373333333333333, | |
| "loss": 2.1964, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005368888888888889, | |
| "loss": 2.179, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005364444444444445, | |
| "loss": 2.2903, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.000536, | |
| "loss": 2.2472, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005355555555555555, | |
| "loss": 2.2579, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005351111111111111, | |
| "loss": 2.2336, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005346666666666667, | |
| "loss": 2.2474, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005342222222222223, | |
| "loss": 2.2482, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005337777777777779, | |
| "loss": 2.2413, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005333333333333334, | |
| "loss": 2.307, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.158494710922241, | |
| "eval_rouge1": 13.1246, | |
| "eval_rouge2": 4.3442, | |
| "eval_rougeL": 8.3035, | |
| "eval_rougeLsum": 8.2848, | |
| "eval_runtime": 9293.9809, | |
| "eval_samples_per_second": 0.054, | |
| "eval_steps_per_second": 0.054, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.000532888888888889, | |
| "loss": 2.2131, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005324444444444444, | |
| "loss": 2.2893, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.000532, | |
| "loss": 2.2613, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005315555555555555, | |
| "loss": 2.2145, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005311111111111111, | |
| "loss": 2.2878, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005306666666666667, | |
| "loss": 2.3126, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005302222222222223, | |
| "loss": 2.2385, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005297777777777779, | |
| "loss": 2.2998, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0005293333333333334, | |
| "loss": 2.3374, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005288888888888889, | |
| "loss": 2.2379, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005284444444444444, | |
| "loss": 2.2535, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000528, | |
| "loss": 2.2459, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005275555555555556, | |
| "loss": 2.2305, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005271111111111111, | |
| "loss": 2.2622, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005266666666666667, | |
| "loss": 2.2552, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005262222222222223, | |
| "loss": 2.1605, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005257777777777778, | |
| "loss": 2.2563, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005253333333333333, | |
| "loss": 2.283, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005248888888888889, | |
| "loss": 2.222, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005244444444444445, | |
| "loss": 2.2059, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000524, | |
| "loss": 2.3158, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005235555555555556, | |
| "loss": 2.2172, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005231111111111111, | |
| "loss": 2.3396, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005226666666666667, | |
| "loss": 2.2428, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005222222222222222, | |
| "loss": 2.2124, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005217777777777778, | |
| "loss": 2.2991, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005213333333333333, | |
| "loss": 2.269, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005208888888888889, | |
| "loss": 2.2231, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005204444444444445, | |
| "loss": 2.2159, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0005200000000000001, | |
| "loss": 2.2804, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005195555555555557, | |
| "loss": 2.2807, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005191111111111111, | |
| "loss": 2.2809, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005186666666666666, | |
| "loss": 2.2411, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005182222222222222, | |
| "loss": 2.2695, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005177777777777778, | |
| "loss": 2.2816, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005173333333333333, | |
| "loss": 2.3263, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005168888888888889, | |
| "loss": 2.2066, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005164444444444445, | |
| "loss": 2.2367, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005160000000000001, | |
| "loss": 2.2495, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005155555555555557, | |
| "loss": 2.2021, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005151111111111112, | |
| "loss": 2.2198, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005146666666666666, | |
| "loss": 2.2663, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005142222222222222, | |
| "loss": 2.2334, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005137777777777778, | |
| "loss": 2.2378, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005133333333333334, | |
| "loss": 2.3457, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005128888888888889, | |
| "loss": 2.2211, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005124444444444445, | |
| "loss": 2.2252, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005120000000000001, | |
| "loss": 2.2159, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005115555555555556, | |
| "loss": 2.2757, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0005111111111111111, | |
| "loss": 2.2497, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005106666666666666, | |
| "loss": 2.2116, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005102222222222222, | |
| "loss": 2.2416, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005097777777777778, | |
| "loss": 2.27, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005093333333333334, | |
| "loss": 2.2762, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005088888888888889, | |
| "loss": 2.3451, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005084444444444445, | |
| "loss": 2.2497, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.000508, | |
| "loss": 2.2762, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005075555555555556, | |
| "loss": 2.2514, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005071111111111111, | |
| "loss": 2.2718, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005066666666666667, | |
| "loss": 2.3076, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005062222222222223, | |
| "loss": 2.2889, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005057777777777778, | |
| "loss": 2.2223, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005053333333333334, | |
| "loss": 2.3181, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005048888888888889, | |
| "loss": 2.1922, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005044444444444444, | |
| "loss": 2.2745, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.000504, | |
| "loss": 2.1848, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005035555555555556, | |
| "loss": 2.2627, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005031111111111112, | |
| "loss": 2.3022, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005026666666666667, | |
| "loss": 2.2763, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0005022222222222223, | |
| "loss": 2.1709, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005017777777777779, | |
| "loss": 2.2644, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005013333333333333, | |
| "loss": 2.2751, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005008888888888888, | |
| "loss": 2.1292, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005004444444444444, | |
| "loss": 2.2523, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0005, | |
| "loss": 2.2041, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004995555555555556, | |
| "loss": 2.2942, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004991111111111112, | |
| "loss": 2.2585, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004986666666666667, | |
| "loss": 2.2354, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004982222222222223, | |
| "loss": 2.1996, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004977777777777778, | |
| "loss": 2.2347, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004973333333333334, | |
| "loss": 2.2003, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004968888888888888, | |
| "loss": 2.184, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004964444444444444, | |
| "loss": 2.2678, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.000496, | |
| "loss": 2.2168, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004955555555555556, | |
| "loss": 2.3202, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004951111111111112, | |
| "loss": 2.2037, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004946666666666667, | |
| "loss": 2.1942, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004942222222222223, | |
| "loss": 2.2402, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0004937777777777778, | |
| "loss": 2.2633, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004933333333333334, | |
| "loss": 2.334, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000492888888888889, | |
| "loss": 2.2998, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004924444444444444, | |
| "loss": 2.1723, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000492, | |
| "loss": 2.2382, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004915555555555556, | |
| "loss": 2.2423, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004911111111111112, | |
| "loss": 2.2987, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004906666666666667, | |
| "loss": 2.3003, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004902222222222222, | |
| "loss": 2.2559, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004897777777777778, | |
| "loss": 2.1604, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004893333333333334, | |
| "loss": 2.2234, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000488888888888889, | |
| "loss": 2.2679, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004884444444444445, | |
| "loss": 2.2542, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.000488, | |
| "loss": 2.2124, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004875555555555556, | |
| "loss": 2.2489, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004871111111111112, | |
| "loss": 2.1796, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00048666666666666666, | |
| "loss": 2.2739, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004862222222222222, | |
| "loss": 2.236, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004857777777777778, | |
| "loss": 2.2216, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004853333333333334, | |
| "loss": 2.2743, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0004848888888888889, | |
| "loss": 2.2245, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004844444444444444, | |
| "loss": 2.2072, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.000484, | |
| "loss": 2.1766, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004835555555555556, | |
| "loss": 2.2321, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00048311111111111113, | |
| "loss": 2.2446, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004826666666666667, | |
| "loss": 2.2288, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004822222222222222, | |
| "loss": 2.1877, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004817777777777778, | |
| "loss": 2.1947, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004813333333333334, | |
| "loss": 2.209, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00048088888888888893, | |
| "loss": 2.2743, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004804444444444444, | |
| "loss": 2.2639, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00048, | |
| "loss": 2.2736, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004795555555555556, | |
| "loss": 2.1983, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00047911111111111114, | |
| "loss": 2.2792, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00047866666666666674, | |
| "loss": 2.2279, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004782222222222222, | |
| "loss": 2.256, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004777777777777778, | |
| "loss": 2.2014, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00047733333333333335, | |
| "loss": 2.2711, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00047688888888888895, | |
| "loss": 2.2451, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00047644444444444443, | |
| "loss": 2.194, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.000476, | |
| "loss": 2.2712, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047555555555555556, | |
| "loss": 2.1745, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047511111111111116, | |
| "loss": 2.2669, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047466666666666675, | |
| "loss": 2.2416, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047422222222222223, | |
| "loss": 2.3204, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047377777777777777, | |
| "loss": 2.2696, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047333333333333336, | |
| "loss": 2.3084, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047288888888888896, | |
| "loss": 2.1746, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047244444444444444, | |
| "loss": 2.2771, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.000472, | |
| "loss": 2.263, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004715555555555556, | |
| "loss": 2.3035, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00047111111111111117, | |
| "loss": 2.3043, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004706666666666667, | |
| "loss": 2.2641, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004702222222222222, | |
| "loss": 2.3417, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004697777777777778, | |
| "loss": 2.2475, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004693333333333334, | |
| "loss": 2.2561, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004688888888888889, | |
| "loss": 2.2323, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004684444444444445, | |
| "loss": 2.3017, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.000468, | |
| "loss": 2.2547, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004675555555555556, | |
| "loss": 2.2893, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004671111111111111, | |
| "loss": 2.2159, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0004666666666666667, | |
| "loss": 2.2515, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0004662222222222222, | |
| "loss": 2.3363, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0004657777777777778, | |
| "loss": 2.2332, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00046533333333333333, | |
| "loss": 2.2265, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0004648888888888889, | |
| "loss": 2.2548, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0004644444444444445, | |
| "loss": 2.2425, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.000464, | |
| "loss": 2.2064, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00046355555555555554, | |
| "loss": 2.1424, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00046311111111111113, | |
| "loss": 2.2388, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00046266666666666673, | |
| "loss": 2.2085, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0004622222222222222, | |
| "loss": 2.2261, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00046177777777777775, | |
| "loss": 2.3124, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00046133333333333334, | |
| "loss": 2.1604, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00046088888888888894, | |
| "loss": 2.31, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0004604444444444445, | |
| "loss": 2.2804, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00045999999999999996, | |
| "loss": 2.2126, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00045955555555555555, | |
| "loss": 2.3328, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00045911111111111115, | |
| "loss": 2.1793, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0004586666666666667, | |
| "loss": 2.2972, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0004582222222222223, | |
| "loss": 2.2291, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00045777777777777776, | |
| "loss": 2.2419, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00045733333333333335, | |
| "loss": 2.1676, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00045688888888888895, | |
| "loss": 2.2912, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004564444444444445, | |
| "loss": 2.2042, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00045599999999999997, | |
| "loss": 2.2094, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00045555555555555556, | |
| "loss": 2.2714, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00045511111111111116, | |
| "loss": 2.239, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004546666666666667, | |
| "loss": 2.2624, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004542222222222223, | |
| "loss": 2.2396, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004537777777777778, | |
| "loss": 2.2247, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00045333333333333337, | |
| "loss": 2.3075, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004528888888888889, | |
| "loss": 2.2015, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004524444444444445, | |
| "loss": 2.266, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.000452, | |
| "loss": 2.2702, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004515555555555556, | |
| "loss": 2.1914, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004511111111111111, | |
| "loss": 2.2207, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004506666666666667, | |
| "loss": 2.2352, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004502222222222223, | |
| "loss": 2.2538, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004497777777777778, | |
| "loss": 2.3808, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004493333333333333, | |
| "loss": 2.2134, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.0004488888888888889, | |
| "loss": 2.2423, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0004484444444444445, | |
| "loss": 2.1774, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044800000000000005, | |
| "loss": 2.1726, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044755555555555553, | |
| "loss": 2.2462, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0004471111111111111, | |
| "loss": 2.2331, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0004466666666666667, | |
| "loss": 2.2157, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044622222222222226, | |
| "loss": 2.2249, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044577777777777774, | |
| "loss": 2.2261, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044533333333333333, | |
| "loss": 2.2146, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0004448888888888889, | |
| "loss": 2.2727, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044444444444444447, | |
| "loss": 2.257, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.1341190338134766, | |
| "eval_rouge1": 13.159, | |
| "eval_rouge2": 4.3165, | |
| "eval_rougeL": 8.2736, | |
| "eval_rougeLsum": 8.2843, | |
| "eval_runtime": 9293.0241, | |
| "eval_samples_per_second": 0.054, | |
| "eval_steps_per_second": 0.054, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044400000000000006, | |
| "loss": 2.2323, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044355555555555554, | |
| "loss": 2.3168, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044311111111111114, | |
| "loss": 2.2113, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0004426666666666667, | |
| "loss": 2.2115, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044222222222222227, | |
| "loss": 2.2126, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044177777777777775, | |
| "loss": 2.1807, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00044133333333333335, | |
| "loss": 2.2357, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0004408888888888889, | |
| "loss": 2.1952, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0004404444444444445, | |
| "loss": 2.2606, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00044000000000000007, | |
| "loss": 2.1503, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00043955555555555555, | |
| "loss": 2.2449, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004391111111111111, | |
| "loss": 2.1983, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004386666666666667, | |
| "loss": 2.1528, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004382222222222223, | |
| "loss": 2.1265, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004377777777777778, | |
| "loss": 2.2405, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004373333333333333, | |
| "loss": 2.2103, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004368888888888889, | |
| "loss": 2.266, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004364444444444445, | |
| "loss": 2.255, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.000436, | |
| "loss": 2.2256, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00043555555555555557, | |
| "loss": 2.2561, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004351111111111111, | |
| "loss": 2.2569, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004346666666666667, | |
| "loss": 2.1995, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004342222222222223, | |
| "loss": 2.193, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00043377777777777783, | |
| "loss": 2.1382, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004333333333333333, | |
| "loss": 2.2134, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004328888888888889, | |
| "loss": 2.173, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004324444444444445, | |
| "loss": 2.2619, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00043200000000000004, | |
| "loss": 2.1414, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004315555555555555, | |
| "loss": 2.1653, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004311111111111111, | |
| "loss": 2.2585, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004306666666666667, | |
| "loss": 2.1986, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00043022222222222225, | |
| "loss": 2.1942, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042977777777777784, | |
| "loss": 2.2071, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004293333333333333, | |
| "loss": 2.2311, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004288888888888889, | |
| "loss": 2.2495, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042844444444444446, | |
| "loss": 2.1843, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042800000000000005, | |
| "loss": 2.2238, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042755555555555564, | |
| "loss": 2.2338, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004271111111111111, | |
| "loss": 2.2446, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042666666666666667, | |
| "loss": 2.1949, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042622222222222226, | |
| "loss": 2.2357, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042577777777777785, | |
| "loss": 2.1674, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042533333333333334, | |
| "loss": 2.2584, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004248888888888889, | |
| "loss": 2.2447, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042444444444444447, | |
| "loss": 2.1795, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00042400000000000006, | |
| "loss": 2.2055, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004235555555555556, | |
| "loss": 2.2359, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004231111111111111, | |
| "loss": 2.1674, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0004226666666666667, | |
| "loss": 2.2057, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00042222222222222227, | |
| "loss": 2.2296, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004217777777777778, | |
| "loss": 2.2603, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004213333333333333, | |
| "loss": 2.3112, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004208888888888889, | |
| "loss": 2.317, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004204444444444445, | |
| "loss": 2.226, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00042, | |
| "loss": 2.2722, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004195555555555556, | |
| "loss": 2.2752, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004191111111111111, | |
| "loss": 2.2753, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004186666666666667, | |
| "loss": 2.2863, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004182222222222222, | |
| "loss": 2.2846, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004177777777777778, | |
| "loss": 2.2366, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004173333333333333, | |
| "loss": 2.2447, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004168888888888889, | |
| "loss": 2.3256, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00041644444444444444, | |
| "loss": 2.2203, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00041600000000000003, | |
| "loss": 2.1463, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004155555555555556, | |
| "loss": 2.2303, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004151111111111111, | |
| "loss": 2.1686, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00041466666666666664, | |
| "loss": 2.2296, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00041422222222222224, | |
| "loss": 2.2437, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00041377777777777783, | |
| "loss": 2.2731, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00041333333333333337, | |
| "loss": 2.249, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00041288888888888885, | |
| "loss": 2.2399, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00041244444444444445, | |
| "loss": 2.1833, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00041200000000000004, | |
| "loss": 2.18, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004115555555555556, | |
| "loss": 2.1966, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004111111111111111, | |
| "loss": 2.1802, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00041066666666666666, | |
| "loss": 2.2319, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00041022222222222225, | |
| "loss": 2.2461, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00040977777777777784, | |
| "loss": 2.2113, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004093333333333334, | |
| "loss": 2.2787, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00040888888888888887, | |
| "loss": 2.1956, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00040844444444444446, | |
| "loss": 2.2832, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00040800000000000005, | |
| "loss": 2.2853, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004075555555555556, | |
| "loss": 2.1744, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004071111111111111, | |
| "loss": 2.214, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00040666666666666667, | |
| "loss": 2.2348, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00040622222222222226, | |
| "loss": 2.1959, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004057777777777778, | |
| "loss": 2.2039, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004053333333333334, | |
| "loss": 2.1746, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0004048888888888889, | |
| "loss": 2.2353, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00040444444444444447, | |
| "loss": 2.238, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.000404, | |
| "loss": 2.254, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004035555555555556, | |
| "loss": 2.214, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004031111111111112, | |
| "loss": 2.2567, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004026666666666667, | |
| "loss": 2.1921, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004022222222222222, | |
| "loss": 2.2253, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004017777777777778, | |
| "loss": 2.2083, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004013333333333334, | |
| "loss": 2.2661, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004008888888888889, | |
| "loss": 2.2081, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004004444444444444, | |
| "loss": 2.105, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0004, | |
| "loss": 2.2621, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0003995555555555556, | |
| "loss": 2.2716, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00039911111111111115, | |
| "loss": 2.2039, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0003986666666666667, | |
| "loss": 2.2143, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00039822222222222223, | |
| "loss": 2.3052, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0003977777777777778, | |
| "loss": 2.2629, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00039733333333333336, | |
| "loss": 2.2976, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0003968888888888889, | |
| "loss": 2.2596, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0003964444444444445, | |
| "loss": 2.2815, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00039600000000000003, | |
| "loss": 2.2405, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00039555555555555557, | |
| "loss": 2.2578, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003951111111111111, | |
| "loss": 2.1971, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003946666666666667, | |
| "loss": 2.2151, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00039422222222222224, | |
| "loss": 2.2425, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003937777777777778, | |
| "loss": 2.2328, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003933333333333333, | |
| "loss": 2.2582, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003928888888888889, | |
| "loss": 2.1567, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003924444444444445, | |
| "loss": 2.2728, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.000392, | |
| "loss": 2.1876, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003915555555555556, | |
| "loss": 2.2007, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003911111111111111, | |
| "loss": 2.2632, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003906666666666667, | |
| "loss": 2.1838, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003902222222222222, | |
| "loss": 2.1546, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003897777777777778, | |
| "loss": 2.2672, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003893333333333334, | |
| "loss": 2.2595, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003888888888888889, | |
| "loss": 2.1636, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00038844444444444446, | |
| "loss": 2.2421, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.000388, | |
| "loss": 2.2063, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0003875555555555556, | |
| "loss": 2.3393, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00038711111111111113, | |
| "loss": 2.2483, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00038666666666666667, | |
| "loss": 2.2506, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003862222222222222, | |
| "loss": 2.2516, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003857777777777778, | |
| "loss": 2.2543, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003853333333333334, | |
| "loss": 2.2858, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003848888888888889, | |
| "loss": 2.1998, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00038444444444444447, | |
| "loss": 2.2668, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.000384, | |
| "loss": 2.2263, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003835555555555556, | |
| "loss": 2.2531, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003831111111111111, | |
| "loss": 2.2176, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003826666666666667, | |
| "loss": 2.2106, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003822222222222223, | |
| "loss": 2.2093, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003817777777777778, | |
| "loss": 2.2806, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00038133333333333335, | |
| "loss": 2.2486, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003808888888888889, | |
| "loss": 2.1462, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003804444444444445, | |
| "loss": 2.1784, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00038, | |
| "loss": 2.2601, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00037955555555555556, | |
| "loss": 2.1816, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00037911111111111115, | |
| "loss": 2.2372, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0003786666666666667, | |
| "loss": 2.1934, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00037822222222222223, | |
| "loss": 2.151, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00037777777777777777, | |
| "loss": 2.2499, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037733333333333336, | |
| "loss": 2.1494, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003768888888888889, | |
| "loss": 2.2536, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037644444444444444, | |
| "loss": 2.186, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.000376, | |
| "loss": 2.2463, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037555555555555557, | |
| "loss": 2.2183, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037511111111111116, | |
| "loss": 2.2226, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003746666666666667, | |
| "loss": 2.2353, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037422222222222224, | |
| "loss": 2.2144, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003737777777777778, | |
| "loss": 2.232, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003733333333333334, | |
| "loss": 2.2233, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003728888888888889, | |
| "loss": 2.1968, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037244444444444445, | |
| "loss": 2.2326, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037200000000000004, | |
| "loss": 2.1779, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003715555555555556, | |
| "loss": 2.1174, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003711111111111111, | |
| "loss": 2.2201, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037066666666666666, | |
| "loss": 2.1531, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00037022222222222225, | |
| "loss": 2.2818, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003697777777777778, | |
| "loss": 2.2397, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.00036933333333333333, | |
| "loss": 2.2505, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0003688888888888889, | |
| "loss": 2.3194, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036844444444444446, | |
| "loss": 2.2095, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036800000000000005, | |
| "loss": 2.2788, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036755555555555554, | |
| "loss": 2.1899, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036711111111111113, | |
| "loss": 2.1734, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036666666666666667, | |
| "loss": 2.2329, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036622222222222226, | |
| "loss": 2.2439, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036577777777777775, | |
| "loss": 2.1828, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036533333333333334, | |
| "loss": 2.2352, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036488888888888893, | |
| "loss": 2.2226, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036444444444444447, | |
| "loss": 2.1416, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.000364, | |
| "loss": 2.2727, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036355555555555555, | |
| "loss": 2.2861, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036311111111111114, | |
| "loss": 2.1931, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0003626666666666667, | |
| "loss": 2.2435, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0003622222222222222, | |
| "loss": 2.2815, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0003617777777777778, | |
| "loss": 2.257, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036133333333333335, | |
| "loss": 2.2182, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036088888888888895, | |
| "loss": 2.2526, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036044444444444443, | |
| "loss": 2.2212, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00036, | |
| "loss": 2.2718, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035955555555555556, | |
| "loss": 2.3178, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035911111111111115, | |
| "loss": 2.189, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035866666666666664, | |
| "loss": 2.2615, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035822222222222223, | |
| "loss": 2.1864, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003577777777777778, | |
| "loss": 2.1476, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035733333333333336, | |
| "loss": 2.2484, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003568888888888889, | |
| "loss": 2.2128, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035644444444444444, | |
| "loss": 2.1948, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035600000000000003, | |
| "loss": 2.292, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035555555555555557, | |
| "loss": 2.1969, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.126681089401245, | |
| "eval_rouge1": 13.5606, | |
| "eval_rouge2": 4.5005, | |
| "eval_rougeL": 8.5459, | |
| "eval_rougeLsum": 8.5241, | |
| "eval_runtime": 9294.4875, | |
| "eval_samples_per_second": 0.054, | |
| "eval_steps_per_second": 0.054, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003551111111111111, | |
| "loss": 2.2215, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003546666666666667, | |
| "loss": 2.189, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035422222222222224, | |
| "loss": 2.2094, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003537777777777778, | |
| "loss": 2.2869, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003533333333333333, | |
| "loss": 2.2525, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003528888888888889, | |
| "loss": 2.1887, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035244444444444445, | |
| "loss": 2.2738, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00035200000000000005, | |
| "loss": 2.2358, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003515555555555556, | |
| "loss": 2.2106, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0003511111111111111, | |
| "loss": 2.2539, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003506666666666667, | |
| "loss": 2.2395, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00035022222222222225, | |
| "loss": 2.146, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003497777777777778, | |
| "loss": 2.2555, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00034933333333333333, | |
| "loss": 2.2447, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003488888888888889, | |
| "loss": 2.3177, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00034844444444444446, | |
| "loss": 2.2334, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.000348, | |
| "loss": 2.1884, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003475555555555556, | |
| "loss": 2.2108, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00034711111111111113, | |
| "loss": 2.2498, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00034666666666666667, | |
| "loss": 2.242, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003462222222222222, | |
| "loss": 2.2292, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003457777777777778, | |
| "loss": 2.2706, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00034533333333333334, | |
| "loss": 2.2784, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003448888888888889, | |
| "loss": 2.1807, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0003444444444444445, | |
| "loss": 2.211, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.000344, | |
| "loss": 2.2172, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0003435555555555556, | |
| "loss": 2.1684, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0003431111111111111, | |
| "loss": 2.2307, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0003426666666666667, | |
| "loss": 2.2313, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0003422222222222222, | |
| "loss": 2.1823, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0003417777777777778, | |
| "loss": 2.1682, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00034133333333333335, | |
| "loss": 2.2691, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0003408888888888889, | |
| "loss": 2.1713, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0003404444444444445, | |
| "loss": 2.1428, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00034, | |
| "loss": 2.2978, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00033955555555555556, | |
| "loss": 2.2355, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003391111111111111, | |
| "loss": 2.2244, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003386666666666667, | |
| "loss": 2.2772, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00033822222222222223, | |
| "loss": 2.2602, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00033777777777777777, | |
| "loss": 2.2692, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00033733333333333337, | |
| "loss": 2.2571, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003368888888888889, | |
| "loss": 2.1604, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003364444444444445, | |
| "loss": 2.2203, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.000336, | |
| "loss": 2.233, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003355555555555556, | |
| "loss": 2.2651, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003351111111111111, | |
| "loss": 2.1877, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003346666666666667, | |
| "loss": 2.2141, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00033422222222222224, | |
| "loss": 2.2046, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003337777777777778, | |
| "loss": 2.188, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003333333333333334, | |
| "loss": 2.1651, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003328888888888889, | |
| "loss": 2.1521, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00033244444444444445, | |
| "loss": 2.1967, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.000332, | |
| "loss": 2.1108, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003315555555555556, | |
| "loss": 2.1332, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0003311111111111111, | |
| "loss": 2.2851, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00033066666666666666, | |
| "loss": 2.254, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00033022222222222226, | |
| "loss": 2.17, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0003297777777777778, | |
| "loss": 2.1728, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032933333333333333, | |
| "loss": 2.154, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032888888888888887, | |
| "loss": 2.2122, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032844444444444447, | |
| "loss": 2.2088, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.000328, | |
| "loss": 2.2363, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0003275555555555556, | |
| "loss": 2.1167, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032711111111111114, | |
| "loss": 2.0969, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0003266666666666667, | |
| "loss": 2.197, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032622222222222227, | |
| "loss": 2.1962, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0003257777777777778, | |
| "loss": 2.1862, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032533333333333334, | |
| "loss": 2.1808, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0003248888888888889, | |
| "loss": 2.2553, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0003244444444444445, | |
| "loss": 2.1425, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.000324, | |
| "loss": 2.1588, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032355555555555555, | |
| "loss": 2.1203, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032311111111111115, | |
| "loss": 2.188, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0003226666666666667, | |
| "loss": 2.1015, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0003222222222222222, | |
| "loss": 2.1881, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00032177777777777776, | |
| "loss": 2.2133, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00032133333333333336, | |
| "loss": 2.181, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0003208888888888889, | |
| "loss": 2.1, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00032044444444444443, | |
| "loss": 2.1577, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00032, | |
| "loss": 2.2327, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031955555555555556, | |
| "loss": 2.182, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031911111111111116, | |
| "loss": 2.1862, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031866666666666664, | |
| "loss": 2.1597, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031822222222222224, | |
| "loss": 2.2375, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0003177777777777778, | |
| "loss": 2.1721, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031733333333333337, | |
| "loss": 2.1676, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0003168888888888889, | |
| "loss": 2.1929, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031644444444444444, | |
| "loss": 2.1102, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031600000000000004, | |
| "loss": 2.199, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0003155555555555556, | |
| "loss": 2.2004, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0003151111111111111, | |
| "loss": 2.2512, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031466666666666665, | |
| "loss": 2.1994, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031422222222222225, | |
| "loss": 2.1508, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00031377777777777784, | |
| "loss": 2.1607, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0003133333333333333, | |
| "loss": 2.2171, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0003128888888888889, | |
| "loss": 2.2, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00031244444444444446, | |
| "loss": 2.1907, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00031200000000000005, | |
| "loss": 2.2026, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00031155555555555553, | |
| "loss": 2.176, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0003111111111111111, | |
| "loss": 2.2456, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00031066666666666666, | |
| "loss": 2.2479, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00031022222222222226, | |
| "loss": 2.1432, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0003097777777777778, | |
| "loss": 2.178, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00030933333333333334, | |
| "loss": 2.1422, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00030888888888888893, | |
| "loss": 2.1744, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00030844444444444447, | |
| "loss": 2.2191, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.000308, | |
| "loss": 2.2665, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00030755555555555554, | |
| "loss": 2.2301, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00030711111111111114, | |
| "loss": 2.2328, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0003066666666666667, | |
| "loss": 2.2027, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0003062222222222222, | |
| "loss": 2.1684, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0003057777777777778, | |
| "loss": 2.2214, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00030533333333333335, | |
| "loss": 2.1323, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00030488888888888894, | |
| "loss": 2.1567, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0003044444444444444, | |
| "loss": 2.1194, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.000304, | |
| "loss": 2.1019, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00030355555555555556, | |
| "loss": 2.15, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00030311111111111115, | |
| "loss": 2.1954, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0003026666666666667, | |
| "loss": 2.1297, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0003022222222222222, | |
| "loss": 2.1343, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0003017777777777778, | |
| "loss": 2.1918, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00030133333333333336, | |
| "loss": 2.2325, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0003008888888888889, | |
| "loss": 2.2012, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00030044444444444443, | |
| "loss": 2.1262, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00030000000000000003, | |
| "loss": 2.1162, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00029955555555555557, | |
| "loss": 2.1746, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002991111111111111, | |
| "loss": 2.1809, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002986666666666667, | |
| "loss": 2.268, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00029822222222222224, | |
| "loss": 2.2361, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002977777777777778, | |
| "loss": 2.1774, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002973333333333333, | |
| "loss": 2.1574, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002968888888888889, | |
| "loss": 2.1475, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002964444444444445, | |
| "loss": 2.1711, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.000296, | |
| "loss": 2.2065, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002955555555555556, | |
| "loss": 2.1746, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002951111111111111, | |
| "loss": 2.2004, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002946666666666667, | |
| "loss": 2.0863, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00029422222222222225, | |
| "loss": 2.2187, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002937777777777778, | |
| "loss": 2.1756, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002933333333333333, | |
| "loss": 2.1669, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002928888888888889, | |
| "loss": 2.0371, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00029244444444444446, | |
| "loss": 2.2079, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.000292, | |
| "loss": 2.138, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002915555555555556, | |
| "loss": 2.1843, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00029111111111111113, | |
| "loss": 2.0759, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00029066666666666667, | |
| "loss": 2.1002, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002902222222222222, | |
| "loss": 2.1626, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002897777777777778, | |
| "loss": 2.2071, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002893333333333334, | |
| "loss": 2.1772, | |
| "step": 1349 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002888888888888889, | |
| "loss": 2.1284, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00028844444444444447, | |
| "loss": 2.1774, | |
| "step": 1351 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.000288, | |
| "loss": 2.1344, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002875555555555556, | |
| "loss": 2.1619, | |
| "step": 1353 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002871111111111111, | |
| "loss": 2.1673, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002866666666666667, | |
| "loss": 2.1395, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002862222222222222, | |
| "loss": 2.1698, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002857777777777778, | |
| "loss": 2.1331, | |
| "step": 1357 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00028533333333333335, | |
| "loss": 2.0731, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002848888888888889, | |
| "loss": 2.1612, | |
| "step": 1359 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002844444444444445, | |
| "loss": 2.1686, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.000284, | |
| "loss": 2.1915, | |
| "step": 1361 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00028355555555555556, | |
| "loss": 2.1867, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002831111111111111, | |
| "loss": 2.1668, | |
| "step": 1363 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002826666666666667, | |
| "loss": 2.2493, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00028222222222222223, | |
| "loss": 2.0671, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00028177777777777777, | |
| "loss": 2.0957, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00028133333333333336, | |
| "loss": 2.0841, | |
| "step": 1367 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002808888888888889, | |
| "loss": 2.1165, | |
| "step": 1368 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002804444444444445, | |
| "loss": 2.1172, | |
| "step": 1369 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00028, | |
| "loss": 2.1452, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00027955555555555557, | |
| "loss": 2.2362, | |
| "step": 1371 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00027911111111111116, | |
| "loss": 2.1393, | |
| "step": 1372 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002786666666666667, | |
| "loss": 2.1432, | |
| "step": 1373 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00027822222222222224, | |
| "loss": 2.188, | |
| "step": 1374 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0002777777777777778, | |
| "loss": 2.0608, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00027733333333333337, | |
| "loss": 2.176, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002768888888888889, | |
| "loss": 2.1274, | |
| "step": 1377 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027644444444444445, | |
| "loss": 2.1743, | |
| "step": 1378 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.000276, | |
| "loss": 2.0862, | |
| "step": 1379 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002755555555555556, | |
| "loss": 2.1069, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002751111111111111, | |
| "loss": 2.0847, | |
| "step": 1381 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027466666666666666, | |
| "loss": 2.1519, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027422222222222225, | |
| "loss": 2.1348, | |
| "step": 1383 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002737777777777778, | |
| "loss": 2.1674, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027333333333333333, | |
| "loss": 2.1839, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027288888888888887, | |
| "loss": 2.1126, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027244444444444446, | |
| "loss": 2.1085, | |
| "step": 1387 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027200000000000005, | |
| "loss": 2.1103, | |
| "step": 1388 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027155555555555554, | |
| "loss": 2.0676, | |
| "step": 1389 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027111111111111113, | |
| "loss": 2.1052, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027066666666666667, | |
| "loss": 2.124, | |
| "step": 1391 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00027022222222222226, | |
| "loss": 2.1178, | |
| "step": 1392 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002697777777777778, | |
| "loss": 2.0989, | |
| "step": 1393 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00026933333333333334, | |
| "loss": 2.0984, | |
| "step": 1394 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00026888888888888893, | |
| "loss": 2.1175, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00026844444444444447, | |
| "loss": 2.1575, | |
| "step": 1396 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.000268, | |
| "loss": 2.1486, | |
| "step": 1397 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026755555555555555, | |
| "loss": 2.0775, | |
| "step": 1398 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026711111111111114, | |
| "loss": 2.2145, | |
| "step": 1399 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0002666666666666667, | |
| "loss": 2.1359, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.1156013011932373, | |
| "eval_rouge1": 13.7423, | |
| "eval_rouge2": 4.7061, | |
| "eval_rougeL": 8.6172, | |
| "eval_rougeLsum": 8.6284, | |
| "eval_runtime": 9295.5684, | |
| "eval_samples_per_second": 0.054, | |
| "eval_steps_per_second": 0.054, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0002662222222222222, | |
| "loss": 2.0898, | |
| "step": 1401 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026577777777777776, | |
| "loss": 2.1313, | |
| "step": 1402 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026533333333333335, | |
| "loss": 2.1198, | |
| "step": 1403 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026488888888888894, | |
| "loss": 2.1396, | |
| "step": 1404 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026444444444444443, | |
| "loss": 2.1515, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.000264, | |
| "loss": 2.1336, | |
| "step": 1406 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026355555555555556, | |
| "loss": 2.1756, | |
| "step": 1407 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026311111111111115, | |
| "loss": 2.1563, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026266666666666664, | |
| "loss": 2.1252, | |
| "step": 1409 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026222222222222223, | |
| "loss": 2.118, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0002617777777777778, | |
| "loss": 2.142, | |
| "step": 1411 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026133333333333336, | |
| "loss": 2.2093, | |
| "step": 1412 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0002608888888888889, | |
| "loss": 2.1199, | |
| "step": 1413 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026044444444444444, | |
| "loss": 2.0734, | |
| "step": 1414 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00026000000000000003, | |
| "loss": 2.0976, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.00025955555555555557, | |
| "loss": 2.0571, | |
| "step": 1416 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002591111111111111, | |
| "loss": 2.1415, | |
| "step": 1417 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025866666666666665, | |
| "loss": 2.0729, | |
| "step": 1418 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025822222222222224, | |
| "loss": 2.1535, | |
| "step": 1419 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025777777777777783, | |
| "loss": 2.1454, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002573333333333333, | |
| "loss": 2.1955, | |
| "step": 1421 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002568888888888889, | |
| "loss": 2.1006, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025644444444444445, | |
| "loss": 2.0917, | |
| "step": 1423 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025600000000000004, | |
| "loss": 2.0921, | |
| "step": 1424 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025555555555555553, | |
| "loss": 2.1228, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002551111111111111, | |
| "loss": 2.1433, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002546666666666667, | |
| "loss": 2.1351, | |
| "step": 1427 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025422222222222225, | |
| "loss": 2.1113, | |
| "step": 1428 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002537777777777778, | |
| "loss": 2.0421, | |
| "step": 1429 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025333333333333333, | |
| "loss": 2.0756, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002528888888888889, | |
| "loss": 2.1356, | |
| "step": 1431 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025244444444444446, | |
| "loss": 2.1675, | |
| "step": 1432 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.000252, | |
| "loss": 2.1583, | |
| "step": 1433 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0002515555555555556, | |
| "loss": 2.1534, | |
| "step": 1434 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025111111111111113, | |
| "loss": 2.1404, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00025066666666666667, | |
| "loss": 2.1038, | |
| "step": 1436 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002502222222222222, | |
| "loss": 2.1705, | |
| "step": 1437 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002497777777777778, | |
| "loss": 2.1683, | |
| "step": 1438 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00024933333333333334, | |
| "loss": 2.1288, | |
| "step": 1439 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002488888888888889, | |
| "loss": 2.1502, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002484444444444444, | |
| "loss": 2.1077, | |
| "step": 1441 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.000248, | |
| "loss": 2.1261, | |
| "step": 1442 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002475555555555556, | |
| "loss": 2.1351, | |
| "step": 1443 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00024711111111111114, | |
| "loss": 2.1047, | |
| "step": 1444 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002466666666666667, | |
| "loss": 2.1223, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002462222222222222, | |
| "loss": 2.0726, | |
| "step": 1446 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002457777777777778, | |
| "loss": 2.1112, | |
| "step": 1447 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00024533333333333335, | |
| "loss": 2.1283, | |
| "step": 1448 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002448888888888889, | |
| "loss": 2.1205, | |
| "step": 1449 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002444444444444445, | |
| "loss": 2.1715, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.000244, | |
| "loss": 2.1517, | |
| "step": 1451 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002435555555555556, | |
| "loss": 2.1007, | |
| "step": 1452 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002431111111111111, | |
| "loss": 2.1804, | |
| "step": 1453 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002426666666666667, | |
| "loss": 2.1121, | |
| "step": 1454 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002422222222222222, | |
| "loss": 2.1229, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002417777777777778, | |
| "loss": 2.0234, | |
| "step": 1456 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00024133333333333336, | |
| "loss": 2.0615, | |
| "step": 1457 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002408888888888889, | |
| "loss": 2.1373, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00024044444444444447, | |
| "loss": 2.1106, | |
| "step": 1459 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00024, | |
| "loss": 2.1396, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023955555555555557, | |
| "loss": 2.1163, | |
| "step": 1461 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002391111111111111, | |
| "loss": 2.0869, | |
| "step": 1462 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023866666666666668, | |
| "loss": 2.1535, | |
| "step": 1463 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023822222222222222, | |
| "loss": 2.0669, | |
| "step": 1464 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023777777777777778, | |
| "loss": 2.1315, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023733333333333337, | |
| "loss": 2.1385, | |
| "step": 1466 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023688888888888889, | |
| "loss": 2.1548, | |
| "step": 1467 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023644444444444448, | |
| "loss": 2.1021, | |
| "step": 1468 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.000236, | |
| "loss": 2.1829, | |
| "step": 1469 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023555555555555558, | |
| "loss": 2.1002, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002351111111111111, | |
| "loss": 2.162, | |
| "step": 1471 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002346666666666667, | |
| "loss": 2.099, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023422222222222225, | |
| "loss": 2.1477, | |
| "step": 1473 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002337777777777778, | |
| "loss": 2.1612, | |
| "step": 1474 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00023333333333333336, | |
| "loss": 2.0684, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002328888888888889, | |
| "loss": 2.1182, | |
| "step": 1476 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00023244444444444446, | |
| "loss": 2.1298, | |
| "step": 1477 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.000232, | |
| "loss": 2.0761, | |
| "step": 1478 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00023155555555555557, | |
| "loss": 2.1133, | |
| "step": 1479 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0002311111111111111, | |
| "loss": 2.1749, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00023066666666666667, | |
| "loss": 2.1653, | |
| "step": 1481 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00023022222222222224, | |
| "loss": 2.1304, | |
| "step": 1482 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022977777777777778, | |
| "loss": 2.0739, | |
| "step": 1483 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022933333333333334, | |
| "loss": 2.0529, | |
| "step": 1484 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022888888888888888, | |
| "loss": 2.1136, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022844444444444447, | |
| "loss": 2.11, | |
| "step": 1486 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022799999999999999, | |
| "loss": 2.1093, | |
| "step": 1487 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022755555555555558, | |
| "loss": 2.1102, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022711111111111114, | |
| "loss": 2.1033, | |
| "step": 1489 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022666666666666668, | |
| "loss": 2.1908, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022622222222222225, | |
| "loss": 2.0719, | |
| "step": 1491 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0002257777777777778, | |
| "loss": 2.0688, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022533333333333335, | |
| "loss": 2.1859, | |
| "step": 1493 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0002248888888888889, | |
| "loss": 2.1219, | |
| "step": 1494 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022444444444444446, | |
| "loss": 2.0813, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00022400000000000002, | |
| "loss": 2.133, | |
| "step": 1496 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022355555555555556, | |
| "loss": 2.2231, | |
| "step": 1497 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022311111111111113, | |
| "loss": 2.1299, | |
| "step": 1498 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022266666666666667, | |
| "loss": 2.1343, | |
| "step": 1499 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022222222222222223, | |
| "loss": 2.1702, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022177777777777777, | |
| "loss": 2.1268, | |
| "step": 1501 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022133333333333334, | |
| "loss": 2.1472, | |
| "step": 1502 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022088888888888888, | |
| "loss": 2.1203, | |
| "step": 1503 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022044444444444444, | |
| "loss": 2.1968, | |
| "step": 1504 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00022000000000000003, | |
| "loss": 2.0619, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021955555555555555, | |
| "loss": 2.1274, | |
| "step": 1506 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021911111111111114, | |
| "loss": 2.0527, | |
| "step": 1507 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021866666666666665, | |
| "loss": 2.0104, | |
| "step": 1508 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021822222222222224, | |
| "loss": 2.1593, | |
| "step": 1509 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021777777777777778, | |
| "loss": 2.0624, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021733333333333335, | |
| "loss": 2.1088, | |
| "step": 1511 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021688888888888891, | |
| "loss": 2.0986, | |
| "step": 1512 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021644444444444445, | |
| "loss": 2.0843, | |
| "step": 1513 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021600000000000002, | |
| "loss": 2.1256, | |
| "step": 1514 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021555555555555556, | |
| "loss": 2.1612, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00021511111111111112, | |
| "loss": 2.1048, | |
| "step": 1516 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021466666666666666, | |
| "loss": 2.1749, | |
| "step": 1517 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021422222222222223, | |
| "loss": 2.13, | |
| "step": 1518 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021377777777777782, | |
| "loss": 2.1937, | |
| "step": 1519 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021333333333333333, | |
| "loss": 2.1608, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021288888888888893, | |
| "loss": 2.1284, | |
| "step": 1521 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021244444444444444, | |
| "loss": 2.1542, | |
| "step": 1522 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021200000000000003, | |
| "loss": 2.1237, | |
| "step": 1523 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021155555555555554, | |
| "loss": 2.1512, | |
| "step": 1524 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021111111111111113, | |
| "loss": 2.0976, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021066666666666665, | |
| "loss": 2.135, | |
| "step": 1526 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00021022222222222224, | |
| "loss": 2.1227, | |
| "step": 1527 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0002097777777777778, | |
| "loss": 2.114, | |
| "step": 1528 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020933333333333334, | |
| "loss": 2.141, | |
| "step": 1529 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0002088888888888889, | |
| "loss": 2.1855, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020844444444444445, | |
| "loss": 2.1343, | |
| "step": 1531 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020800000000000001, | |
| "loss": 2.0732, | |
| "step": 1532 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020755555555555555, | |
| "loss": 2.1155, | |
| "step": 1533 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020711111111111112, | |
| "loss": 2.1146, | |
| "step": 1534 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020666666666666668, | |
| "loss": 2.1464, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020622222222222222, | |
| "loss": 2.1327, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002057777777777778, | |
| "loss": 2.1498, | |
| "step": 1537 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020533333333333333, | |
| "loss": 2.133, | |
| "step": 1538 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020488888888888892, | |
| "loss": 2.1969, | |
| "step": 1539 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020444444444444443, | |
| "loss": 2.1602, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020400000000000003, | |
| "loss": 2.1356, | |
| "step": 1541 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020355555555555554, | |
| "loss": 2.1506, | |
| "step": 1542 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020311111111111113, | |
| "loss": 2.0696, | |
| "step": 1543 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002026666666666667, | |
| "loss": 2.1018, | |
| "step": 1544 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020222222222222223, | |
| "loss": 2.1302, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002017777777777778, | |
| "loss": 2.1233, | |
| "step": 1546 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020133333333333334, | |
| "loss": 2.1112, | |
| "step": 1547 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002008888888888889, | |
| "loss": 2.219, | |
| "step": 1548 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020044444444444444, | |
| "loss": 2.1709, | |
| "step": 1549 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0002, | |
| "loss": 2.077, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00019955555555555558, | |
| "loss": 2.167, | |
| "step": 1551 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00019911111111111111, | |
| "loss": 2.0899, | |
| "step": 1552 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00019866666666666668, | |
| "loss": 2.1202, | |
| "step": 1553 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00019822222222222225, | |
| "loss": 2.1057, | |
| "step": 1554 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00019777777777777778, | |
| "loss": 2.0946, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00019733333333333335, | |
| "loss": 2.1357, | |
| "step": 1556 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0001968888888888889, | |
| "loss": 2.1182, | |
| "step": 1557 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019644444444444445, | |
| "loss": 2.112, | |
| "step": 1558 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.000196, | |
| "loss": 2.1319, | |
| "step": 1559 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019555555555555556, | |
| "loss": 2.1783, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0001951111111111111, | |
| "loss": 2.1374, | |
| "step": 1561 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0001946666666666667, | |
| "loss": 2.1862, | |
| "step": 1562 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019422222222222223, | |
| "loss": 2.1993, | |
| "step": 1563 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0001937777777777778, | |
| "loss": 2.0878, | |
| "step": 1564 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019333333333333333, | |
| "loss": 2.1082, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0001928888888888889, | |
| "loss": 2.1007, | |
| "step": 1566 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019244444444444444, | |
| "loss": 2.1888, | |
| "step": 1567 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.000192, | |
| "loss": 2.1479, | |
| "step": 1568 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019155555555555554, | |
| "loss": 2.235, | |
| "step": 1569 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019111111111111114, | |
| "loss": 2.1963, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019066666666666668, | |
| "loss": 2.0638, | |
| "step": 1571 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019022222222222224, | |
| "loss": 2.1389, | |
| "step": 1572 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00018977777777777778, | |
| "loss": 2.0982, | |
| "step": 1573 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00018933333333333335, | |
| "loss": 2.1129, | |
| "step": 1574 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 2.1199, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00018844444444444445, | |
| "loss": 2.0823, | |
| "step": 1576 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.000188, | |
| "loss": 2.1113, | |
| "step": 1577 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018755555555555558, | |
| "loss": 2.1034, | |
| "step": 1578 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018711111111111112, | |
| "loss": 2.0663, | |
| "step": 1579 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0001866666666666667, | |
| "loss": 2.0847, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018622222222222223, | |
| "loss": 2.155, | |
| "step": 1581 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0001857777777777778, | |
| "loss": 2.1369, | |
| "step": 1582 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018533333333333333, | |
| "loss": 2.0877, | |
| "step": 1583 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0001848888888888889, | |
| "loss": 2.0629, | |
| "step": 1584 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018444444444444446, | |
| "loss": 2.1677, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018400000000000003, | |
| "loss": 2.0936, | |
| "step": 1586 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018355555555555557, | |
| "loss": 2.0795, | |
| "step": 1587 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018311111111111113, | |
| "loss": 2.0966, | |
| "step": 1588 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018266666666666667, | |
| "loss": 2.1501, | |
| "step": 1589 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018222222222222224, | |
| "loss": 2.1039, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018177777777777778, | |
| "loss": 2.1578, | |
| "step": 1591 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018133333333333334, | |
| "loss": 2.0771, | |
| "step": 1592 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0001808888888888889, | |
| "loss": 2.126, | |
| "step": 1593 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018044444444444447, | |
| "loss": 2.0727, | |
| "step": 1594 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00018, | |
| "loss": 2.1178, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00017955555555555558, | |
| "loss": 2.1327, | |
| "step": 1596 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017911111111111112, | |
| "loss": 2.1465, | |
| "step": 1597 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017866666666666668, | |
| "loss": 2.1755, | |
| "step": 1598 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017822222222222222, | |
| "loss": 2.1052, | |
| "step": 1599 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017777777777777779, | |
| "loss": 2.1484, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.1070616245269775, | |
| "eval_rouge1": 14.2289, | |
| "eval_rouge2": 4.7872, | |
| "eval_rougeL": 8.9958, | |
| "eval_rougeLsum": 9.0075, | |
| "eval_runtime": 9037.4832, | |
| "eval_samples_per_second": 0.055, | |
| "eval_steps_per_second": 0.055, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017733333333333335, | |
| "loss": 2.1553, | |
| "step": 1601 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0001768888888888889, | |
| "loss": 2.1205, | |
| "step": 1602 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017644444444444446, | |
| "loss": 2.0923, | |
| "step": 1603 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017600000000000002, | |
| "loss": 2.1086, | |
| "step": 1604 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017555555555555556, | |
| "loss": 2.1597, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017511111111111113, | |
| "loss": 2.1147, | |
| "step": 1606 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017466666666666667, | |
| "loss": 2.1185, | |
| "step": 1607 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017422222222222223, | |
| "loss": 2.1641, | |
| "step": 1608 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0001737777777777778, | |
| "loss": 2.1046, | |
| "step": 1609 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017333333333333334, | |
| "loss": 2.136, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0001728888888888889, | |
| "loss": 2.1428, | |
| "step": 1611 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017244444444444444, | |
| "loss": 2.1183, | |
| "step": 1612 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.000172, | |
| "loss": 2.1328, | |
| "step": 1613 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017155555555555555, | |
| "loss": 2.1679, | |
| "step": 1614 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0001711111111111111, | |
| "loss": 2.0474, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00017066666666666668, | |
| "loss": 2.1221, | |
| "step": 1616 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00017022222222222224, | |
| "loss": 2.1004, | |
| "step": 1617 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016977777777777778, | |
| "loss": 2.1145, | |
| "step": 1618 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016933333333333335, | |
| "loss": 2.0614, | |
| "step": 1619 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016888888888888889, | |
| "loss": 2.0831, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016844444444444445, | |
| "loss": 2.1027, | |
| "step": 1621 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.000168, | |
| "loss": 2.0872, | |
| "step": 1622 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016755555555555556, | |
| "loss": 2.1903, | |
| "step": 1623 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016711111111111112, | |
| "loss": 2.0553, | |
| "step": 1624 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.0001666666666666667, | |
| "loss": 2.0946, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016622222222222223, | |
| "loss": 2.0989, | |
| "step": 1626 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.0001657777777777778, | |
| "loss": 2.0825, | |
| "step": 1627 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016533333333333333, | |
| "loss": 2.1103, | |
| "step": 1628 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.0001648888888888889, | |
| "loss": 2.0972, | |
| "step": 1629 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016444444444444444, | |
| "loss": 2.1631, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.000164, | |
| "loss": 2.0751, | |
| "step": 1631 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016355555555555557, | |
| "loss": 2.1255, | |
| "step": 1632 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016311111111111113, | |
| "loss": 2.159, | |
| "step": 1633 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016266666666666667, | |
| "loss": 2.1844, | |
| "step": 1634 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016222222222222224, | |
| "loss": 2.1239, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00016177777777777778, | |
| "loss": 2.1313, | |
| "step": 1636 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00016133333333333334, | |
| "loss": 2.075, | |
| "step": 1637 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00016088888888888888, | |
| "loss": 2.1068, | |
| "step": 1638 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00016044444444444445, | |
| "loss": 2.1633, | |
| "step": 1639 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00016, | |
| "loss": 2.073, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015955555555555558, | |
| "loss": 2.0468, | |
| "step": 1641 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015911111111111112, | |
| "loss": 2.0787, | |
| "step": 1642 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015866666666666668, | |
| "loss": 2.123, | |
| "step": 1643 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015822222222222222, | |
| "loss": 2.1059, | |
| "step": 1644 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0001577777777777778, | |
| "loss": 2.0936, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015733333333333333, | |
| "loss": 2.0886, | |
| "step": 1646 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015688888888888892, | |
| "loss": 2.1275, | |
| "step": 1647 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015644444444444446, | |
| "loss": 2.1048, | |
| "step": 1648 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015600000000000002, | |
| "loss": 2.1402, | |
| "step": 1649 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015555555555555556, | |
| "loss": 2.0868, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015511111111111113, | |
| "loss": 2.154, | |
| "step": 1651 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015466666666666667, | |
| "loss": 2.172, | |
| "step": 1652 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015422222222222223, | |
| "loss": 2.1129, | |
| "step": 1653 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015377777777777777, | |
| "loss": 2.1185, | |
| "step": 1654 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00015333333333333334, | |
| "loss": 2.0523, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0001528888888888889, | |
| "loss": 2.0779, | |
| "step": 1656 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00015244444444444447, | |
| "loss": 2.149, | |
| "step": 1657 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.000152, | |
| "loss": 2.0786, | |
| "step": 1658 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00015155555555555557, | |
| "loss": 2.1615, | |
| "step": 1659 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001511111111111111, | |
| "loss": 2.0473, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00015066666666666668, | |
| "loss": 2.1122, | |
| "step": 1661 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00015022222222222222, | |
| "loss": 2.1615, | |
| "step": 1662 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014977777777777778, | |
| "loss": 2.1331, | |
| "step": 1663 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014933333333333335, | |
| "loss": 2.1782, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001488888888888889, | |
| "loss": 2.126, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014844444444444445, | |
| "loss": 2.1666, | |
| "step": 1666 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.000148, | |
| "loss": 2.0928, | |
| "step": 1667 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014755555555555556, | |
| "loss": 2.111, | |
| "step": 1668 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014711111111111112, | |
| "loss": 2.1106, | |
| "step": 1669 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014666666666666666, | |
| "loss": 2.1035, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014622222222222223, | |
| "loss": 2.0571, | |
| "step": 1671 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001457777777777778, | |
| "loss": 2.1375, | |
| "step": 1672 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014533333333333333, | |
| "loss": 2.1541, | |
| "step": 1673 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001448888888888889, | |
| "loss": 2.1483, | |
| "step": 1674 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00014444444444444444, | |
| "loss": 2.1686, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.000144, | |
| "loss": 2.1511, | |
| "step": 1676 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014355555555555554, | |
| "loss": 2.1709, | |
| "step": 1677 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0001431111111111111, | |
| "loss": 2.1034, | |
| "step": 1678 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014266666666666667, | |
| "loss": 2.1392, | |
| "step": 1679 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014222222222222224, | |
| "loss": 2.0913, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014177777777777778, | |
| "loss": 2.1404, | |
| "step": 1681 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014133333333333334, | |
| "loss": 2.0701, | |
| "step": 1682 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014088888888888888, | |
| "loss": 2.0879, | |
| "step": 1683 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014044444444444445, | |
| "loss": 2.1121, | |
| "step": 1684 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00014, | |
| "loss": 2.1578, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013955555555555558, | |
| "loss": 2.1933, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013911111111111112, | |
| "loss": 2.1029, | |
| "step": 1687 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013866666666666669, | |
| "loss": 2.1216, | |
| "step": 1688 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013822222222222222, | |
| "loss": 2.1357, | |
| "step": 1689 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0001377777777777778, | |
| "loss": 2.1176, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013733333333333333, | |
| "loss": 2.1144, | |
| "step": 1691 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.0001368888888888889, | |
| "loss": 2.1074, | |
| "step": 1692 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013644444444444443, | |
| "loss": 2.0608, | |
| "step": 1693 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013600000000000003, | |
| "loss": 2.0729, | |
| "step": 1694 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013555555555555556, | |
| "loss": 2.1246, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00013511111111111113, | |
| "loss": 2.0802, | |
| "step": 1696 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013466666666666667, | |
| "loss": 2.1442, | |
| "step": 1697 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013422222222222224, | |
| "loss": 2.1373, | |
| "step": 1698 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013377777777777777, | |
| "loss": 2.141, | |
| "step": 1699 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013333333333333334, | |
| "loss": 2.0219, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013288888888888888, | |
| "loss": 2.0535, | |
| "step": 1701 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013244444444444447, | |
| "loss": 2.1072, | |
| "step": 1702 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.000132, | |
| "loss": 2.174, | |
| "step": 1703 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013155555555555558, | |
| "loss": 2.0816, | |
| "step": 1704 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013111111111111111, | |
| "loss": 2.0743, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013066666666666668, | |
| "loss": 2.1471, | |
| "step": 1706 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00013022222222222222, | |
| "loss": 2.1651, | |
| "step": 1707 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012977777777777779, | |
| "loss": 2.1085, | |
| "step": 1708 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012933333333333332, | |
| "loss": 2.1157, | |
| "step": 1709 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012888888888888892, | |
| "loss": 2.0453, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012844444444444446, | |
| "loss": 2.0658, | |
| "step": 1711 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012800000000000002, | |
| "loss": 2.1744, | |
| "step": 1712 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012755555555555556, | |
| "loss": 2.1362, | |
| "step": 1713 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012711111111111113, | |
| "loss": 2.1377, | |
| "step": 1714 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012666666666666666, | |
| "loss": 2.1357, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00012622222222222223, | |
| "loss": 2.1446, | |
| "step": 1716 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001257777777777778, | |
| "loss": 2.1889, | |
| "step": 1717 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012533333333333334, | |
| "loss": 2.1327, | |
| "step": 1718 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001248888888888889, | |
| "loss": 2.1451, | |
| "step": 1719 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012444444444444444, | |
| "loss": 2.1374, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.000124, | |
| "loss": 2.1527, | |
| "step": 1721 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012355555555555557, | |
| "loss": 2.1332, | |
| "step": 1722 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001231111111111111, | |
| "loss": 2.1022, | |
| "step": 1723 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012266666666666668, | |
| "loss": 2.1166, | |
| "step": 1724 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012222222222222224, | |
| "loss": 2.1562, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001217777777777778, | |
| "loss": 2.1021, | |
| "step": 1726 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012133333333333335, | |
| "loss": 2.1367, | |
| "step": 1727 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001208888888888889, | |
| "loss": 2.1476, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012044444444444445, | |
| "loss": 2.1072, | |
| "step": 1729 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00012, | |
| "loss": 2.1132, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00011955555555555556, | |
| "loss": 2.0795, | |
| "step": 1731 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00011911111111111111, | |
| "loss": 2.0882, | |
| "step": 1732 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00011866666666666669, | |
| "loss": 2.1346, | |
| "step": 1733 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00011822222222222224, | |
| "loss": 2.0931, | |
| "step": 1734 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00011777777777777779, | |
| "loss": 2.0273, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00011733333333333334, | |
| "loss": 2.0873, | |
| "step": 1736 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0001168888888888889, | |
| "loss": 2.0911, | |
| "step": 1737 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011644444444444445, | |
| "loss": 2.1585, | |
| "step": 1738 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.000116, | |
| "loss": 2.0838, | |
| "step": 1739 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011555555555555555, | |
| "loss": 2.0459, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011511111111111112, | |
| "loss": 2.1807, | |
| "step": 1741 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011466666666666667, | |
| "loss": 2.0983, | |
| "step": 1742 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011422222222222224, | |
| "loss": 2.0333, | |
| "step": 1743 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011377777777777779, | |
| "loss": 2.0963, | |
| "step": 1744 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011333333333333334, | |
| "loss": 2.0992, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0001128888888888889, | |
| "loss": 2.0909, | |
| "step": 1746 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011244444444444445, | |
| "loss": 2.1078, | |
| "step": 1747 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011200000000000001, | |
| "loss": 2.0909, | |
| "step": 1748 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011155555555555556, | |
| "loss": 2.1248, | |
| "step": 1749 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011111111111111112, | |
| "loss": 1.9917, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011066666666666667, | |
| "loss": 2.046, | |
| "step": 1751 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00011022222222222222, | |
| "loss": 2.0792, | |
| "step": 1752 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00010977777777777777, | |
| "loss": 2.2136, | |
| "step": 1753 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00010933333333333333, | |
| "loss": 2.1238, | |
| "step": 1754 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00010888888888888889, | |
| "loss": 2.1477, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00010844444444444446, | |
| "loss": 2.0884, | |
| "step": 1756 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010800000000000001, | |
| "loss": 2.1012, | |
| "step": 1757 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010755555555555556, | |
| "loss": 2.1251, | |
| "step": 1758 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010711111111111111, | |
| "loss": 2.1066, | |
| "step": 1759 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010666666666666667, | |
| "loss": 2.2252, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010622222222222222, | |
| "loss": 2.1278, | |
| "step": 1761 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010577777777777777, | |
| "loss": 2.1932, | |
| "step": 1762 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010533333333333332, | |
| "loss": 2.1579, | |
| "step": 1763 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0001048888888888889, | |
| "loss": 2.1337, | |
| "step": 1764 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010444444444444445, | |
| "loss": 2.1615, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010400000000000001, | |
| "loss": 2.1607, | |
| "step": 1766 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010355555555555556, | |
| "loss": 2.1803, | |
| "step": 1767 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010311111111111111, | |
| "loss": 2.0833, | |
| "step": 1768 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010266666666666666, | |
| "loss": 2.1253, | |
| "step": 1769 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010222222222222222, | |
| "loss": 2.1848, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010177777777777777, | |
| "loss": 2.0917, | |
| "step": 1771 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010133333333333335, | |
| "loss": 2.1113, | |
| "step": 1772 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0001008888888888889, | |
| "loss": 2.1489, | |
| "step": 1773 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00010044444444444445, | |
| "loss": 2.182, | |
| "step": 1774 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0001, | |
| "loss": 2.103, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.955555555555556e-05, | |
| "loss": 2.1152, | |
| "step": 1776 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.911111111111112e-05, | |
| "loss": 2.0967, | |
| "step": 1777 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.866666666666668e-05, | |
| "loss": 2.097, | |
| "step": 1778 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.822222222222223e-05, | |
| "loss": 2.1097, | |
| "step": 1779 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.777777777777778e-05, | |
| "loss": 2.1377, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.733333333333335e-05, | |
| "loss": 2.0993, | |
| "step": 1781 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.68888888888889e-05, | |
| "loss": 2.1712, | |
| "step": 1782 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.644444444444445e-05, | |
| "loss": 2.1424, | |
| "step": 1783 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.6e-05, | |
| "loss": 2.1878, | |
| "step": 1784 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.555555555555557e-05, | |
| "loss": 2.1306, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.511111111111112e-05, | |
| "loss": 2.0727, | |
| "step": 1786 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.466666666666667e-05, | |
| "loss": 2.1407, | |
| "step": 1787 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.422222222222223e-05, | |
| "loss": 2.0953, | |
| "step": 1788 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.377777777777779e-05, | |
| "loss": 2.1386, | |
| "step": 1789 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.333333333333334e-05, | |
| "loss": 2.0762, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.28888888888889e-05, | |
| "loss": 2.1777, | |
| "step": 1791 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.244444444444445e-05, | |
| "loss": 2.1934, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 2.0668, | |
| "step": 1793 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.155555555555557e-05, | |
| "loss": 2.1243, | |
| "step": 1794 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.111111111111112e-05, | |
| "loss": 2.0199, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.066666666666667e-05, | |
| "loss": 2.0717, | |
| "step": 1796 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 9.022222222222224e-05, | |
| "loss": 2.1475, | |
| "step": 1797 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.977777777777779e-05, | |
| "loss": 2.0687, | |
| "step": 1798 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.933333333333334e-05, | |
| "loss": 2.0979, | |
| "step": 1799 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.888888888888889e-05, | |
| "loss": 2.105, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.099489688873291, | |
| "eval_rouge1": 13.9638, | |
| "eval_rouge2": 4.7209, | |
| "eval_rougeL": 8.7889, | |
| "eval_rougeLsum": 8.7778, | |
| "eval_runtime": 9040.193, | |
| "eval_samples_per_second": 0.055, | |
| "eval_steps_per_second": 0.055, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.844444444444445e-05, | |
| "loss": 2.0691, | |
| "step": 1801 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 2.0835, | |
| "step": 1802 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.755555555555556e-05, | |
| "loss": 2.0686, | |
| "step": 1803 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.711111111111112e-05, | |
| "loss": 2.2013, | |
| "step": 1804 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.666666666666667e-05, | |
| "loss": 2.0697, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.622222222222222e-05, | |
| "loss": 2.15, | |
| "step": 1806 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.577777777777777e-05, | |
| "loss": 2.079, | |
| "step": 1807 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.533333333333334e-05, | |
| "loss": 2.1418, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.488888888888889e-05, | |
| "loss": 2.1518, | |
| "step": 1809 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.444444444444444e-05, | |
| "loss": 2.12, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.4e-05, | |
| "loss": 2.1908, | |
| "step": 1811 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.355555555555556e-05, | |
| "loss": 2.1359, | |
| "step": 1812 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.311111111111111e-05, | |
| "loss": 2.1208, | |
| "step": 1813 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.266666666666667e-05, | |
| "loss": 2.1014, | |
| "step": 1814 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.222222222222222e-05, | |
| "loss": 2.1451, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.177777777777778e-05, | |
| "loss": 2.0757, | |
| "step": 1816 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 8.133333333333334e-05, | |
| "loss": 2.0587, | |
| "step": 1817 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 8.088888888888889e-05, | |
| "loss": 2.1008, | |
| "step": 1818 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 8.044444444444444e-05, | |
| "loss": 2.0753, | |
| "step": 1819 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 8e-05, | |
| "loss": 2.0912, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.955555555555556e-05, | |
| "loss": 2.0648, | |
| "step": 1821 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.911111111111111e-05, | |
| "loss": 2.0948, | |
| "step": 1822 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.866666666666666e-05, | |
| "loss": 2.1252, | |
| "step": 1823 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.822222222222223e-05, | |
| "loss": 2.1548, | |
| "step": 1824 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.777777777777778e-05, | |
| "loss": 2.1061, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.733333333333333e-05, | |
| "loss": 2.2017, | |
| "step": 1826 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.688888888888889e-05, | |
| "loss": 2.1743, | |
| "step": 1827 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.644444444444445e-05, | |
| "loss": 2.1126, | |
| "step": 1828 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.6e-05, | |
| "loss": 2.0536, | |
| "step": 1829 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.555555555555556e-05, | |
| "loss": 2.0527, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.511111111111111e-05, | |
| "loss": 2.0798, | |
| "step": 1831 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.466666666666667e-05, | |
| "loss": 2.1436, | |
| "step": 1832 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.422222222222223e-05, | |
| "loss": 2.1289, | |
| "step": 1833 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.377777777777778e-05, | |
| "loss": 2.1487, | |
| "step": 1834 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.333333333333333e-05, | |
| "loss": 2.0779, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 7.28888888888889e-05, | |
| "loss": 2.1045, | |
| "step": 1836 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.244444444444445e-05, | |
| "loss": 2.0659, | |
| "step": 1837 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.2e-05, | |
| "loss": 2.0704, | |
| "step": 1838 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.155555555555555e-05, | |
| "loss": 2.0901, | |
| "step": 1839 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.111111111111112e-05, | |
| "loss": 2.104, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.066666666666667e-05, | |
| "loss": 2.1784, | |
| "step": 1841 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.022222222222222e-05, | |
| "loss": 2.0993, | |
| "step": 1842 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.977777777777779e-05, | |
| "loss": 2.1538, | |
| "step": 1843 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.933333333333334e-05, | |
| "loss": 2.0939, | |
| "step": 1844 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.88888888888889e-05, | |
| "loss": 2.2134, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.844444444444445e-05, | |
| "loss": 2.0872, | |
| "step": 1846 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 2.0762, | |
| "step": 1847 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.755555555555557e-05, | |
| "loss": 2.1751, | |
| "step": 1848 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.711111111111112e-05, | |
| "loss": 2.1107, | |
| "step": 1849 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 2.1072, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.622222222222224e-05, | |
| "loss": 2.1557, | |
| "step": 1851 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.577777777777779e-05, | |
| "loss": 2.1582, | |
| "step": 1852 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.533333333333334e-05, | |
| "loss": 2.1279, | |
| "step": 1853 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.488888888888889e-05, | |
| "loss": 2.1797, | |
| "step": 1854 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.444444444444446e-05, | |
| "loss": 2.1951, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 2.018, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.355555555555556e-05, | |
| "loss": 2.1234, | |
| "step": 1857 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.311111111111112e-05, | |
| "loss": 2.1022, | |
| "step": 1858 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.266666666666667e-05, | |
| "loss": 2.0655, | |
| "step": 1859 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.222222222222222e-05, | |
| "loss": 2.138, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.177777777777779e-05, | |
| "loss": 2.1443, | |
| "step": 1861 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.133333333333334e-05, | |
| "loss": 2.1101, | |
| "step": 1862 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.08888888888889e-05, | |
| "loss": 2.0674, | |
| "step": 1863 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6.044444444444445e-05, | |
| "loss": 2.1124, | |
| "step": 1864 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 6e-05, | |
| "loss": 2.1006, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.9555555555555554e-05, | |
| "loss": 2.1313, | |
| "step": 1866 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.911111111111112e-05, | |
| "loss": 2.0989, | |
| "step": 1867 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.866666666666667e-05, | |
| "loss": 2.0372, | |
| "step": 1868 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.8222222222222224e-05, | |
| "loss": 2.1849, | |
| "step": 1869 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.7777777777777776e-05, | |
| "loss": 2.1394, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.7333333333333336e-05, | |
| "loss": 2.1078, | |
| "step": 1871 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.6888888888888895e-05, | |
| "loss": 2.2069, | |
| "step": 1872 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.644444444444445e-05, | |
| "loss": 2.0549, | |
| "step": 1873 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 2.0907, | |
| "step": 1874 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.555555555555556e-05, | |
| "loss": 2.1449, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.511111111111111e-05, | |
| "loss": 2.1387, | |
| "step": 1876 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.466666666666666e-05, | |
| "loss": 2.0901, | |
| "step": 1877 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.422222222222223e-05, | |
| "loss": 2.0245, | |
| "step": 1878 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.377777777777778e-05, | |
| "loss": 2.1397, | |
| "step": 1879 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 2.1229, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.2888888888888885e-05, | |
| "loss": 2.1114, | |
| "step": 1881 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.244444444444445e-05, | |
| "loss": 2.1436, | |
| "step": 1882 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 2.0848, | |
| "step": 1883 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.1555555555555556e-05, | |
| "loss": 2.1456, | |
| "step": 1884 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.111111111111111e-05, | |
| "loss": 2.0572, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.0666666666666674e-05, | |
| "loss": 2.1019, | |
| "step": 1886 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.0222222222222226e-05, | |
| "loss": 2.0934, | |
| "step": 1887 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.977777777777778e-05, | |
| "loss": 2.1123, | |
| "step": 1888 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.933333333333334e-05, | |
| "loss": 2.1475, | |
| "step": 1889 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.888888888888889e-05, | |
| "loss": 2.14, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.844444444444445e-05, | |
| "loss": 2.1103, | |
| "step": 1891 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.8e-05, | |
| "loss": 2.0636, | |
| "step": 1892 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.755555555555556e-05, | |
| "loss": 2.0378, | |
| "step": 1893 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.711111111111111e-05, | |
| "loss": 2.1109, | |
| "step": 1894 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 2.0696, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.6222222222222224e-05, | |
| "loss": 2.117, | |
| "step": 1896 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.577777777777778e-05, | |
| "loss": 2.0447, | |
| "step": 1897 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.5333333333333335e-05, | |
| "loss": 2.1653, | |
| "step": 1898 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.4888888888888894e-05, | |
| "loss": 2.0236, | |
| "step": 1899 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "loss": 2.1244, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 2.1211, | |
| "step": 1901 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.355555555555556e-05, | |
| "loss": 2.2072, | |
| "step": 1902 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.311111111111111e-05, | |
| "loss": 2.1041, | |
| "step": 1903 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.266666666666667e-05, | |
| "loss": 2.071, | |
| "step": 1904 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.222222222222222e-05, | |
| "loss": 2.0229, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.177777777777778e-05, | |
| "loss": 2.0974, | |
| "step": 1906 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.133333333333333e-05, | |
| "loss": 2.1007, | |
| "step": 1907 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.088888888888889e-05, | |
| "loss": 2.1028, | |
| "step": 1908 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.0444444444444444e-05, | |
| "loss": 2.097, | |
| "step": 1909 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4e-05, | |
| "loss": 2.1031, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.9555555555555556e-05, | |
| "loss": 2.0905, | |
| "step": 1911 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.9111111111111115e-05, | |
| "loss": 2.1131, | |
| "step": 1912 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.866666666666667e-05, | |
| "loss": 2.1349, | |
| "step": 1913 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.8222222222222226e-05, | |
| "loss": 2.1394, | |
| "step": 1914 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.777777777777778e-05, | |
| "loss": 2.0631, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.733333333333334e-05, | |
| "loss": 2.1042, | |
| "step": 1916 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.688888888888889e-05, | |
| "loss": 2.1382, | |
| "step": 1917 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.644444444444445e-05, | |
| "loss": 2.1726, | |
| "step": 1918 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.6e-05, | |
| "loss": 2.1671, | |
| "step": 1919 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.555555555555556e-05, | |
| "loss": 2.0649, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.511111111111111e-05, | |
| "loss": 2.0917, | |
| "step": 1921 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.466666666666667e-05, | |
| "loss": 2.0466, | |
| "step": 1922 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.4222222222222224e-05, | |
| "loss": 2.0903, | |
| "step": 1923 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.377777777777778e-05, | |
| "loss": 2.0701, | |
| "step": 1924 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 2.0987, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.2888888888888894e-05, | |
| "loss": 2.1408, | |
| "step": 1926 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.2444444444444446e-05, | |
| "loss": 2.104, | |
| "step": 1927 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 2.088, | |
| "step": 1928 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.155555555555556e-05, | |
| "loss": 2.0903, | |
| "step": 1929 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.111111111111111e-05, | |
| "loss": 2.0852, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.066666666666667e-05, | |
| "loss": 2.0749, | |
| "step": 1931 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.0222222222222225e-05, | |
| "loss": 2.0794, | |
| "step": 1932 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.9777777777777777e-05, | |
| "loss": 2.0552, | |
| "step": 1933 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.9333333333333336e-05, | |
| "loss": 2.0896, | |
| "step": 1934 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.8888888888888888e-05, | |
| "loss": 2.1586, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.8444444444444447e-05, | |
| "loss": 2.1477, | |
| "step": 1936 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 2.0847, | |
| "step": 1937 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7555555555555555e-05, | |
| "loss": 2.0892, | |
| "step": 1938 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7111111111111114e-05, | |
| "loss": 2.1891, | |
| "step": 1939 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 2.0856, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.6222222222222226e-05, | |
| "loss": 2.1166, | |
| "step": 1941 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.5777777777777778e-05, | |
| "loss": 2.15, | |
| "step": 1942 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.5333333333333337e-05, | |
| "loss": 2.0967, | |
| "step": 1943 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.488888888888889e-05, | |
| "loss": 2.1111, | |
| "step": 1944 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.4444444444444445e-05, | |
| "loss": 2.097, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.4e-05, | |
| "loss": 2.1431, | |
| "step": 1946 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.3555555555555556e-05, | |
| "loss": 2.1265, | |
| "step": 1947 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.3111111111111112e-05, | |
| "loss": 2.2027, | |
| "step": 1948 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.2666666666666668e-05, | |
| "loss": 2.0808, | |
| "step": 1949 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.2222222222222223e-05, | |
| "loss": 2.0719, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.177777777777778e-05, | |
| "loss": 2.0903, | |
| "step": 1951 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.1333333333333335e-05, | |
| "loss": 2.0455, | |
| "step": 1952 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.088888888888889e-05, | |
| "loss": 2.1111, | |
| "step": 1953 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.0444444444444446e-05, | |
| "loss": 2.1216, | |
| "step": 1954 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2e-05, | |
| "loss": 2.1088, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.9555555555555557e-05, | |
| "loss": 2.0789, | |
| "step": 1956 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.9111111111111113e-05, | |
| "loss": 2.0954, | |
| "step": 1957 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.866666666666667e-05, | |
| "loss": 2.0936, | |
| "step": 1958 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.8222222222222224e-05, | |
| "loss": 2.1256, | |
| "step": 1959 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.777777777777778e-05, | |
| "loss": 2.1112, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.7333333333333336e-05, | |
| "loss": 2.0945, | |
| "step": 1961 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.688888888888889e-05, | |
| "loss": 2.1225, | |
| "step": 1962 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.6444444444444447e-05, | |
| "loss": 2.0004, | |
| "step": 1963 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.1143, | |
| "step": 1964 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.5555555555555555e-05, | |
| "loss": 2.0987, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.5111111111111112e-05, | |
| "loss": 2.0606, | |
| "step": 1966 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.4666666666666668e-05, | |
| "loss": 2.0714, | |
| "step": 1967 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.4222222222222224e-05, | |
| "loss": 2.0926, | |
| "step": 1968 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.3777777777777778e-05, | |
| "loss": 2.139, | |
| "step": 1969 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 2.1304, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.2888888888888889e-05, | |
| "loss": 2.1202, | |
| "step": 1971 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.2444444444444445e-05, | |
| "loss": 2.1232, | |
| "step": 1972 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.1089, | |
| "step": 1973 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.1555555555555556e-05, | |
| "loss": 2.1162, | |
| "step": 1974 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.1111111111111112e-05, | |
| "loss": 2.0465, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.0666666666666667e-05, | |
| "loss": 2.1557, | |
| "step": 1976 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.0222222222222223e-05, | |
| "loss": 2.1571, | |
| "step": 1977 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 2.0609, | |
| "step": 1978 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 2.1276, | |
| "step": 1979 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.88888888888889e-06, | |
| "loss": 2.0715, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 2.0817, | |
| "step": 1981 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 2.1797, | |
| "step": 1982 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 2.1437, | |
| "step": 1983 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 2.1116, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 2.0851, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.222222222222222e-06, | |
| "loss": 2.1341, | |
| "step": 1986 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 2.1147, | |
| "step": 1987 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 2.1872, | |
| "step": 1988 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 2.1453, | |
| "step": 1989 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.444444444444445e-06, | |
| "loss": 2.1466, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.213, | |
| "step": 1991 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 2.0331, | |
| "step": 1992 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.111111111111111e-06, | |
| "loss": 2.0973, | |
| "step": 1993 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 2.0288, | |
| "step": 1994 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.2222222222222225e-06, | |
| "loss": 2.1309, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 2.1932, | |
| "step": 1996 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 2.037, | |
| "step": 1997 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 2.1351, | |
| "step": 1998 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 2.1637, | |
| "step": 1999 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0, | |
| "loss": 2.1263, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "eval_gen_len": 1023.0, | |
| "eval_loss": 2.0949487686157227, | |
| "eval_rouge1": 14.0237, | |
| "eval_rouge2": 4.8219, | |
| "eval_rougeL": 8.8145, | |
| "eval_rougeLsum": 8.793, | |
| "eval_runtime": 9159.2647, | |
| "eval_samples_per_second": 0.055, | |
| "eval_steps_per_second": 0.055, | |
| "step": 2000 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 1.608210613960658e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |