{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.999338361783777, "global_step": 9445, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 7.042253521126761e-08, "loss": 0.9991, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.4084507042253522e-07, "loss": 0.982, "step": 2 }, { "epoch": 0.0, "learning_rate": 2.1126760563380284e-07, "loss": 1.0135, "step": 3 }, { "epoch": 0.0, "learning_rate": 2.8169014084507043e-07, "loss": 0.9558, "step": 4 }, { "epoch": 0.0, "learning_rate": 3.521126760563381e-07, "loss": 1.0749, "step": 5 }, { "epoch": 0.0, "learning_rate": 4.225352112676057e-07, "loss": 1.0304, "step": 6 }, { "epoch": 0.0, "learning_rate": 4.929577464788733e-07, "loss": 0.9809, "step": 7 }, { "epoch": 0.0, "learning_rate": 5.633802816901409e-07, "loss": 0.9271, "step": 8 }, { "epoch": 0.0, "learning_rate": 6.338028169014085e-07, "loss": 0.911, "step": 9 }, { "epoch": 0.01, "learning_rate": 7.042253521126762e-07, "loss": 0.8964, "step": 10 }, { "epoch": 0.01, "learning_rate": 7.746478873239437e-07, "loss": 0.9058, "step": 11 }, { "epoch": 0.01, "learning_rate": 8.450704225352114e-07, "loss": 0.9289, "step": 12 }, { "epoch": 0.01, "learning_rate": 9.154929577464789e-07, "loss": 0.7715, "step": 13 }, { "epoch": 0.01, "learning_rate": 9.859154929577465e-07, "loss": 0.745, "step": 14 }, { "epoch": 0.01, "learning_rate": 1.0563380281690142e-06, "loss": 0.7159, "step": 15 }, { "epoch": 0.01, "learning_rate": 1.1267605633802817e-06, "loss": 0.698, "step": 16 }, { "epoch": 0.01, "learning_rate": 1.1971830985915492e-06, "loss": 0.6938, "step": 17 }, { "epoch": 0.01, "learning_rate": 1.267605633802817e-06, "loss": 0.5836, "step": 18 }, { "epoch": 0.01, "learning_rate": 1.3380281690140844e-06, "loss": 0.674, "step": 19 }, { "epoch": 0.01, "learning_rate": 1.4084507042253523e-06, "loss": 0.6405, "step": 20 }, { "epoch": 0.01, "learning_rate": 1.4788732394366198e-06, "loss": 0.5486, "step": 21 }, { "epoch": 0.01, "learning_rate": 1.5492957746478873e-06, "loss": 0.5166, "step": 22 }, { "epoch": 0.01, "learning_rate": 1.6197183098591552e-06, "loss": 0.338, "step": 23 }, { "epoch": 0.01, "learning_rate": 1.6901408450704227e-06, "loss": 0.5036, "step": 24 }, { "epoch": 0.01, "learning_rate": 1.7605633802816902e-06, "loss": 0.4492, "step": 25 }, { "epoch": 0.01, "learning_rate": 1.8309859154929579e-06, "loss": 0.3996, "step": 26 }, { "epoch": 0.01, "learning_rate": 1.9014084507042254e-06, "loss": 0.5513, "step": 27 }, { "epoch": 0.01, "learning_rate": 1.971830985915493e-06, "loss": 0.3885, "step": 28 }, { "epoch": 0.02, "learning_rate": 2.0422535211267608e-06, "loss": 0.3533, "step": 29 }, { "epoch": 0.02, "learning_rate": 2.1126760563380285e-06, "loss": 0.4806, "step": 30 }, { "epoch": 0.02, "learning_rate": 2.1830985915492958e-06, "loss": 0.3584, "step": 31 }, { "epoch": 0.02, "learning_rate": 2.2535211267605635e-06, "loss": 0.3971, "step": 32 }, { "epoch": 0.02, "learning_rate": 2.323943661971831e-06, "loss": 0.3297, "step": 33 }, { "epoch": 0.02, "learning_rate": 2.3943661971830984e-06, "loss": 0.3274, "step": 34 }, { "epoch": 0.02, "learning_rate": 2.4647887323943666e-06, "loss": 0.3345, "step": 35 }, { "epoch": 0.02, "learning_rate": 2.535211267605634e-06, "loss": 0.4462, "step": 36 }, { "epoch": 0.02, "learning_rate": 2.6056338028169015e-06, "loss": 0.4242, "step": 37 }, { "epoch": 0.02, "learning_rate": 2.676056338028169e-06, "loss": 0.3258, "step": 38 }, { "epoch": 0.02, "learning_rate": 2.746478873239437e-06, "loss": 0.3861, "step": 39 }, { "epoch": 0.02, "learning_rate": 2.8169014084507046e-06, "loss": 0.5329, "step": 40 }, { "epoch": 0.02, "learning_rate": 2.887323943661972e-06, "loss": 0.5445, "step": 41 }, { "epoch": 0.02, "learning_rate": 2.9577464788732396e-06, "loss": 0.3269, "step": 42 }, { "epoch": 0.02, "learning_rate": 3.0281690140845073e-06, "loss": 0.485, "step": 43 }, { "epoch": 0.02, "learning_rate": 3.0985915492957746e-06, "loss": 0.2517, "step": 44 }, { "epoch": 0.02, "learning_rate": 3.1690140845070427e-06, "loss": 0.3704, "step": 45 }, { "epoch": 0.02, "learning_rate": 3.2394366197183104e-06, "loss": 0.2578, "step": 46 }, { "epoch": 0.02, "learning_rate": 3.3098591549295777e-06, "loss": 0.2914, "step": 47 }, { "epoch": 0.03, "learning_rate": 3.3802816901408454e-06, "loss": 0.3726, "step": 48 }, { "epoch": 0.03, "learning_rate": 3.4507042253521127e-06, "loss": 0.268, "step": 49 }, { "epoch": 0.03, "learning_rate": 3.5211267605633804e-06, "loss": 0.5147, "step": 50 }, { "epoch": 0.03, "learning_rate": 3.5915492957746485e-06, "loss": 0.4073, "step": 51 }, { "epoch": 0.03, "learning_rate": 3.6619718309859158e-06, "loss": 0.4051, "step": 52 }, { "epoch": 0.03, "learning_rate": 3.7323943661971835e-06, "loss": 0.3266, "step": 53 }, { "epoch": 0.03, "learning_rate": 3.8028169014084508e-06, "loss": 0.3473, "step": 54 }, { "epoch": 0.03, "learning_rate": 3.873239436619718e-06, "loss": 0.3463, "step": 55 }, { "epoch": 0.03, "learning_rate": 3.943661971830986e-06, "loss": 0.2735, "step": 56 }, { "epoch": 0.03, "learning_rate": 4.014084507042254e-06, "loss": 0.3144, "step": 57 }, { "epoch": 0.03, "learning_rate": 4.0845070422535216e-06, "loss": 0.3105, "step": 58 }, { "epoch": 0.03, "learning_rate": 4.154929577464789e-06, "loss": 0.4126, "step": 59 }, { "epoch": 0.03, "learning_rate": 4.225352112676057e-06, "loss": 0.3294, "step": 60 }, { "epoch": 0.03, "learning_rate": 4.295774647887324e-06, "loss": 0.2541, "step": 61 }, { "epoch": 0.03, "learning_rate": 4.3661971830985915e-06, "loss": 0.2963, "step": 62 }, { "epoch": 0.03, "learning_rate": 4.43661971830986e-06, "loss": 0.2646, "step": 63 }, { "epoch": 0.03, "learning_rate": 4.507042253521127e-06, "loss": 0.3772, "step": 64 }, { "epoch": 0.03, "learning_rate": 4.577464788732395e-06, "loss": 0.4184, "step": 65 }, { "epoch": 0.03, "learning_rate": 4.647887323943662e-06, "loss": 0.2736, "step": 66 }, { "epoch": 0.04, "learning_rate": 4.71830985915493e-06, "loss": 0.3726, "step": 67 }, { "epoch": 0.04, "learning_rate": 4.788732394366197e-06, "loss": 0.3049, "step": 68 }, { "epoch": 0.04, "learning_rate": 4.859154929577465e-06, "loss": 0.3652, "step": 69 }, { "epoch": 0.04, "learning_rate": 4.929577464788733e-06, "loss": 0.4649, "step": 70 }, { "epoch": 0.04, "learning_rate": 5e-06, "loss": 0.4129, "step": 71 }, { "epoch": 0.04, "learning_rate": 5.070422535211268e-06, "loss": 0.3197, "step": 72 }, { "epoch": 0.04, "learning_rate": 5.140845070422536e-06, "loss": 0.2889, "step": 73 }, { "epoch": 0.04, "learning_rate": 5.211267605633803e-06, "loss": 0.3439, "step": 74 }, { "epoch": 0.04, "learning_rate": 5.28169014084507e-06, "loss": 0.3384, "step": 75 }, { "epoch": 0.04, "learning_rate": 5.352112676056338e-06, "loss": 0.2421, "step": 76 }, { "epoch": 0.04, "learning_rate": 5.422535211267607e-06, "loss": 0.2434, "step": 77 }, { "epoch": 0.04, "learning_rate": 5.492957746478874e-06, "loss": 0.3097, "step": 78 }, { "epoch": 0.04, "learning_rate": 5.563380281690142e-06, "loss": 0.253, "step": 79 }, { "epoch": 0.04, "learning_rate": 5.633802816901409e-06, "loss": 0.3247, "step": 80 }, { "epoch": 0.04, "learning_rate": 5.7042253521126766e-06, "loss": 0.3341, "step": 81 }, { "epoch": 0.04, "learning_rate": 5.774647887323944e-06, "loss": 0.2774, "step": 82 }, { "epoch": 0.04, "learning_rate": 5.845070422535212e-06, "loss": 0.3111, "step": 83 }, { "epoch": 0.04, "learning_rate": 5.915492957746479e-06, "loss": 0.3988, "step": 84 }, { "epoch": 0.04, "learning_rate": 5.9859154929577465e-06, "loss": 0.3226, "step": 85 }, { "epoch": 0.05, "learning_rate": 6.056338028169015e-06, "loss": 0.2715, "step": 86 }, { "epoch": 0.05, "learning_rate": 6.126760563380282e-06, "loss": 0.3528, "step": 87 }, { "epoch": 0.05, "learning_rate": 6.197183098591549e-06, "loss": 0.3711, "step": 88 }, { "epoch": 0.05, "learning_rate": 6.267605633802818e-06, "loss": 0.3835, "step": 89 }, { "epoch": 0.05, "learning_rate": 6.3380281690140855e-06, "loss": 0.2041, "step": 90 }, { "epoch": 0.05, "learning_rate": 6.408450704225353e-06, "loss": 0.2666, "step": 91 }, { "epoch": 0.05, "learning_rate": 6.478873239436621e-06, "loss": 0.3453, "step": 92 }, { "epoch": 0.05, "learning_rate": 6.549295774647888e-06, "loss": 0.2053, "step": 93 }, { "epoch": 0.05, "learning_rate": 6.619718309859155e-06, "loss": 0.4689, "step": 94 }, { "epoch": 0.05, "learning_rate": 6.690140845070423e-06, "loss": 0.2694, "step": 95 }, { "epoch": 0.05, "learning_rate": 6.760563380281691e-06, "loss": 0.2822, "step": 96 }, { "epoch": 0.05, "learning_rate": 6.830985915492958e-06, "loss": 0.3153, "step": 97 }, { "epoch": 0.05, "learning_rate": 6.901408450704225e-06, "loss": 0.2786, "step": 98 }, { "epoch": 0.05, "learning_rate": 6.9718309859154935e-06, "loss": 0.3277, "step": 99 }, { "epoch": 0.05, "learning_rate": 7.042253521126761e-06, "loss": 0.2813, "step": 100 }, { "epoch": 0.05, "learning_rate": 7.112676056338029e-06, "loss": 0.3475, "step": 101 }, { "epoch": 0.05, "learning_rate": 7.183098591549297e-06, "loss": 0.2379, "step": 102 }, { "epoch": 0.05, "learning_rate": 7.253521126760564e-06, "loss": 0.3561, "step": 103 }, { "epoch": 0.06, "learning_rate": 7.3239436619718316e-06, "loss": 0.2051, "step": 104 }, { "epoch": 0.06, "learning_rate": 7.3943661971831e-06, "loss": 0.3269, "step": 105 }, { "epoch": 0.06, "learning_rate": 7.464788732394367e-06, "loss": 0.3184, "step": 106 }, { "epoch": 0.06, "learning_rate": 7.535211267605634e-06, "loss": 0.5486, "step": 107 }, { "epoch": 0.06, "learning_rate": 7.6056338028169015e-06, "loss": 0.2629, "step": 108 }, { "epoch": 0.06, "learning_rate": 7.67605633802817e-06, "loss": 0.3492, "step": 109 }, { "epoch": 0.06, "learning_rate": 7.746478873239436e-06, "loss": 0.217, "step": 110 }, { "epoch": 0.06, "learning_rate": 7.816901408450704e-06, "loss": 0.2889, "step": 111 }, { "epoch": 0.06, "learning_rate": 7.887323943661972e-06, "loss": 0.2451, "step": 112 }, { "epoch": 0.06, "learning_rate": 7.95774647887324e-06, "loss": 0.2728, "step": 113 }, { "epoch": 0.06, "learning_rate": 8.028169014084509e-06, "loss": 0.2953, "step": 114 }, { "epoch": 0.06, "learning_rate": 8.098591549295775e-06, "loss": 0.2847, "step": 115 }, { "epoch": 0.06, "learning_rate": 8.169014084507043e-06, "loss": 0.2623, "step": 116 }, { "epoch": 0.06, "learning_rate": 8.239436619718311e-06, "loss": 0.3969, "step": 117 }, { "epoch": 0.06, "learning_rate": 8.309859154929578e-06, "loss": 0.2579, "step": 118 }, { "epoch": 0.06, "learning_rate": 8.380281690140846e-06, "loss": 0.245, "step": 119 }, { "epoch": 0.06, "learning_rate": 8.450704225352114e-06, "loss": 0.2743, "step": 120 }, { "epoch": 0.06, "learning_rate": 8.52112676056338e-06, "loss": 0.2273, "step": 121 }, { "epoch": 0.06, "learning_rate": 8.591549295774648e-06, "loss": 0.3096, "step": 122 }, { "epoch": 0.07, "learning_rate": 8.661971830985915e-06, "loss": 0.415, "step": 123 }, { "epoch": 0.07, "learning_rate": 8.732394366197183e-06, "loss": 0.2351, "step": 124 }, { "epoch": 0.07, "learning_rate": 8.802816901408451e-06, "loss": 0.237, "step": 125 }, { "epoch": 0.07, "learning_rate": 8.87323943661972e-06, "loss": 0.3535, "step": 126 }, { "epoch": 0.07, "learning_rate": 8.943661971830987e-06, "loss": 0.4072, "step": 127 }, { "epoch": 0.07, "learning_rate": 9.014084507042254e-06, "loss": 0.3245, "step": 128 }, { "epoch": 0.07, "learning_rate": 9.084507042253522e-06, "loss": 0.2921, "step": 129 }, { "epoch": 0.07, "learning_rate": 9.15492957746479e-06, "loss": 0.1987, "step": 130 }, { "epoch": 0.07, "learning_rate": 9.225352112676057e-06, "loss": 0.2812, "step": 131 }, { "epoch": 0.07, "learning_rate": 9.295774647887325e-06, "loss": 0.2603, "step": 132 }, { "epoch": 0.07, "learning_rate": 9.366197183098593e-06, "loss": 0.4538, "step": 133 }, { "epoch": 0.07, "learning_rate": 9.43661971830986e-06, "loss": 0.2754, "step": 134 }, { "epoch": 0.07, "learning_rate": 9.507042253521127e-06, "loss": 0.3688, "step": 135 }, { "epoch": 0.07, "learning_rate": 9.577464788732394e-06, "loss": 0.378, "step": 136 }, { "epoch": 0.07, "learning_rate": 9.647887323943664e-06, "loss": 0.4352, "step": 137 }, { "epoch": 0.07, "learning_rate": 9.71830985915493e-06, "loss": 0.1091, "step": 138 }, { "epoch": 0.07, "learning_rate": 9.788732394366198e-06, "loss": 0.3099, "step": 139 }, { "epoch": 0.07, "learning_rate": 9.859154929577466e-06, "loss": 0.2694, "step": 140 }, { "epoch": 0.07, "learning_rate": 9.929577464788733e-06, "loss": 0.4208, "step": 141 }, { "epoch": 0.08, "learning_rate": 1e-05, "loss": 0.2774, "step": 142 }, { "epoch": 0.08, "learning_rate": 1.0070422535211269e-05, "loss": 0.2744, "step": 143 }, { "epoch": 0.08, "learning_rate": 1.0140845070422535e-05, "loss": 0.3598, "step": 144 }, { "epoch": 0.08, "learning_rate": 1.0211267605633803e-05, "loss": 0.3036, "step": 145 }, { "epoch": 0.08, "learning_rate": 1.0281690140845072e-05, "loss": 0.1622, "step": 146 }, { "epoch": 0.08, "learning_rate": 1.0352112676056338e-05, "loss": 0.2657, "step": 147 }, { "epoch": 0.08, "learning_rate": 1.0422535211267606e-05, "loss": 0.2632, "step": 148 }, { "epoch": 0.08, "learning_rate": 1.0492957746478873e-05, "loss": 0.1815, "step": 149 }, { "epoch": 0.08, "learning_rate": 1.056338028169014e-05, "loss": 0.2577, "step": 150 }, { "epoch": 0.08, "learning_rate": 1.0633802816901409e-05, "loss": 0.2063, "step": 151 }, { "epoch": 0.08, "learning_rate": 1.0704225352112675e-05, "loss": 0.2859, "step": 152 }, { "epoch": 0.08, "learning_rate": 1.0774647887323943e-05, "loss": 0.406, "step": 153 }, { "epoch": 0.08, "learning_rate": 1.0845070422535213e-05, "loss": 0.4142, "step": 154 }, { "epoch": 0.08, "learning_rate": 1.0915492957746481e-05, "loss": 0.1504, "step": 155 }, { "epoch": 0.08, "learning_rate": 1.0985915492957748e-05, "loss": 0.1746, "step": 156 }, { "epoch": 0.08, "learning_rate": 1.1056338028169016e-05, "loss": 0.1952, "step": 157 }, { "epoch": 0.08, "learning_rate": 1.1126760563380284e-05, "loss": 0.2084, "step": 158 }, { "epoch": 0.08, "learning_rate": 1.119718309859155e-05, "loss": 0.2945, "step": 159 }, { "epoch": 0.08, "learning_rate": 1.1267605633802819e-05, "loss": 0.247, "step": 160 }, { "epoch": 0.09, "learning_rate": 1.1338028169014087e-05, "loss": 0.1968, "step": 161 }, { "epoch": 0.09, "learning_rate": 1.1408450704225353e-05, "loss": 0.1935, "step": 162 }, { "epoch": 0.09, "learning_rate": 1.1478873239436621e-05, "loss": 0.3113, "step": 163 }, { "epoch": 0.09, "learning_rate": 1.1549295774647888e-05, "loss": 0.1949, "step": 164 }, { "epoch": 0.09, "learning_rate": 1.1619718309859156e-05, "loss": 0.3254, "step": 165 }, { "epoch": 0.09, "learning_rate": 1.1690140845070424e-05, "loss": 0.2308, "step": 166 }, { "epoch": 0.09, "learning_rate": 1.176056338028169e-05, "loss": 0.2349, "step": 167 }, { "epoch": 0.09, "learning_rate": 1.1830985915492958e-05, "loss": 0.37, "step": 168 }, { "epoch": 0.09, "learning_rate": 1.1901408450704227e-05, "loss": 0.2988, "step": 169 }, { "epoch": 0.09, "learning_rate": 1.1971830985915493e-05, "loss": 0.3991, "step": 170 }, { "epoch": 0.09, "learning_rate": 1.2042253521126761e-05, "loss": 0.2378, "step": 171 }, { "epoch": 0.09, "learning_rate": 1.211267605633803e-05, "loss": 0.3736, "step": 172 }, { "epoch": 0.09, "learning_rate": 1.2183098591549296e-05, "loss": 0.3531, "step": 173 }, { "epoch": 0.09, "learning_rate": 1.2253521126760564e-05, "loss": 0.2288, "step": 174 }, { "epoch": 0.09, "learning_rate": 1.232394366197183e-05, "loss": 0.1799, "step": 175 }, { "epoch": 0.09, "learning_rate": 1.2394366197183098e-05, "loss": 0.3801, "step": 176 }, { "epoch": 0.09, "learning_rate": 1.2464788732394367e-05, "loss": 0.3145, "step": 177 }, { "epoch": 0.09, "learning_rate": 1.2535211267605636e-05, "loss": 0.2927, "step": 178 }, { "epoch": 0.09, "learning_rate": 1.2605633802816903e-05, "loss": 0.3182, "step": 179 }, { "epoch": 0.1, "learning_rate": 1.2676056338028171e-05, "loss": 0.27, "step": 180 }, { "epoch": 0.1, "learning_rate": 1.2746478873239439e-05, "loss": 0.2261, "step": 181 }, { "epoch": 0.1, "learning_rate": 1.2816901408450705e-05, "loss": 0.2645, "step": 182 }, { "epoch": 0.1, "learning_rate": 1.2887323943661974e-05, "loss": 0.2815, "step": 183 }, { "epoch": 0.1, "learning_rate": 1.2957746478873242e-05, "loss": 0.2574, "step": 184 }, { "epoch": 0.1, "learning_rate": 1.3028169014084508e-05, "loss": 0.3003, "step": 185 }, { "epoch": 0.1, "learning_rate": 1.3098591549295776e-05, "loss": 0.2545, "step": 186 }, { "epoch": 0.1, "learning_rate": 1.3169014084507044e-05, "loss": 0.2763, "step": 187 }, { "epoch": 0.1, "learning_rate": 1.323943661971831e-05, "loss": 0.36, "step": 188 }, { "epoch": 0.1, "learning_rate": 1.3309859154929579e-05, "loss": 0.1868, "step": 189 }, { "epoch": 0.1, "learning_rate": 1.3380281690140845e-05, "loss": 0.4571, "step": 190 }, { "epoch": 0.1, "learning_rate": 1.3450704225352114e-05, "loss": 0.2419, "step": 191 }, { "epoch": 0.1, "learning_rate": 1.3521126760563382e-05, "loss": 0.2845, "step": 192 }, { "epoch": 0.1, "learning_rate": 1.3591549295774648e-05, "loss": 0.3147, "step": 193 }, { "epoch": 0.1, "learning_rate": 1.3661971830985916e-05, "loss": 0.419, "step": 194 }, { "epoch": 0.1, "learning_rate": 1.3732394366197184e-05, "loss": 0.2945, "step": 195 }, { "epoch": 0.1, "learning_rate": 1.380281690140845e-05, "loss": 0.1917, "step": 196 }, { "epoch": 0.1, "learning_rate": 1.3873239436619719e-05, "loss": 0.2909, "step": 197 }, { "epoch": 0.1, "learning_rate": 1.3943661971830987e-05, "loss": 0.1775, "step": 198 }, { "epoch": 0.11, "learning_rate": 1.4014084507042253e-05, "loss": 0.2548, "step": 199 }, { "epoch": 0.11, "learning_rate": 1.4084507042253522e-05, "loss": 0.2051, "step": 200 }, { "epoch": 0.11, "learning_rate": 1.4154929577464788e-05, "loss": 0.3303, "step": 201 }, { "epoch": 0.11, "learning_rate": 1.4225352112676058e-05, "loss": 0.2845, "step": 202 }, { "epoch": 0.11, "learning_rate": 1.4295774647887326e-05, "loss": 0.2016, "step": 203 }, { "epoch": 0.11, "learning_rate": 1.4366197183098594e-05, "loss": 0.1657, "step": 204 }, { "epoch": 0.11, "learning_rate": 1.443661971830986e-05, "loss": 0.338, "step": 205 }, { "epoch": 0.11, "learning_rate": 1.4507042253521129e-05, "loss": 0.197, "step": 206 }, { "epoch": 0.11, "learning_rate": 1.4577464788732397e-05, "loss": 0.2388, "step": 207 }, { "epoch": 0.11, "learning_rate": 1.4647887323943663e-05, "loss": 0.1975, "step": 208 }, { "epoch": 0.11, "learning_rate": 1.4718309859154931e-05, "loss": 0.2463, "step": 209 }, { "epoch": 0.11, "learning_rate": 1.47887323943662e-05, "loss": 0.3282, "step": 210 }, { "epoch": 0.11, "learning_rate": 1.4859154929577466e-05, "loss": 0.3105, "step": 211 }, { "epoch": 0.11, "learning_rate": 1.4929577464788734e-05, "loss": 0.2413, "step": 212 }, { "epoch": 0.11, "learning_rate": 1.5000000000000002e-05, "loss": 0.3187, "step": 213 }, { "epoch": 0.11, "learning_rate": 1.5070422535211269e-05, "loss": 0.3096, "step": 214 }, { "epoch": 0.11, "learning_rate": 1.5140845070422537e-05, "loss": 0.1883, "step": 215 }, { "epoch": 0.11, "learning_rate": 1.5211267605633803e-05, "loss": 0.1222, "step": 216 }, { "epoch": 0.11, "learning_rate": 1.528169014084507e-05, "loss": 0.254, "step": 217 }, { "epoch": 0.12, "learning_rate": 1.535211267605634e-05, "loss": 0.297, "step": 218 }, { "epoch": 0.12, "learning_rate": 1.5422535211267607e-05, "loss": 0.3285, "step": 219 }, { "epoch": 0.12, "learning_rate": 1.5492957746478872e-05, "loss": 0.2907, "step": 220 }, { "epoch": 0.12, "learning_rate": 1.556338028169014e-05, "loss": 0.3265, "step": 221 }, { "epoch": 0.12, "learning_rate": 1.563380281690141e-05, "loss": 0.2819, "step": 222 }, { "epoch": 0.12, "learning_rate": 1.5704225352112677e-05, "loss": 0.3837, "step": 223 }, { "epoch": 0.12, "learning_rate": 1.5774647887323945e-05, "loss": 0.3022, "step": 224 }, { "epoch": 0.12, "learning_rate": 1.5845070422535213e-05, "loss": 0.2491, "step": 225 }, { "epoch": 0.12, "learning_rate": 1.591549295774648e-05, "loss": 0.2202, "step": 226 }, { "epoch": 0.12, "learning_rate": 1.598591549295775e-05, "loss": 0.2314, "step": 227 }, { "epoch": 0.12, "learning_rate": 1.6056338028169017e-05, "loss": 0.3388, "step": 228 }, { "epoch": 0.12, "learning_rate": 1.6126760563380285e-05, "loss": 0.3595, "step": 229 }, { "epoch": 0.12, "learning_rate": 1.619718309859155e-05, "loss": 0.2253, "step": 230 }, { "epoch": 0.12, "learning_rate": 1.6267605633802818e-05, "loss": 0.4379, "step": 231 }, { "epoch": 0.12, "learning_rate": 1.6338028169014086e-05, "loss": 0.2713, "step": 232 }, { "epoch": 0.12, "learning_rate": 1.6408450704225354e-05, "loss": 0.209, "step": 233 }, { "epoch": 0.12, "learning_rate": 1.6478873239436623e-05, "loss": 0.3046, "step": 234 }, { "epoch": 0.12, "learning_rate": 1.6549295774647887e-05, "loss": 0.2628, "step": 235 }, { "epoch": 0.12, "learning_rate": 1.6619718309859155e-05, "loss": 0.2569, "step": 236 }, { "epoch": 0.13, "learning_rate": 1.6690140845070424e-05, "loss": 0.2466, "step": 237 }, { "epoch": 0.13, "learning_rate": 1.676056338028169e-05, "loss": 0.2371, "step": 238 }, { "epoch": 0.13, "learning_rate": 1.683098591549296e-05, "loss": 0.1569, "step": 239 }, { "epoch": 0.13, "learning_rate": 1.6901408450704228e-05, "loss": 0.2176, "step": 240 }, { "epoch": 0.13, "learning_rate": 1.6971830985915493e-05, "loss": 0.264, "step": 241 }, { "epoch": 0.13, "learning_rate": 1.704225352112676e-05, "loss": 0.25, "step": 242 }, { "epoch": 0.13, "learning_rate": 1.711267605633803e-05, "loss": 0.2626, "step": 243 }, { "epoch": 0.13, "learning_rate": 1.7183098591549297e-05, "loss": 0.2664, "step": 244 }, { "epoch": 0.13, "learning_rate": 1.7253521126760565e-05, "loss": 0.3253, "step": 245 }, { "epoch": 0.13, "learning_rate": 1.732394366197183e-05, "loss": 0.3211, "step": 246 }, { "epoch": 0.13, "learning_rate": 1.7394366197183098e-05, "loss": 0.2281, "step": 247 }, { "epoch": 0.13, "learning_rate": 1.7464788732394366e-05, "loss": 0.2939, "step": 248 }, { "epoch": 0.13, "learning_rate": 1.7535211267605638e-05, "loss": 0.2551, "step": 249 }, { "epoch": 0.13, "learning_rate": 1.7605633802816902e-05, "loss": 0.2686, "step": 250 }, { "epoch": 0.13, "learning_rate": 1.767605633802817e-05, "loss": 0.2199, "step": 251 }, { "epoch": 0.13, "learning_rate": 1.774647887323944e-05, "loss": 0.3774, "step": 252 }, { "epoch": 0.13, "learning_rate": 1.7816901408450707e-05, "loss": 0.3094, "step": 253 }, { "epoch": 0.13, "learning_rate": 1.7887323943661975e-05, "loss": 0.1645, "step": 254 }, { "epoch": 0.13, "learning_rate": 1.7957746478873243e-05, "loss": 0.2424, "step": 255 }, { "epoch": 0.14, "learning_rate": 1.8028169014084508e-05, "loss": 0.2802, "step": 256 }, { "epoch": 0.14, "learning_rate": 1.8098591549295776e-05, "loss": 0.3561, "step": 257 }, { "epoch": 0.14, "learning_rate": 1.8169014084507044e-05, "loss": 0.1656, "step": 258 }, { "epoch": 0.14, "learning_rate": 1.8239436619718312e-05, "loss": 0.2367, "step": 259 }, { "epoch": 0.14, "learning_rate": 1.830985915492958e-05, "loss": 0.1392, "step": 260 }, { "epoch": 0.14, "learning_rate": 1.8380281690140845e-05, "loss": 0.3206, "step": 261 }, { "epoch": 0.14, "learning_rate": 1.8450704225352113e-05, "loss": 0.2734, "step": 262 }, { "epoch": 0.14, "learning_rate": 1.852112676056338e-05, "loss": 0.389, "step": 263 }, { "epoch": 0.14, "learning_rate": 1.859154929577465e-05, "loss": 0.1136, "step": 264 }, { "epoch": 0.14, "learning_rate": 1.8661971830985917e-05, "loss": 0.1316, "step": 265 }, { "epoch": 0.14, "learning_rate": 1.8732394366197186e-05, "loss": 0.3205, "step": 266 }, { "epoch": 0.14, "learning_rate": 1.880281690140845e-05, "loss": 0.2652, "step": 267 }, { "epoch": 0.14, "learning_rate": 1.887323943661972e-05, "loss": 0.2805, "step": 268 }, { "epoch": 0.14, "learning_rate": 1.8943661971830987e-05, "loss": 0.177, "step": 269 }, { "epoch": 0.14, "learning_rate": 1.9014084507042255e-05, "loss": 0.1931, "step": 270 }, { "epoch": 0.14, "learning_rate": 1.9084507042253523e-05, "loss": 0.3576, "step": 271 }, { "epoch": 0.14, "learning_rate": 1.9154929577464788e-05, "loss": 0.1907, "step": 272 }, { "epoch": 0.14, "learning_rate": 1.922535211267606e-05, "loss": 0.2938, "step": 273 }, { "epoch": 0.15, "learning_rate": 1.9295774647887327e-05, "loss": 0.289, "step": 274 }, { "epoch": 0.15, "learning_rate": 1.9366197183098595e-05, "loss": 0.2156, "step": 275 }, { "epoch": 0.15, "learning_rate": 1.943661971830986e-05, "loss": 0.4307, "step": 276 }, { "epoch": 0.15, "learning_rate": 1.9507042253521128e-05, "loss": 0.1977, "step": 277 }, { "epoch": 0.15, "learning_rate": 1.9577464788732396e-05, "loss": 0.2752, "step": 278 }, { "epoch": 0.15, "learning_rate": 1.9647887323943664e-05, "loss": 0.3232, "step": 279 }, { "epoch": 0.15, "learning_rate": 1.9718309859154933e-05, "loss": 0.3029, "step": 280 }, { "epoch": 0.15, "learning_rate": 1.97887323943662e-05, "loss": 0.2517, "step": 281 }, { "epoch": 0.15, "learning_rate": 1.9859154929577465e-05, "loss": 0.2347, "step": 282 }, { "epoch": 0.15, "learning_rate": 1.9929577464788734e-05, "loss": 0.29, "step": 283 }, { "epoch": 0.15, "learning_rate": 2e-05, "loss": 0.1168, "step": 284 }, { "epoch": 0.15, "learning_rate": 1.9999999411990993e-05, "loss": 0.2454, "step": 285 }, { "epoch": 0.15, "learning_rate": 1.999999764796404e-05, "loss": 0.2419, "step": 286 }, { "epoch": 0.15, "learning_rate": 1.9999994707919344e-05, "loss": 0.1554, "step": 287 }, { "epoch": 0.15, "learning_rate": 1.999999059185726e-05, "loss": 0.262, "step": 288 }, { "epoch": 0.15, "learning_rate": 1.9999985299778262e-05, "loss": 0.3607, "step": 289 }, { "epoch": 0.15, "learning_rate": 1.999997883168298e-05, "loss": 0.2037, "step": 290 }, { "epoch": 0.15, "learning_rate": 1.9999971187572172e-05, "loss": 0.332, "step": 291 }, { "epoch": 0.15, "learning_rate": 1.9999962367446736e-05, "loss": 0.3099, "step": 292 }, { "epoch": 0.16, "learning_rate": 1.9999952371307708e-05, "loss": 0.2294, "step": 293 }, { "epoch": 0.16, "learning_rate": 1.9999941199156264e-05, "loss": 0.1785, "step": 294 }, { "epoch": 0.16, "learning_rate": 1.9999928850993725e-05, "loss": 0.2078, "step": 295 }, { "epoch": 0.16, "learning_rate": 1.9999915326821536e-05, "loss": 0.2526, "step": 296 }, { "epoch": 0.16, "learning_rate": 1.9999900626641288e-05, "loss": 0.3691, "step": 297 }, { "epoch": 0.16, "learning_rate": 1.999988475045471e-05, "loss": 0.2773, "step": 298 }, { "epoch": 0.16, "learning_rate": 1.999986769826367e-05, "loss": 0.2654, "step": 299 }, { "epoch": 0.16, "learning_rate": 1.9999849470070175e-05, "loss": 0.2491, "step": 300 }, { "epoch": 0.16, "learning_rate": 1.9999830065876364e-05, "loss": 0.2324, "step": 301 }, { "epoch": 0.16, "learning_rate": 1.9999809485684524e-05, "loss": 0.362, "step": 302 }, { "epoch": 0.16, "learning_rate": 1.9999787729497072e-05, "loss": 0.1768, "step": 303 }, { "epoch": 0.16, "learning_rate": 1.999976479731657e-05, "loss": 0.206, "step": 304 }, { "epoch": 0.16, "learning_rate": 1.999974068914571e-05, "loss": 0.2639, "step": 305 }, { "epoch": 0.16, "learning_rate": 1.999971540498733e-05, "loss": 0.2274, "step": 306 }, { "epoch": 0.16, "learning_rate": 1.9999688944844404e-05, "loss": 0.2431, "step": 307 }, { "epoch": 0.16, "learning_rate": 1.999966130872004e-05, "loss": 0.1362, "step": 308 }, { "epoch": 0.16, "learning_rate": 1.99996324966175e-05, "loss": 0.1769, "step": 309 }, { "epoch": 0.16, "learning_rate": 1.9999602508540153e-05, "loss": 0.1815, "step": 310 }, { "epoch": 0.16, "learning_rate": 1.999957134449154e-05, "loss": 0.257, "step": 311 }, { "epoch": 0.17, "learning_rate": 1.9999539004475323e-05, "loss": 0.3034, "step": 312 }, { "epoch": 0.17, "learning_rate": 1.9999505488495303e-05, "loss": 0.0926, "step": 313 }, { "epoch": 0.17, "learning_rate": 1.9999470796555423e-05, "loss": 0.2726, "step": 314 }, { "epoch": 0.17, "learning_rate": 1.999943492865976e-05, "loss": 0.1474, "step": 315 }, { "epoch": 0.17, "learning_rate": 1.999939788481254e-05, "loss": 0.2079, "step": 316 }, { "epoch": 0.17, "learning_rate": 1.999935966501811e-05, "loss": 0.2725, "step": 317 }, { "epoch": 0.17, "learning_rate": 1.9999320269280974e-05, "loss": 0.3008, "step": 318 }, { "epoch": 0.17, "learning_rate": 1.9999279697605757e-05, "loss": 0.3026, "step": 319 }, { "epoch": 0.17, "learning_rate": 1.999923794999723e-05, "loss": 0.2887, "step": 320 }, { "epoch": 0.17, "learning_rate": 1.999919502646031e-05, "loss": 0.2222, "step": 321 }, { "epoch": 0.17, "learning_rate": 1.9999150927000037e-05, "loss": 0.2462, "step": 322 }, { "epoch": 0.17, "learning_rate": 1.9999105651621603e-05, "loss": 0.2317, "step": 323 }, { "epoch": 0.17, "learning_rate": 1.999905920033033e-05, "loss": 0.3508, "step": 324 }, { "epoch": 0.17, "learning_rate": 1.999901157313168e-05, "loss": 0.1809, "step": 325 }, { "epoch": 0.17, "learning_rate": 1.9998962770031258e-05, "loss": 0.18, "step": 326 }, { "epoch": 0.17, "learning_rate": 1.9998912791034798e-05, "loss": 0.1345, "step": 327 }, { "epoch": 0.17, "learning_rate": 1.999886163614818e-05, "loss": 0.2936, "step": 328 }, { "epoch": 0.17, "learning_rate": 1.999880930537742e-05, "loss": 0.341, "step": 329 }, { "epoch": 0.17, "learning_rate": 1.9998755798728673e-05, "loss": 0.4943, "step": 330 }, { "epoch": 0.18, "learning_rate": 1.999870111620823e-05, "loss": 0.2253, "step": 331 }, { "epoch": 0.18, "learning_rate": 1.9998645257822523e-05, "loss": 0.2927, "step": 332 }, { "epoch": 0.18, "learning_rate": 1.999858822357812e-05, "loss": 0.2199, "step": 333 }, { "epoch": 0.18, "learning_rate": 1.999853001348173e-05, "loss": 0.3093, "step": 334 }, { "epoch": 0.18, "learning_rate": 1.9998470627540195e-05, "loss": 0.2779, "step": 335 }, { "epoch": 0.18, "learning_rate": 1.9998410065760503e-05, "loss": 0.3728, "step": 336 }, { "epoch": 0.18, "learning_rate": 1.9998348328149775e-05, "loss": 0.1684, "step": 337 }, { "epoch": 0.18, "learning_rate": 1.999828541471527e-05, "loss": 0.1665, "step": 338 }, { "epoch": 0.18, "learning_rate": 1.9998221325464385e-05, "loss": 0.3486, "step": 339 }, { "epoch": 0.18, "learning_rate": 1.9998156060404665e-05, "loss": 0.3234, "step": 340 }, { "epoch": 0.18, "learning_rate": 1.9998089619543775e-05, "loss": 0.259, "step": 341 }, { "epoch": 0.18, "learning_rate": 1.9998022002889535e-05, "loss": 0.1971, "step": 342 }, { "epoch": 0.18, "learning_rate": 1.9997953210449895e-05, "loss": 0.234, "step": 343 }, { "epoch": 0.18, "learning_rate": 1.9997883242232948e-05, "loss": 0.2632, "step": 344 }, { "epoch": 0.18, "learning_rate": 1.9997812098246918e-05, "loss": 0.2022, "step": 345 }, { "epoch": 0.18, "learning_rate": 1.9997739778500172e-05, "loss": 0.2579, "step": 346 }, { "epoch": 0.18, "learning_rate": 1.999766628300122e-05, "loss": 0.2116, "step": 347 }, { "epoch": 0.18, "learning_rate": 1.99975916117587e-05, "loss": 0.2291, "step": 348 }, { "epoch": 0.18, "learning_rate": 1.9997515764781393e-05, "loss": 0.3236, "step": 349 }, { "epoch": 0.19, "learning_rate": 1.9997438742078223e-05, "loss": 0.1856, "step": 350 }, { "epoch": 0.19, "learning_rate": 1.9997360543658244e-05, "loss": 0.3259, "step": 351 }, { "epoch": 0.19, "learning_rate": 1.9997281169530657e-05, "loss": 0.3582, "step": 352 }, { "epoch": 0.19, "learning_rate": 1.999720061970479e-05, "loss": 0.3517, "step": 353 }, { "epoch": 0.19, "learning_rate": 1.9997118894190126e-05, "loss": 0.2094, "step": 354 }, { "epoch": 0.19, "learning_rate": 1.9997035992996264e-05, "loss": 0.2639, "step": 355 }, { "epoch": 0.19, "learning_rate": 1.9996951916132958e-05, "loss": 0.2242, "step": 356 }, { "epoch": 0.19, "learning_rate": 1.99968666636101e-05, "loss": 0.2927, "step": 357 }, { "epoch": 0.19, "learning_rate": 1.999678023543771e-05, "loss": 0.2414, "step": 358 }, { "epoch": 0.19, "learning_rate": 1.9996692631625956e-05, "loss": 0.1805, "step": 359 }, { "epoch": 0.19, "learning_rate": 1.9996603852185136e-05, "loss": 0.2372, "step": 360 }, { "epoch": 0.19, "learning_rate": 1.9996513897125696e-05, "loss": 0.2431, "step": 361 }, { "epoch": 0.19, "learning_rate": 1.9996422766458214e-05, "loss": 0.2567, "step": 362 }, { "epoch": 0.19, "learning_rate": 1.9996330460193402e-05, "loss": 0.2399, "step": 363 }, { "epoch": 0.19, "learning_rate": 1.999623697834212e-05, "loss": 0.3421, "step": 364 }, { "epoch": 0.19, "learning_rate": 1.999614232091536e-05, "loss": 0.302, "step": 365 }, { "epoch": 0.19, "learning_rate": 1.9996046487924255e-05, "loss": 0.1755, "step": 366 }, { "epoch": 0.19, "learning_rate": 1.9995949479380075e-05, "loss": 0.2457, "step": 367 }, { "epoch": 0.19, "learning_rate": 1.9995851295294227e-05, "loss": 0.2446, "step": 368 }, { "epoch": 0.2, "learning_rate": 1.9995751935678257e-05, "loss": 0.2594, "step": 369 }, { "epoch": 0.2, "learning_rate": 1.999565140054385e-05, "loss": 0.312, "step": 370 }, { "epoch": 0.2, "learning_rate": 1.9995549689902836e-05, "loss": 0.2861, "step": 371 }, { "epoch": 0.2, "learning_rate": 1.999544680376717e-05, "loss": 0.3332, "step": 372 }, { "epoch": 0.2, "learning_rate": 1.999534274214895e-05, "loss": 0.2258, "step": 373 }, { "epoch": 0.2, "learning_rate": 1.9995237505060413e-05, "loss": 0.2773, "step": 374 }, { "epoch": 0.2, "learning_rate": 1.9995131092513944e-05, "loss": 0.3378, "step": 375 }, { "epoch": 0.2, "learning_rate": 1.999502350452205e-05, "loss": 0.1891, "step": 376 }, { "epoch": 0.2, "learning_rate": 1.9994914741097385e-05, "loss": 0.22, "step": 377 }, { "epoch": 0.2, "learning_rate": 1.999480480225274e-05, "loss": 0.1847, "step": 378 }, { "epoch": 0.2, "learning_rate": 1.9994693688001042e-05, "loss": 0.2696, "step": 379 }, { "epoch": 0.2, "learning_rate": 1.999458139835536e-05, "loss": 0.2121, "step": 380 }, { "epoch": 0.2, "learning_rate": 1.9994467933328904e-05, "loss": 0.1706, "step": 381 }, { "epoch": 0.2, "learning_rate": 1.9994353292935012e-05, "loss": 0.4377, "step": 382 }, { "epoch": 0.2, "learning_rate": 1.9994237477187165e-05, "loss": 0.2472, "step": 383 }, { "epoch": 0.2, "learning_rate": 1.9994120486098985e-05, "loss": 0.2426, "step": 384 }, { "epoch": 0.2, "learning_rate": 1.999400231968423e-05, "loss": 0.3053, "step": 385 }, { "epoch": 0.2, "learning_rate": 1.99938829779568e-05, "loss": 0.2884, "step": 386 }, { "epoch": 0.2, "learning_rate": 1.9993762460930726e-05, "loss": 0.2118, "step": 387 }, { "epoch": 0.21, "learning_rate": 1.9993640768620184e-05, "loss": 0.2006, "step": 388 }, { "epoch": 0.21, "learning_rate": 1.999351790103948e-05, "loss": 0.2924, "step": 389 }, { "epoch": 0.21, "learning_rate": 1.999339385820307e-05, "loss": 0.1175, "step": 390 }, { "epoch": 0.21, "learning_rate": 1.9993268640125536e-05, "loss": 0.3781, "step": 391 }, { "epoch": 0.21, "learning_rate": 1.9993142246821608e-05, "loss": 0.3718, "step": 392 }, { "epoch": 0.21, "learning_rate": 1.9993014678306147e-05, "loss": 0.2057, "step": 393 }, { "epoch": 0.21, "learning_rate": 1.999288593459416e-05, "loss": 0.2265, "step": 394 }, { "epoch": 0.21, "learning_rate": 1.999275601570078e-05, "loss": 0.241, "step": 395 }, { "epoch": 0.21, "learning_rate": 1.999262492164129e-05, "loss": 0.2133, "step": 396 }, { "epoch": 0.21, "learning_rate": 1.999249265243111e-05, "loss": 0.1932, "step": 397 }, { "epoch": 0.21, "learning_rate": 1.9992359208085793e-05, "loss": 0.3148, "step": 398 }, { "epoch": 0.21, "learning_rate": 1.9992224588621027e-05, "loss": 0.1627, "step": 399 }, { "epoch": 0.21, "learning_rate": 1.999208879405265e-05, "loss": 0.2231, "step": 400 }, { "epoch": 0.21, "learning_rate": 1.999195182439663e-05, "loss": 0.2576, "step": 401 }, { "epoch": 0.21, "learning_rate": 1.9991813679669072e-05, "loss": 0.2693, "step": 402 }, { "epoch": 0.21, "learning_rate": 1.9991674359886226e-05, "loss": 0.1994, "step": 403 }, { "epoch": 0.21, "learning_rate": 1.9991533865064473e-05, "loss": 0.1777, "step": 404 }, { "epoch": 0.21, "learning_rate": 1.999139219522034e-05, "loss": 0.2621, "step": 405 }, { "epoch": 0.21, "learning_rate": 1.999124935037048e-05, "loss": 0.3486, "step": 406 }, { "epoch": 0.22, "learning_rate": 1.99911053305317e-05, "loss": 0.3706, "step": 407 }, { "epoch": 0.22, "learning_rate": 1.9990960135720932e-05, "loss": 0.1464, "step": 408 }, { "epoch": 0.22, "learning_rate": 1.9990813765955256e-05, "loss": 0.2552, "step": 409 }, { "epoch": 0.22, "learning_rate": 1.9990666221251875e-05, "loss": 0.1483, "step": 410 }, { "epoch": 0.22, "learning_rate": 1.9990517501628152e-05, "loss": 0.3424, "step": 411 }, { "epoch": 0.22, "learning_rate": 1.9990367607101572e-05, "loss": 0.3158, "step": 412 }, { "epoch": 0.22, "learning_rate": 1.9990216537689763e-05, "loss": 0.2016, "step": 413 }, { "epoch": 0.22, "learning_rate": 1.999006429341049e-05, "loss": 0.1819, "step": 414 }, { "epoch": 0.22, "learning_rate": 1.9989910874281662e-05, "loss": 0.1576, "step": 415 }, { "epoch": 0.22, "learning_rate": 1.9989756280321315e-05, "loss": 0.3027, "step": 416 }, { "epoch": 0.22, "learning_rate": 1.998960051154763e-05, "loss": 0.1799, "step": 417 }, { "epoch": 0.22, "learning_rate": 1.9989443567978932e-05, "loss": 0.2549, "step": 418 }, { "epoch": 0.22, "learning_rate": 1.998928544963367e-05, "loss": 0.2166, "step": 419 }, { "epoch": 0.22, "learning_rate": 1.9989126156530444e-05, "loss": 0.24, "step": 420 }, { "epoch": 0.22, "learning_rate": 1.9988965688687987e-05, "loss": 0.3242, "step": 421 }, { "epoch": 0.22, "learning_rate": 1.998880404612517e-05, "loss": 0.2642, "step": 422 }, { "epoch": 0.22, "learning_rate": 1.9988641228861e-05, "loss": 0.2312, "step": 423 }, { "epoch": 0.22, "learning_rate": 1.9988477236914625e-05, "loss": 0.2722, "step": 424 }, { "epoch": 0.22, "learning_rate": 1.9988312070305336e-05, "loss": 0.3362, "step": 425 }, { "epoch": 0.23, "learning_rate": 1.998814572905255e-05, "loss": 0.1802, "step": 426 }, { "epoch": 0.23, "learning_rate": 1.998797821317583e-05, "loss": 0.1532, "step": 427 }, { "epoch": 0.23, "learning_rate": 1.9987809522694884e-05, "loss": 0.3054, "step": 428 }, { "epoch": 0.23, "learning_rate": 1.998763965762954e-05, "loss": 0.2366, "step": 429 }, { "epoch": 0.23, "learning_rate": 1.998746861799978e-05, "loss": 0.2022, "step": 430 }, { "epoch": 0.23, "learning_rate": 1.9987296403825715e-05, "loss": 0.2095, "step": 431 }, { "epoch": 0.23, "learning_rate": 1.9987123015127602e-05, "loss": 0.2762, "step": 432 }, { "epoch": 0.23, "learning_rate": 1.998694845192583e-05, "loss": 0.2508, "step": 433 }, { "epoch": 0.23, "learning_rate": 1.998677271424093e-05, "loss": 0.2044, "step": 434 }, { "epoch": 0.23, "learning_rate": 1.9986595802093566e-05, "loss": 0.257, "step": 435 }, { "epoch": 0.23, "learning_rate": 1.9986417715504542e-05, "loss": 0.2175, "step": 436 }, { "epoch": 0.23, "learning_rate": 1.9986238454494805e-05, "loss": 0.2883, "step": 437 }, { "epoch": 0.23, "learning_rate": 1.998605801908543e-05, "loss": 0.2902, "step": 438 }, { "epoch": 0.23, "learning_rate": 1.998587640929765e-05, "loss": 0.2407, "step": 439 }, { "epoch": 0.23, "learning_rate": 1.9985693625152807e-05, "loss": 0.2437, "step": 440 }, { "epoch": 0.23, "learning_rate": 1.998550966667241e-05, "loss": 0.2797, "step": 441 }, { "epoch": 0.23, "learning_rate": 1.9985324533878082e-05, "loss": 0.2519, "step": 442 }, { "epoch": 0.23, "learning_rate": 1.99851382267916e-05, "loss": 0.3, "step": 443 }, { "epoch": 0.24, "learning_rate": 1.9984950745434875e-05, "loss": 0.3161, "step": 444 }, { "epoch": 0.24, "learning_rate": 1.9984762089829953e-05, "loss": 0.2461, "step": 445 }, { "epoch": 0.24, "learning_rate": 1.998457225999902e-05, "loss": 0.2453, "step": 446 }, { "epoch": 0.24, "learning_rate": 1.99843812559644e-05, "loss": 0.3031, "step": 447 }, { "epoch": 0.24, "learning_rate": 1.9984189077748562e-05, "loss": 0.2504, "step": 448 }, { "epoch": 0.24, "learning_rate": 1.9983995725374096e-05, "loss": 0.3365, "step": 449 }, { "epoch": 0.24, "learning_rate": 1.9983801198863752e-05, "loss": 0.1846, "step": 450 }, { "epoch": 0.24, "learning_rate": 1.9983605498240395e-05, "loss": 0.1735, "step": 451 }, { "epoch": 0.24, "learning_rate": 1.998340862352705e-05, "loss": 0.3807, "step": 452 }, { "epoch": 0.24, "learning_rate": 1.9983210574746857e-05, "loss": 0.3203, "step": 453 }, { "epoch": 0.24, "learning_rate": 1.9983011351923123e-05, "loss": 0.3364, "step": 454 }, { "epoch": 0.24, "learning_rate": 1.9982810955079266e-05, "loss": 0.4128, "step": 455 }, { "epoch": 0.24, "learning_rate": 1.9982609384238857e-05, "loss": 0.1329, "step": 456 }, { "epoch": 0.24, "learning_rate": 1.99824066394256e-05, "loss": 0.2151, "step": 457 }, { "epoch": 0.24, "learning_rate": 1.998220272066333e-05, "loss": 0.3818, "step": 458 }, { "epoch": 0.24, "learning_rate": 1.9981997627976046e-05, "loss": 0.2789, "step": 459 }, { "epoch": 0.24, "learning_rate": 1.9981791361387858e-05, "loss": 0.2509, "step": 460 }, { "epoch": 0.24, "learning_rate": 1.9981583920923018e-05, "loss": 0.2289, "step": 461 }, { "epoch": 0.24, "learning_rate": 1.9981375306605927e-05, "loss": 0.3319, "step": 462 }, { "epoch": 0.25, "learning_rate": 1.998116551846112e-05, "loss": 0.2209, "step": 463 }, { "epoch": 0.25, "learning_rate": 1.998095455651327e-05, "loss": 0.201, "step": 464 }, { "epoch": 0.25, "learning_rate": 1.9980742420787176e-05, "loss": 0.1265, "step": 465 }, { "epoch": 0.25, "learning_rate": 1.9980529111307792e-05, "loss": 0.2153, "step": 466 }, { "epoch": 0.25, "learning_rate": 1.9980314628100207e-05, "loss": 0.3453, "step": 467 }, { "epoch": 0.25, "learning_rate": 1.9980098971189644e-05, "loss": 0.1681, "step": 468 }, { "epoch": 0.25, "learning_rate": 1.9979882140601458e-05, "loss": 0.4614, "step": 469 }, { "epoch": 0.25, "learning_rate": 1.9979664136361153e-05, "loss": 0.1611, "step": 470 }, { "epoch": 0.25, "learning_rate": 1.997944495849437e-05, "loss": 0.2278, "step": 471 }, { "epoch": 0.25, "learning_rate": 1.9979224607026876e-05, "loss": 0.2224, "step": 472 }, { "epoch": 0.25, "learning_rate": 1.9979003081984595e-05, "loss": 0.2981, "step": 473 }, { "epoch": 0.25, "learning_rate": 1.997878038339357e-05, "loss": 0.2581, "step": 474 }, { "epoch": 0.25, "learning_rate": 1.9978556511279995e-05, "loss": 0.1634, "step": 475 }, { "epoch": 0.25, "learning_rate": 1.99783314656702e-05, "loss": 0.1611, "step": 476 }, { "epoch": 0.25, "learning_rate": 1.9978105246590645e-05, "loss": 0.2558, "step": 477 }, { "epoch": 0.25, "learning_rate": 1.997787785406794e-05, "loss": 0.4443, "step": 478 }, { "epoch": 0.25, "learning_rate": 1.997764928812882e-05, "loss": 0.3482, "step": 479 }, { "epoch": 0.25, "learning_rate": 1.997741954880017e-05, "loss": 0.2938, "step": 480 }, { "epoch": 0.25, "learning_rate": 1.9977188636109007e-05, "loss": 0.2157, "step": 481 }, { "epoch": 0.26, "learning_rate": 1.9976956550082486e-05, "loss": 0.275, "step": 482 }, { "epoch": 0.26, "learning_rate": 1.9976723290747895e-05, "loss": 0.1706, "step": 483 }, { "epoch": 0.26, "learning_rate": 1.9976488858132675e-05, "loss": 0.1644, "step": 484 }, { "epoch": 0.26, "learning_rate": 1.9976253252264394e-05, "loss": 0.3645, "step": 485 }, { "epoch": 0.26, "learning_rate": 1.9976016473170755e-05, "loss": 0.2113, "step": 486 }, { "epoch": 0.26, "learning_rate": 1.997577852087961e-05, "loss": 0.261, "step": 487 }, { "epoch": 0.26, "learning_rate": 1.9975539395418936e-05, "loss": 0.1485, "step": 488 }, { "epoch": 0.26, "learning_rate": 1.9975299096816857e-05, "loss": 0.2286, "step": 489 }, { "epoch": 0.26, "learning_rate": 1.997505762510163e-05, "loss": 0.1943, "step": 490 }, { "epoch": 0.26, "learning_rate": 1.997481498030166e-05, "loss": 0.3522, "step": 491 }, { "epoch": 0.26, "learning_rate": 1.9974571162445477e-05, "loss": 0.2328, "step": 492 }, { "epoch": 0.26, "learning_rate": 1.9974326171561757e-05, "loss": 0.2486, "step": 493 }, { "epoch": 0.26, "learning_rate": 1.9974080007679307e-05, "loss": 0.261, "step": 494 }, { "epoch": 0.26, "learning_rate": 1.997383267082708e-05, "loss": 0.1919, "step": 495 }, { "epoch": 0.26, "learning_rate": 1.9973584161034164e-05, "loss": 0.3078, "step": 496 }, { "epoch": 0.26, "learning_rate": 1.9973334478329782e-05, "loss": 0.183, "step": 497 }, { "epoch": 0.26, "learning_rate": 1.9973083622743295e-05, "loss": 0.2306, "step": 498 }, { "epoch": 0.26, "learning_rate": 1.997283159430421e-05, "loss": 0.2703, "step": 499 }, { "epoch": 0.26, "learning_rate": 1.9972578393042162e-05, "loss": 0.2295, "step": 500 }, { "epoch": 0.27, "learning_rate": 1.997232401898693e-05, "loss": 0.1347, "step": 501 }, { "epoch": 0.27, "learning_rate": 1.9972068472168424e-05, "loss": 0.4344, "step": 502 }, { "epoch": 0.27, "learning_rate": 1.99718117526167e-05, "loss": 0.1095, "step": 503 }, { "epoch": 0.27, "learning_rate": 1.9971553860361954e-05, "loss": 0.2552, "step": 504 }, { "epoch": 0.27, "learning_rate": 1.9971294795434505e-05, "loss": 0.2313, "step": 505 }, { "epoch": 0.27, "learning_rate": 1.9971034557864825e-05, "loss": 0.2918, "step": 506 }, { "epoch": 0.27, "learning_rate": 1.997077314768352e-05, "loss": 0.2088, "step": 507 }, { "epoch": 0.27, "learning_rate": 1.9970510564921328e-05, "loss": 0.2522, "step": 508 }, { "epoch": 0.27, "learning_rate": 1.9970246809609127e-05, "loss": 0.3004, "step": 509 }, { "epoch": 0.27, "learning_rate": 1.9969981881777946e-05, "loss": 0.1378, "step": 510 }, { "epoch": 0.27, "learning_rate": 1.996971578145893e-05, "loss": 0.2116, "step": 511 }, { "epoch": 0.27, "learning_rate": 1.9969448508683377e-05, "loss": 0.1752, "step": 512 }, { "epoch": 0.27, "learning_rate": 1.996918006348272e-05, "loss": 0.1693, "step": 513 }, { "epoch": 0.27, "learning_rate": 1.996891044588853e-05, "loss": 0.2801, "step": 514 }, { "epoch": 0.27, "learning_rate": 1.9968639655932506e-05, "loss": 0.2676, "step": 515 }, { "epoch": 0.27, "learning_rate": 1.9968367693646505e-05, "loss": 0.1486, "step": 516 }, { "epoch": 0.27, "learning_rate": 1.99680945590625e-05, "loss": 0.2472, "step": 517 }, { "epoch": 0.27, "learning_rate": 1.996782025221262e-05, "loss": 0.2849, "step": 518 }, { "epoch": 0.27, "learning_rate": 1.9967544773129118e-05, "loss": 0.3063, "step": 519 }, { "epoch": 0.28, "learning_rate": 1.9967268121844395e-05, "loss": 0.1517, "step": 520 }, { "epoch": 0.28, "learning_rate": 1.9966990298390984e-05, "loss": 0.3536, "step": 521 }, { "epoch": 0.28, "learning_rate": 1.996671130280156e-05, "loss": 0.2268, "step": 522 }, { "epoch": 0.28, "learning_rate": 1.9966431135108925e-05, "loss": 0.3127, "step": 523 }, { "epoch": 0.28, "learning_rate": 1.9966149795346037e-05, "loss": 0.337, "step": 524 }, { "epoch": 0.28, "learning_rate": 1.9965867283545978e-05, "loss": 0.246, "step": 525 }, { "epoch": 0.28, "learning_rate": 1.9965583599741974e-05, "loss": 0.2052, "step": 526 }, { "epoch": 0.28, "learning_rate": 1.9965298743967382e-05, "loss": 0.1672, "step": 527 }, { "epoch": 0.28, "learning_rate": 1.9965012716255704e-05, "loss": 0.2795, "step": 528 }, { "epoch": 0.28, "learning_rate": 1.996472551664058e-05, "loss": 0.113, "step": 529 }, { "epoch": 0.28, "learning_rate": 1.9964437145155785e-05, "loss": 0.2437, "step": 530 }, { "epoch": 0.28, "learning_rate": 1.9964147601835228e-05, "loss": 0.1965, "step": 531 }, { "epoch": 0.28, "learning_rate": 1.996385688671296e-05, "loss": 0.3071, "step": 532 }, { "epoch": 0.28, "learning_rate": 1.9963564999823174e-05, "loss": 0.328, "step": 533 }, { "epoch": 0.28, "learning_rate": 1.996327194120019e-05, "loss": 0.2102, "step": 534 }, { "epoch": 0.28, "learning_rate": 1.9962977710878485e-05, "loss": 0.2243, "step": 535 }, { "epoch": 0.28, "learning_rate": 1.9962682308892643e-05, "loss": 0.3501, "step": 536 }, { "epoch": 0.28, "learning_rate": 1.996238573527742e-05, "loss": 0.194, "step": 537 }, { "epoch": 0.28, "learning_rate": 1.9962087990067682e-05, "loss": 0.3538, "step": 538 }, { "epoch": 0.29, "learning_rate": 1.9961789073298453e-05, "loss": 0.2567, "step": 539 }, { "epoch": 0.29, "learning_rate": 1.996148898500488e-05, "loss": 0.3082, "step": 540 }, { "epoch": 0.29, "learning_rate": 1.9961187725222256e-05, "loss": 0.1733, "step": 541 }, { "epoch": 0.29, "learning_rate": 1.9960885293986006e-05, "loss": 0.2668, "step": 542 }, { "epoch": 0.29, "learning_rate": 1.9960581691331705e-05, "loss": 0.135, "step": 543 }, { "epoch": 0.29, "learning_rate": 1.9960276917295052e-05, "loss": 0.4188, "step": 544 }, { "epoch": 0.29, "learning_rate": 1.9959970971911885e-05, "loss": 0.1251, "step": 545 }, { "epoch": 0.29, "learning_rate": 1.9959663855218193e-05, "loss": 0.2158, "step": 546 }, { "epoch": 0.29, "learning_rate": 1.995935556725009e-05, "loss": 0.2193, "step": 547 }, { "epoch": 0.29, "learning_rate": 1.9959046108043822e-05, "loss": 0.2622, "step": 548 }, { "epoch": 0.29, "learning_rate": 1.9958735477635795e-05, "loss": 0.1161, "step": 549 }, { "epoch": 0.29, "learning_rate": 1.9958423676062537e-05, "loss": 0.2069, "step": 550 }, { "epoch": 0.29, "learning_rate": 1.995811070336071e-05, "loss": 0.2278, "step": 551 }, { "epoch": 0.29, "learning_rate": 1.9957796559567125e-05, "loss": 0.2833, "step": 552 }, { "epoch": 0.29, "learning_rate": 1.9957481244718723e-05, "loss": 0.1651, "step": 553 }, { "epoch": 0.29, "learning_rate": 1.9957164758852593e-05, "loss": 0.2054, "step": 554 }, { "epoch": 0.29, "learning_rate": 1.9956847102005942e-05, "loss": 0.2573, "step": 555 }, { "epoch": 0.29, "learning_rate": 1.9956528274216137e-05, "loss": 0.179, "step": 556 }, { "epoch": 0.29, "learning_rate": 1.995620827552067e-05, "loss": 0.1858, "step": 557 }, { "epoch": 0.3, "learning_rate": 1.9955887105957175e-05, "loss": 0.293, "step": 558 }, { "epoch": 0.3, "learning_rate": 1.995556476556342e-05, "loss": 0.4158, "step": 559 }, { "epoch": 0.3, "learning_rate": 1.995524125437731e-05, "loss": 0.2615, "step": 560 }, { "epoch": 0.3, "learning_rate": 1.995491657243689e-05, "loss": 0.17, "step": 561 }, { "epoch": 0.3, "learning_rate": 1.9954590719780355e-05, "loss": 0.2396, "step": 562 }, { "epoch": 0.3, "learning_rate": 1.9954263696446012e-05, "loss": 0.1827, "step": 563 }, { "epoch": 0.3, "learning_rate": 1.995393550247233e-05, "loss": 0.233, "step": 564 }, { "epoch": 0.3, "learning_rate": 1.9953606137897893e-05, "loss": 0.2635, "step": 565 }, { "epoch": 0.3, "learning_rate": 1.9953275602761445e-05, "loss": 0.2609, "step": 566 }, { "epoch": 0.3, "learning_rate": 1.9952943897101855e-05, "loss": 0.2655, "step": 567 }, { "epoch": 0.3, "learning_rate": 1.9952611020958136e-05, "loss": 0.1995, "step": 568 }, { "epoch": 0.3, "learning_rate": 1.9952276974369426e-05, "loss": 0.2869, "step": 569 }, { "epoch": 0.3, "learning_rate": 1.9951941757375015e-05, "loss": 0.1026, "step": 570 }, { "epoch": 0.3, "learning_rate": 1.9951605370014323e-05, "loss": 0.2273, "step": 571 }, { "epoch": 0.3, "learning_rate": 1.995126781232691e-05, "loss": 0.3283, "step": 572 }, { "epoch": 0.3, "learning_rate": 1.9950929084352476e-05, "loss": 0.1861, "step": 573 }, { "epoch": 0.3, "learning_rate": 1.9950589186130857e-05, "loss": 0.2848, "step": 574 }, { "epoch": 0.3, "learning_rate": 1.9950248117702017e-05, "loss": 0.2382, "step": 575 }, { "epoch": 0.3, "learning_rate": 1.994990587910608e-05, "loss": 0.2179, "step": 576 }, { "epoch": 0.31, "learning_rate": 1.994956247038328e-05, "loss": 0.1382, "step": 577 }, { "epoch": 0.31, "learning_rate": 1.994921789157401e-05, "loss": 0.3586, "step": 578 }, { "epoch": 0.31, "learning_rate": 1.9948872142718795e-05, "loss": 0.2055, "step": 579 }, { "epoch": 0.31, "learning_rate": 1.9948525223858288e-05, "loss": 0.1407, "step": 580 }, { "epoch": 0.31, "learning_rate": 1.9948177135033295e-05, "loss": 0.2875, "step": 581 }, { "epoch": 0.31, "learning_rate": 1.9947827876284745e-05, "loss": 0.1412, "step": 582 }, { "epoch": 0.31, "learning_rate": 1.994747744765372e-05, "loss": 0.1601, "step": 583 }, { "epoch": 0.31, "learning_rate": 1.9947125849181422e-05, "loss": 0.1432, "step": 584 }, { "epoch": 0.31, "learning_rate": 1.9946773080909207e-05, "loss": 0.2344, "step": 585 }, { "epoch": 0.31, "learning_rate": 1.9946419142878556e-05, "loss": 0.2231, "step": 586 }, { "epoch": 0.31, "learning_rate": 1.9946064035131095e-05, "loss": 0.2926, "step": 587 }, { "epoch": 0.31, "learning_rate": 1.9945707757708583e-05, "loss": 0.1906, "step": 588 }, { "epoch": 0.31, "learning_rate": 1.9945350310652922e-05, "loss": 0.2445, "step": 589 }, { "epoch": 0.31, "learning_rate": 1.9944991694006147e-05, "loss": 0.2385, "step": 590 }, { "epoch": 0.31, "learning_rate": 1.9944631907810435e-05, "loss": 0.1737, "step": 591 }, { "epoch": 0.31, "learning_rate": 1.9944270952108092e-05, "loss": 0.1244, "step": 592 }, { "epoch": 0.31, "learning_rate": 1.9943908826941572e-05, "loss": 0.2167, "step": 593 }, { "epoch": 0.31, "learning_rate": 1.9943545532353458e-05, "loss": 0.3113, "step": 594 }, { "epoch": 0.31, "learning_rate": 1.9943181068386473e-05, "loss": 0.153, "step": 595 }, { "epoch": 0.32, "learning_rate": 1.9942815435083484e-05, "loss": 0.2814, "step": 596 }, { "epoch": 0.32, "learning_rate": 1.9942448632487486e-05, "loss": 0.2517, "step": 597 }, { "epoch": 0.32, "learning_rate": 1.9942080660641615e-05, "loss": 0.2224, "step": 598 }, { "epoch": 0.32, "learning_rate": 1.9941711519589152e-05, "loss": 0.1844, "step": 599 }, { "epoch": 0.32, "learning_rate": 1.99413412093735e-05, "loss": 0.3252, "step": 600 }, { "epoch": 0.32, "learning_rate": 1.994096973003821e-05, "loss": 0.2876, "step": 601 }, { "epoch": 0.32, "learning_rate": 1.9940597081626975e-05, "loss": 0.1736, "step": 602 }, { "epoch": 0.32, "learning_rate": 1.9940223264183612e-05, "loss": 0.3604, "step": 603 }, { "epoch": 0.32, "learning_rate": 1.9939848277752085e-05, "loss": 0.1095, "step": 604 }, { "epoch": 0.32, "learning_rate": 1.9939472122376492e-05, "loss": 0.3865, "step": 605 }, { "epoch": 0.32, "learning_rate": 1.9939094798101074e-05, "loss": 0.1472, "step": 606 }, { "epoch": 0.32, "learning_rate": 1.99387163049702e-05, "loss": 0.3937, "step": 607 }, { "epoch": 0.32, "learning_rate": 1.9938336643028383e-05, "loss": 0.2002, "step": 608 }, { "epoch": 0.32, "learning_rate": 1.9937955812320275e-05, "loss": 0.2333, "step": 609 }, { "epoch": 0.32, "learning_rate": 1.9937573812890655e-05, "loss": 0.189, "step": 610 }, { "epoch": 0.32, "learning_rate": 1.9937190644784453e-05, "loss": 0.1206, "step": 611 }, { "epoch": 0.32, "learning_rate": 1.993680630804673e-05, "loss": 0.3613, "step": 612 }, { "epoch": 0.32, "learning_rate": 1.993642080272268e-05, "loss": 0.2806, "step": 613 }, { "epoch": 0.32, "learning_rate": 1.9936034128857645e-05, "loss": 0.229, "step": 614 }, { "epoch": 0.33, "learning_rate": 1.99356462864971e-05, "loss": 0.2066, "step": 615 }, { "epoch": 0.33, "learning_rate": 1.9935257275686647e-05, "loss": 0.2905, "step": 616 }, { "epoch": 0.33, "learning_rate": 1.993486709647204e-05, "loss": 0.1671, "step": 617 }, { "epoch": 0.33, "learning_rate": 1.9934475748899164e-05, "loss": 0.1844, "step": 618 }, { "epoch": 0.33, "learning_rate": 1.993408323301404e-05, "loss": 0.2943, "step": 619 }, { "epoch": 0.33, "learning_rate": 1.9933689548862837e-05, "loss": 0.1762, "step": 620 }, { "epoch": 0.33, "learning_rate": 1.9933294696491843e-05, "loss": 0.2095, "step": 621 }, { "epoch": 0.33, "learning_rate": 1.9932898675947496e-05, "loss": 0.3674, "step": 622 }, { "epoch": 0.33, "learning_rate": 1.9932501487276373e-05, "loss": 0.2781, "step": 623 }, { "epoch": 0.33, "learning_rate": 1.993210313052518e-05, "loss": 0.1451, "step": 624 }, { "epoch": 0.33, "learning_rate": 1.993170360574076e-05, "loss": 0.2386, "step": 625 }, { "epoch": 0.33, "learning_rate": 1.993130291297011e-05, "loss": 0.1373, "step": 626 }, { "epoch": 0.33, "learning_rate": 1.9930901052260342e-05, "loss": 0.2261, "step": 627 }, { "epoch": 0.33, "learning_rate": 1.993049802365872e-05, "loss": 0.1439, "step": 628 }, { "epoch": 0.33, "learning_rate": 1.9930093827212646e-05, "loss": 0.2097, "step": 629 }, { "epoch": 0.33, "learning_rate": 1.992968846296964e-05, "loss": 0.2284, "step": 630 }, { "epoch": 0.33, "learning_rate": 1.9929281930977386e-05, "loss": 0.235, "step": 631 }, { "epoch": 0.33, "learning_rate": 1.9928874231283685e-05, "loss": 0.2238, "step": 632 }, { "epoch": 0.34, "learning_rate": 1.992846536393649e-05, "loss": 0.1817, "step": 633 }, { "epoch": 0.34, "learning_rate": 1.992805532898388e-05, "loss": 0.3337, "step": 634 }, { "epoch": 0.34, "learning_rate": 1.9927644126474076e-05, "loss": 0.1553, "step": 635 }, { "epoch": 0.34, "learning_rate": 1.992723175645544e-05, "loss": 0.1621, "step": 636 }, { "epoch": 0.34, "learning_rate": 1.9926818218976465e-05, "loss": 0.1426, "step": 637 }, { "epoch": 0.34, "learning_rate": 1.992640351408578e-05, "loss": 0.239, "step": 638 }, { "epoch": 0.34, "learning_rate": 1.992598764183216e-05, "loss": 0.2589, "step": 639 }, { "epoch": 0.34, "learning_rate": 1.9925570602264518e-05, "loss": 0.1584, "step": 640 }, { "epoch": 0.34, "learning_rate": 1.9925152395431882e-05, "loss": 0.1688, "step": 641 }, { "epoch": 0.34, "learning_rate": 1.9924733021383448e-05, "loss": 0.2596, "step": 642 }, { "epoch": 0.34, "learning_rate": 1.992431248016853e-05, "loss": 0.1687, "step": 643 }, { "epoch": 0.34, "learning_rate": 1.9923890771836584e-05, "loss": 0.2695, "step": 644 }, { "epoch": 0.34, "learning_rate": 1.9923467896437204e-05, "loss": 0.1995, "step": 645 }, { "epoch": 0.34, "learning_rate": 1.9923043854020125e-05, "loss": 0.2822, "step": 646 }, { "epoch": 0.34, "learning_rate": 1.992261864463521e-05, "loss": 0.2606, "step": 647 }, { "epoch": 0.34, "learning_rate": 1.9922192268332464e-05, "loss": 0.2154, "step": 648 }, { "epoch": 0.34, "learning_rate": 1.9921764725162034e-05, "loss": 0.2106, "step": 649 }, { "epoch": 0.34, "learning_rate": 1.9921336015174196e-05, "loss": 0.357, "step": 650 }, { "epoch": 0.34, "learning_rate": 1.992090613841937e-05, "loss": 0.1654, "step": 651 }, { "epoch": 0.35, "learning_rate": 1.992047509494811e-05, "loss": 0.2548, "step": 652 }, { "epoch": 0.35, "learning_rate": 1.9920042884811105e-05, "loss": 0.1341, "step": 653 }, { "epoch": 0.35, "learning_rate": 1.9919609508059184e-05, "loss": 0.234, "step": 654 }, { "epoch": 0.35, "learning_rate": 1.9919174964743316e-05, "loss": 0.2678, "step": 655 }, { "epoch": 0.35, "learning_rate": 1.9918739254914602e-05, "loss": 0.2128, "step": 656 }, { "epoch": 0.35, "learning_rate": 1.991830237862428e-05, "loss": 0.1537, "step": 657 }, { "epoch": 0.35, "learning_rate": 1.9917864335923732e-05, "loss": 0.2586, "step": 658 }, { "epoch": 0.35, "learning_rate": 1.9917425126864467e-05, "loss": 0.1207, "step": 659 }, { "epoch": 0.35, "learning_rate": 1.9916984751498143e-05, "loss": 0.2626, "step": 660 }, { "epoch": 0.35, "learning_rate": 1.9916543209876545e-05, "loss": 0.2261, "step": 661 }, { "epoch": 0.35, "learning_rate": 1.9916100502051604e-05, "loss": 0.2174, "step": 662 }, { "epoch": 0.35, "learning_rate": 1.9915656628075374e-05, "loss": 0.1861, "step": 663 }, { "epoch": 0.35, "learning_rate": 1.9915211588000062e-05, "loss": 0.3252, "step": 664 }, { "epoch": 0.35, "learning_rate": 1.9914765381878006e-05, "loss": 0.3271, "step": 665 }, { "epoch": 0.35, "learning_rate": 1.991431800976168e-05, "loss": 0.2647, "step": 666 }, { "epoch": 0.35, "learning_rate": 1.9913869471703695e-05, "loss": 0.2541, "step": 667 }, { "epoch": 0.35, "learning_rate": 1.9913419767756796e-05, "loss": 0.2386, "step": 668 }, { "epoch": 0.35, "learning_rate": 1.9912968897973874e-05, "loss": 0.2756, "step": 669 }, { "epoch": 0.35, "learning_rate": 1.991251686240795e-05, "loss": 0.27, "step": 670 }, { "epoch": 0.36, "learning_rate": 1.9912063661112188e-05, "loss": 0.2722, "step": 671 }, { "epoch": 0.36, "learning_rate": 1.991160929413988e-05, "loss": 0.1791, "step": 672 }, { "epoch": 0.36, "learning_rate": 1.991115376154446e-05, "loss": 0.1848, "step": 673 }, { "epoch": 0.36, "learning_rate": 1.991069706337951e-05, "loss": 0.2836, "step": 674 }, { "epoch": 0.36, "learning_rate": 1.991023919969872e-05, "loss": 0.2531, "step": 675 }, { "epoch": 0.36, "learning_rate": 1.9909780170555955e-05, "loss": 0.2832, "step": 676 }, { "epoch": 0.36, "learning_rate": 1.9909319976005184e-05, "loss": 0.1364, "step": 677 }, { "epoch": 0.36, "learning_rate": 1.9908858616100534e-05, "loss": 0.1761, "step": 678 }, { "epoch": 0.36, "learning_rate": 1.9908396090896257e-05, "loss": 0.257, "step": 679 }, { "epoch": 0.36, "learning_rate": 1.990793240044675e-05, "loss": 0.2609, "step": 680 }, { "epoch": 0.36, "learning_rate": 1.9907467544806542e-05, "loss": 0.2986, "step": 681 }, { "epoch": 0.36, "learning_rate": 1.9907001524030303e-05, "loss": 0.268, "step": 682 }, { "epoch": 0.36, "learning_rate": 1.9906534338172835e-05, "loss": 0.3071, "step": 683 }, { "epoch": 0.36, "learning_rate": 1.9906065987289084e-05, "loss": 0.2191, "step": 684 }, { "epoch": 0.36, "learning_rate": 1.9905596471434124e-05, "loss": 0.1482, "step": 685 }, { "epoch": 0.36, "learning_rate": 1.9905125790663174e-05, "loss": 0.3271, "step": 686 }, { "epoch": 0.36, "learning_rate": 1.9904653945031587e-05, "loss": 0.2501, "step": 687 }, { "epoch": 0.36, "learning_rate": 1.9904180934594853e-05, "loss": 0.2252, "step": 688 }, { "epoch": 0.36, "learning_rate": 1.99037067594086e-05, "loss": 0.2543, "step": 689 }, { "epoch": 0.37, "learning_rate": 1.9903231419528582e-05, "loss": 0.1956, "step": 690 }, { "epoch": 0.37, "learning_rate": 1.9902754915010715e-05, "loss": 0.2151, "step": 691 }, { "epoch": 0.37, "learning_rate": 1.9902277245911027e-05, "loss": 0.1808, "step": 692 }, { "epoch": 0.37, "learning_rate": 1.9901798412285695e-05, "loss": 0.176, "step": 693 }, { "epoch": 0.37, "learning_rate": 1.9901318414191032e-05, "loss": 0.2241, "step": 694 }, { "epoch": 0.37, "learning_rate": 1.9900837251683486e-05, "loss": 0.2601, "step": 695 }, { "epoch": 0.37, "learning_rate": 1.990035492481964e-05, "loss": 0.3599, "step": 696 }, { "epoch": 0.37, "learning_rate": 1.989987143365622e-05, "loss": 0.2882, "step": 697 }, { "epoch": 0.37, "learning_rate": 1.9899386778250084e-05, "loss": 0.1632, "step": 698 }, { "epoch": 0.37, "learning_rate": 1.9898900958658226e-05, "loss": 0.1783, "step": 699 }, { "epoch": 0.37, "learning_rate": 1.9898413974937783e-05, "loss": 0.3442, "step": 700 }, { "epoch": 0.37, "learning_rate": 1.9897925827146026e-05, "loss": 0.2605, "step": 701 }, { "epoch": 0.37, "learning_rate": 1.989743651534036e-05, "loss": 0.2263, "step": 702 }, { "epoch": 0.37, "learning_rate": 1.9896946039578324e-05, "loss": 0.2173, "step": 703 }, { "epoch": 0.37, "learning_rate": 1.989645439991761e-05, "loss": 0.2647, "step": 704 }, { "epoch": 0.37, "learning_rate": 1.9895961596416024e-05, "loss": 0.1204, "step": 705 }, { "epoch": 0.37, "learning_rate": 1.9895467629131532e-05, "loss": 0.2769, "step": 706 }, { "epoch": 0.37, "learning_rate": 1.9894972498122213e-05, "loss": 0.2311, "step": 707 }, { "epoch": 0.37, "learning_rate": 1.9894476203446308e-05, "loss": 0.1996, "step": 708 }, { "epoch": 0.38, "learning_rate": 1.989397874516217e-05, "loss": 0.191, "step": 709 }, { "epoch": 0.38, "learning_rate": 1.989348012332831e-05, "loss": 0.3325, "step": 710 }, { "epoch": 0.38, "learning_rate": 1.9892980338003364e-05, "loss": 0.2061, "step": 711 }, { "epoch": 0.38, "learning_rate": 1.9892479389246106e-05, "loss": 0.1651, "step": 712 }, { "epoch": 0.38, "learning_rate": 1.989197727711545e-05, "loss": 0.1653, "step": 713 }, { "epoch": 0.38, "learning_rate": 1.9891474001670447e-05, "loss": 0.2746, "step": 714 }, { "epoch": 0.38, "learning_rate": 1.989096956297028e-05, "loss": 0.2196, "step": 715 }, { "epoch": 0.38, "learning_rate": 1.9890463961074273e-05, "loss": 0.3351, "step": 716 }, { "epoch": 0.38, "learning_rate": 1.9889957196041886e-05, "loss": 0.1057, "step": 717 }, { "epoch": 0.38, "learning_rate": 1.9889449267932713e-05, "loss": 0.1467, "step": 718 }, { "epoch": 0.38, "learning_rate": 1.9888940176806494e-05, "loss": 0.1975, "step": 719 }, { "epoch": 0.38, "learning_rate": 1.988842992272309e-05, "loss": 0.1672, "step": 720 }, { "epoch": 0.38, "learning_rate": 1.9887918505742514e-05, "loss": 0.2443, "step": 721 }, { "epoch": 0.38, "learning_rate": 1.988740592592491e-05, "loss": 0.2059, "step": 722 }, { "epoch": 0.38, "learning_rate": 1.9886892183330557e-05, "loss": 0.3004, "step": 723 }, { "epoch": 0.38, "learning_rate": 1.9886377278019866e-05, "loss": 0.2759, "step": 724 }, { "epoch": 0.38, "learning_rate": 1.98858612100534e-05, "loss": 0.3338, "step": 725 }, { "epoch": 0.38, "learning_rate": 1.9885343979491846e-05, "loss": 0.3445, "step": 726 }, { "epoch": 0.38, "learning_rate": 1.9884825586396032e-05, "loss": 0.1364, "step": 727 }, { "epoch": 0.39, "learning_rate": 1.9884306030826915e-05, "loss": 0.1738, "step": 728 }, { "epoch": 0.39, "learning_rate": 1.9883785312845605e-05, "loss": 0.1064, "step": 729 }, { "epoch": 0.39, "learning_rate": 1.9883263432513334e-05, "loss": 0.2541, "step": 730 }, { "epoch": 0.39, "learning_rate": 1.9882740389891483e-05, "loss": 0.141, "step": 731 }, { "epoch": 0.39, "learning_rate": 1.9882216185041555e-05, "loss": 0.2095, "step": 732 }, { "epoch": 0.39, "learning_rate": 1.98816908180252e-05, "loss": 0.2766, "step": 733 }, { "epoch": 0.39, "learning_rate": 1.98811642889042e-05, "loss": 0.3181, "step": 734 }, { "epoch": 0.39, "learning_rate": 1.988063659774048e-05, "loss": 0.3473, "step": 735 }, { "epoch": 0.39, "learning_rate": 1.9880107744596095e-05, "loss": 0.2065, "step": 736 }, { "epoch": 0.39, "learning_rate": 1.987957772953324e-05, "loss": 0.2802, "step": 737 }, { "epoch": 0.39, "learning_rate": 1.987904655261425e-05, "loss": 0.174, "step": 738 }, { "epoch": 0.39, "learning_rate": 1.987851421390158e-05, "loss": 0.2709, "step": 739 }, { "epoch": 0.39, "learning_rate": 1.9877980713457844e-05, "loss": 0.2233, "step": 740 }, { "epoch": 0.39, "learning_rate": 1.987744605134578e-05, "loss": 0.2174, "step": 741 }, { "epoch": 0.39, "learning_rate": 1.9876910227628267e-05, "loss": 0.3169, "step": 742 }, { "epoch": 0.39, "learning_rate": 1.9876373242368318e-05, "loss": 0.1722, "step": 743 }, { "epoch": 0.39, "learning_rate": 1.987583509562908e-05, "loss": 0.2045, "step": 744 }, { "epoch": 0.39, "learning_rate": 1.9875295787473845e-05, "loss": 0.188, "step": 745 }, { "epoch": 0.39, "learning_rate": 1.9874755317966033e-05, "loss": 0.214, "step": 746 }, { "epoch": 0.4, "learning_rate": 1.9874213687169206e-05, "loss": 0.263, "step": 747 }, { "epoch": 0.4, "learning_rate": 1.987367089514706e-05, "loss": 0.1311, "step": 748 }, { "epoch": 0.4, "learning_rate": 1.987312694196343e-05, "loss": 0.1629, "step": 749 }, { "epoch": 0.4, "learning_rate": 1.987258182768228e-05, "loss": 0.2826, "step": 750 }, { "epoch": 0.4, "learning_rate": 1.9872035552367727e-05, "loss": 0.2814, "step": 751 }, { "epoch": 0.4, "learning_rate": 1.9871488116084007e-05, "loss": 0.2751, "step": 752 }, { "epoch": 0.4, "learning_rate": 1.9870939518895497e-05, "loss": 0.312, "step": 753 }, { "epoch": 0.4, "learning_rate": 1.9870389760866717e-05, "loss": 0.2431, "step": 754 }, { "epoch": 0.4, "learning_rate": 1.9869838842062322e-05, "loss": 0.3343, "step": 755 }, { "epoch": 0.4, "learning_rate": 1.9869286762547094e-05, "loss": 0.118, "step": 756 }, { "epoch": 0.4, "learning_rate": 1.9868733522385966e-05, "loss": 0.3153, "step": 757 }, { "epoch": 0.4, "learning_rate": 1.9868179121643997e-05, "loss": 0.3495, "step": 758 }, { "epoch": 0.4, "learning_rate": 1.986762356038638e-05, "loss": 0.2987, "step": 759 }, { "epoch": 0.4, "learning_rate": 1.986706683867846e-05, "loss": 0.2979, "step": 760 }, { "epoch": 0.4, "learning_rate": 1.98665089565857e-05, "loss": 0.3024, "step": 761 }, { "epoch": 0.4, "learning_rate": 1.986594991417371e-05, "loss": 0.1311, "step": 762 }, { "epoch": 0.4, "learning_rate": 1.9865389711508243e-05, "loss": 0.1167, "step": 763 }, { "epoch": 0.4, "learning_rate": 1.986482834865517e-05, "loss": 0.2428, "step": 764 }, { "epoch": 0.4, "learning_rate": 1.9864265825680507e-05, "loss": 0.1666, "step": 765 }, { "epoch": 0.41, "learning_rate": 1.9863702142650418e-05, "loss": 0.2342, "step": 766 }, { "epoch": 0.41, "learning_rate": 1.986313729963118e-05, "loss": 0.2028, "step": 767 }, { "epoch": 0.41, "learning_rate": 1.986257129668923e-05, "loss": 0.1619, "step": 768 }, { "epoch": 0.41, "learning_rate": 1.986200413389113e-05, "loss": 0.2938, "step": 769 }, { "epoch": 0.41, "learning_rate": 1.986143581130357e-05, "loss": 0.1361, "step": 770 }, { "epoch": 0.41, "learning_rate": 1.98608663289934e-05, "loss": 0.2128, "step": 771 }, { "epoch": 0.41, "learning_rate": 1.9860295687027583e-05, "loss": 0.1582, "step": 772 }, { "epoch": 0.41, "learning_rate": 1.9859723885473225e-05, "loss": 0.2226, "step": 773 }, { "epoch": 0.41, "learning_rate": 1.9859150924397577e-05, "loss": 0.2538, "step": 774 }, { "epoch": 0.41, "learning_rate": 1.985857680386802e-05, "loss": 0.1975, "step": 775 }, { "epoch": 0.41, "learning_rate": 1.985800152395207e-05, "loss": 0.2229, "step": 776 }, { "epoch": 0.41, "learning_rate": 1.9857425084717378e-05, "loss": 0.1774, "step": 777 }, { "epoch": 0.41, "learning_rate": 1.985684748623174e-05, "loss": 0.3467, "step": 778 }, { "epoch": 0.41, "learning_rate": 1.9856268728563078e-05, "loss": 0.3318, "step": 779 }, { "epoch": 0.41, "learning_rate": 1.9855688811779457e-05, "loss": 0.2468, "step": 780 }, { "epoch": 0.41, "learning_rate": 1.9855107735949077e-05, "loss": 0.3241, "step": 781 }, { "epoch": 0.41, "learning_rate": 1.9854525501140273e-05, "loss": 0.2384, "step": 782 }, { "epoch": 0.41, "learning_rate": 1.9853942107421516e-05, "loss": 0.3909, "step": 783 }, { "epoch": 0.41, "learning_rate": 1.9853357554861415e-05, "loss": 0.1282, "step": 784 }, { "epoch": 0.42, "learning_rate": 1.9852771843528712e-05, "loss": 0.403, "step": 785 }, { "epoch": 0.42, "learning_rate": 1.9852184973492292e-05, "loss": 0.3027, "step": 786 }, { "epoch": 0.42, "learning_rate": 1.9851596944821167e-05, "loss": 0.2732, "step": 787 }, { "epoch": 0.42, "learning_rate": 1.98510077575845e-05, "loss": 0.2696, "step": 788 }, { "epoch": 0.42, "learning_rate": 1.9850417411851565e-05, "loss": 0.1355, "step": 789 }, { "epoch": 0.42, "learning_rate": 1.9849825907691798e-05, "loss": 0.1902, "step": 790 }, { "epoch": 0.42, "learning_rate": 1.984923324517476e-05, "loss": 0.174, "step": 791 }, { "epoch": 0.42, "learning_rate": 1.9848639424370152e-05, "loss": 0.1769, "step": 792 }, { "epoch": 0.42, "learning_rate": 1.98480444453478e-05, "loss": 0.2301, "step": 793 }, { "epoch": 0.42, "learning_rate": 1.9847448308177678e-05, "loss": 0.2371, "step": 794 }, { "epoch": 0.42, "learning_rate": 1.98468510129299e-05, "loss": 0.1646, "step": 795 }, { "epoch": 0.42, "learning_rate": 1.9846252559674698e-05, "loss": 0.2075, "step": 796 }, { "epoch": 0.42, "learning_rate": 1.984565294848246e-05, "loss": 0.1589, "step": 797 }, { "epoch": 0.42, "learning_rate": 1.9845052179423695e-05, "loss": 0.1164, "step": 798 }, { "epoch": 0.42, "learning_rate": 1.9844450252569056e-05, "loss": 0.1871, "step": 799 }, { "epoch": 0.42, "learning_rate": 1.9843847167989337e-05, "loss": 0.2884, "step": 800 }, { "epoch": 0.42, "learning_rate": 1.9843242925755452e-05, "loss": 0.2551, "step": 801 }, { "epoch": 0.42, "learning_rate": 1.9842637525938468e-05, "loss": 0.348, "step": 802 }, { "epoch": 0.43, "learning_rate": 1.9842030968609574e-05, "loss": 0.172, "step": 803 }, { "epoch": 0.43, "learning_rate": 1.9841423253840113e-05, "loss": 0.2832, "step": 804 }, { "epoch": 0.43, "learning_rate": 1.9840814381701547e-05, "loss": 0.1925, "step": 805 }, { "epoch": 0.43, "learning_rate": 1.9840204352265476e-05, "loss": 0.1623, "step": 806 }, { "epoch": 0.43, "learning_rate": 1.983959316560365e-05, "loss": 0.2191, "step": 807 }, { "epoch": 0.43, "learning_rate": 1.983898082178794e-05, "loss": 0.3226, "step": 808 }, { "epoch": 0.43, "learning_rate": 1.983836732089036e-05, "loss": 0.2034, "step": 809 }, { "epoch": 0.43, "learning_rate": 1.9837752662983056e-05, "loss": 0.2321, "step": 810 }, { "epoch": 0.43, "learning_rate": 1.9837136848138317e-05, "loss": 0.2435, "step": 811 }, { "epoch": 0.43, "learning_rate": 1.9836519876428563e-05, "loss": 0.1807, "step": 812 }, { "epoch": 0.43, "learning_rate": 1.9835901747926352e-05, "loss": 0.2673, "step": 813 }, { "epoch": 0.43, "learning_rate": 1.9835282462704372e-05, "loss": 0.2116, "step": 814 }, { "epoch": 0.43, "learning_rate": 1.9834662020835456e-05, "loss": 0.1573, "step": 815 }, { "epoch": 0.43, "learning_rate": 1.983404042239257e-05, "loss": 0.2065, "step": 816 }, { "epoch": 0.43, "learning_rate": 1.9833417667448812e-05, "loss": 0.1911, "step": 817 }, { "epoch": 0.43, "learning_rate": 1.9832793756077423e-05, "loss": 0.2327, "step": 818 }, { "epoch": 0.43, "learning_rate": 1.9832168688351772e-05, "loss": 0.3149, "step": 819 }, { "epoch": 0.43, "learning_rate": 1.983154246434537e-05, "loss": 0.3124, "step": 820 }, { "epoch": 0.43, "learning_rate": 1.9830915084131864e-05, "loss": 0.1862, "step": 821 }, { "epoch": 0.44, "learning_rate": 1.983028654778503e-05, "loss": 0.18, "step": 822 }, { "epoch": 0.44, "learning_rate": 1.9829656855378787e-05, "loss": 0.3075, "step": 823 }, { "epoch": 0.44, "learning_rate": 1.9829026006987192e-05, "loss": 0.3959, "step": 824 }, { "epoch": 0.44, "learning_rate": 1.9828394002684432e-05, "loss": 0.2593, "step": 825 }, { "epoch": 0.44, "learning_rate": 1.982776084254483e-05, "loss": 0.128, "step": 826 }, { "epoch": 0.44, "learning_rate": 1.9827126526642843e-05, "loss": 0.174, "step": 827 }, { "epoch": 0.44, "learning_rate": 1.9826491055053075e-05, "loss": 0.2636, "step": 828 }, { "epoch": 0.44, "learning_rate": 1.982585442785026e-05, "loss": 0.1244, "step": 829 }, { "epoch": 0.44, "learning_rate": 1.982521664510926e-05, "loss": 0.1733, "step": 830 }, { "epoch": 0.44, "learning_rate": 1.982457770690508e-05, "loss": 0.2366, "step": 831 }, { "epoch": 0.44, "learning_rate": 1.9823937613312863e-05, "loss": 0.272, "step": 832 }, { "epoch": 0.44, "learning_rate": 1.982329636440788e-05, "loss": 0.1595, "step": 833 }, { "epoch": 0.44, "learning_rate": 1.9822653960265552e-05, "loss": 0.201, "step": 834 }, { "epoch": 0.44, "learning_rate": 1.982201040096142e-05, "loss": 0.2166, "step": 835 }, { "epoch": 0.44, "learning_rate": 1.9821365686571172e-05, "loss": 0.2518, "step": 836 }, { "epoch": 0.44, "learning_rate": 1.9820719817170625e-05, "loss": 0.1737, "step": 837 }, { "epoch": 0.44, "learning_rate": 1.9820072792835738e-05, "loss": 0.2317, "step": 838 }, { "epoch": 0.44, "learning_rate": 1.981942461364259e-05, "loss": 0.2609, "step": 839 }, { "epoch": 0.44, "learning_rate": 1.9818775279667425e-05, "loss": 0.2377, "step": 840 }, { "epoch": 0.45, "learning_rate": 1.9818124790986595e-05, "loss": 0.2015, "step": 841 }, { "epoch": 0.45, "learning_rate": 1.98174731476766e-05, "loss": 0.2761, "step": 842 }, { "epoch": 0.45, "learning_rate": 1.9816820349814078e-05, "loss": 0.1878, "step": 843 }, { "epoch": 0.45, "learning_rate": 1.9816166397475796e-05, "loss": 0.3091, "step": 844 }, { "epoch": 0.45, "learning_rate": 1.9815511290738666e-05, "loss": 0.4403, "step": 845 }, { "epoch": 0.45, "learning_rate": 1.981485502967972e-05, "loss": 0.1977, "step": 846 }, { "epoch": 0.45, "learning_rate": 1.9814197614376137e-05, "loss": 0.167, "step": 847 }, { "epoch": 0.45, "learning_rate": 1.9813539044905238e-05, "loss": 0.272, "step": 848 }, { "epoch": 0.45, "learning_rate": 1.981287932134447e-05, "loss": 0.2473, "step": 849 }, { "epoch": 0.45, "learning_rate": 1.9812218443771412e-05, "loss": 0.2149, "step": 850 }, { "epoch": 0.45, "learning_rate": 1.9811556412263788e-05, "loss": 0.2295, "step": 851 }, { "epoch": 0.45, "learning_rate": 1.9810893226899453e-05, "loss": 0.3151, "step": 852 }, { "epoch": 0.45, "learning_rate": 1.9810228887756403e-05, "loss": 0.4157, "step": 853 }, { "epoch": 0.45, "learning_rate": 1.9809563394912757e-05, "loss": 0.2262, "step": 854 }, { "epoch": 0.45, "learning_rate": 1.9808896748446788e-05, "loss": 0.1367, "step": 855 }, { "epoch": 0.45, "learning_rate": 1.9808228948436883e-05, "loss": 0.1457, "step": 856 }, { "epoch": 0.45, "learning_rate": 1.980755999496159e-05, "loss": 0.1301, "step": 857 }, { "epoch": 0.45, "learning_rate": 1.9806889888099573e-05, "loss": 0.1416, "step": 858 }, { "epoch": 0.45, "learning_rate": 1.9806218627929633e-05, "loss": 0.3045, "step": 859 }, { "epoch": 0.46, "learning_rate": 1.980554621453072e-05, "loss": 0.2247, "step": 860 }, { "epoch": 0.46, "learning_rate": 1.9804872647981903e-05, "loss": 0.1511, "step": 861 }, { "epoch": 0.46, "learning_rate": 1.98041979283624e-05, "loss": 0.1647, "step": 862 }, { "epoch": 0.46, "learning_rate": 1.9803522055751557e-05, "loss": 0.2222, "step": 863 }, { "epoch": 0.46, "learning_rate": 1.980284503022886e-05, "loss": 0.2678, "step": 864 }, { "epoch": 0.46, "learning_rate": 1.9802166851873925e-05, "loss": 0.3829, "step": 865 }, { "epoch": 0.46, "learning_rate": 1.9801487520766513e-05, "loss": 0.2994, "step": 866 }, { "epoch": 0.46, "learning_rate": 1.9800807036986504e-05, "loss": 0.2398, "step": 867 }, { "epoch": 0.46, "learning_rate": 1.9800125400613935e-05, "loss": 0.3297, "step": 868 }, { "epoch": 0.46, "learning_rate": 1.9799442611728964e-05, "loss": 0.2428, "step": 869 }, { "epoch": 0.46, "learning_rate": 1.9798758670411886e-05, "loss": 0.2451, "step": 870 }, { "epoch": 0.46, "learning_rate": 1.9798073576743137e-05, "loss": 0.2817, "step": 871 }, { "epoch": 0.46, "learning_rate": 1.979738733080328e-05, "loss": 0.2186, "step": 872 }, { "epoch": 0.46, "learning_rate": 1.9796699932673023e-05, "loss": 0.2541, "step": 873 }, { "epoch": 0.46, "learning_rate": 1.9796011382433206e-05, "loss": 0.1578, "step": 874 }, { "epoch": 0.46, "learning_rate": 1.9795321680164798e-05, "loss": 0.1974, "step": 875 }, { "epoch": 0.46, "learning_rate": 1.979463082594892e-05, "loss": 0.313, "step": 876 }, { "epoch": 0.46, "learning_rate": 1.9793938819866806e-05, "loss": 0.2081, "step": 877 }, { "epoch": 0.46, "learning_rate": 1.9793245661999845e-05, "loss": 0.1746, "step": 878 }, { "epoch": 0.47, "learning_rate": 1.979255135242955e-05, "loss": 0.1484, "step": 879 }, { "epoch": 0.47, "learning_rate": 1.9791855891237577e-05, "loss": 0.2144, "step": 880 }, { "epoch": 0.47, "learning_rate": 1.9791159278505706e-05, "loss": 0.3924, "step": 881 }, { "epoch": 0.47, "learning_rate": 1.9790461514315868e-05, "loss": 0.1738, "step": 882 }, { "epoch": 0.47, "learning_rate": 1.9789762598750115e-05, "loss": 0.1677, "step": 883 }, { "epoch": 0.47, "learning_rate": 1.9789062531890642e-05, "loss": 0.2613, "step": 884 }, { "epoch": 0.47, "learning_rate": 1.9788361313819783e-05, "loss": 0.2109, "step": 885 }, { "epoch": 0.47, "learning_rate": 1.9787658944619998e-05, "loss": 0.165, "step": 886 }, { "epoch": 0.47, "learning_rate": 1.978695542437389e-05, "loss": 0.3092, "step": 887 }, { "epoch": 0.47, "learning_rate": 1.978625075316419e-05, "loss": 0.2844, "step": 888 }, { "epoch": 0.47, "learning_rate": 1.9785544931073773e-05, "loss": 0.2652, "step": 889 }, { "epoch": 0.47, "learning_rate": 1.9784837958185637e-05, "loss": 0.274, "step": 890 }, { "epoch": 0.47, "learning_rate": 1.9784129834582934e-05, "loss": 0.2027, "step": 891 }, { "epoch": 0.47, "learning_rate": 1.9783420560348933e-05, "loss": 0.2614, "step": 892 }, { "epoch": 0.47, "learning_rate": 1.978271013556705e-05, "loss": 0.2472, "step": 893 }, { "epoch": 0.47, "learning_rate": 1.9781998560320832e-05, "loss": 0.0935, "step": 894 }, { "epoch": 0.47, "learning_rate": 1.978128583469396e-05, "loss": 0.1297, "step": 895 }, { "epoch": 0.47, "learning_rate": 1.9780571958770253e-05, "loss": 0.1567, "step": 896 }, { "epoch": 0.47, "learning_rate": 1.977985693263366e-05, "loss": 0.2363, "step": 897 }, { "epoch": 0.48, "learning_rate": 1.9779140756368276e-05, "loss": 0.2857, "step": 898 }, { "epoch": 0.48, "learning_rate": 1.977842343005832e-05, "loss": 0.2544, "step": 899 }, { "epoch": 0.48, "learning_rate": 1.977770495378815e-05, "loss": 0.1812, "step": 900 }, { "epoch": 0.48, "learning_rate": 1.9776985327642264e-05, "loss": 0.3876, "step": 901 }, { "epoch": 0.48, "learning_rate": 1.9776264551705292e-05, "loss": 0.2631, "step": 902 }, { "epoch": 0.48, "learning_rate": 1.9775542626061993e-05, "loss": 0.3296, "step": 903 }, { "epoch": 0.48, "learning_rate": 1.977481955079727e-05, "loss": 0.2612, "step": 904 }, { "epoch": 0.48, "learning_rate": 1.9774095325996158e-05, "loss": 0.1275, "step": 905 }, { "epoch": 0.48, "learning_rate": 1.977336995174383e-05, "loss": 0.1828, "step": 906 }, { "epoch": 0.48, "learning_rate": 1.9772643428125585e-05, "loss": 0.2965, "step": 907 }, { "epoch": 0.48, "learning_rate": 1.9771915755226865e-05, "loss": 0.3017, "step": 908 }, { "epoch": 0.48, "learning_rate": 1.977118693313325e-05, "loss": 0.1835, "step": 909 }, { "epoch": 0.48, "learning_rate": 1.977045696193045e-05, "loss": 0.2463, "step": 910 }, { "epoch": 0.48, "learning_rate": 1.9769725841704307e-05, "loss": 0.3461, "step": 911 }, { "epoch": 0.48, "learning_rate": 1.97689935725408e-05, "loss": 0.2464, "step": 912 }, { "epoch": 0.48, "learning_rate": 1.9768260154526056e-05, "loss": 0.1981, "step": 913 }, { "epoch": 0.48, "learning_rate": 1.9767525587746318e-05, "loss": 0.1704, "step": 914 }, { "epoch": 0.48, "learning_rate": 1.976678987228797e-05, "loss": 0.3053, "step": 915 }, { "epoch": 0.48, "learning_rate": 1.9766053008237543e-05, "loss": 0.2918, "step": 916 }, { "epoch": 0.49, "learning_rate": 1.9765314995681687e-05, "loss": 0.2886, "step": 917 }, { "epoch": 0.49, "learning_rate": 1.9764575834707194e-05, "loss": 0.2715, "step": 918 }, { "epoch": 0.49, "learning_rate": 1.976383552540099e-05, "loss": 0.2385, "step": 919 }, { "epoch": 0.49, "learning_rate": 1.976309406785014e-05, "loss": 0.275, "step": 920 }, { "epoch": 0.49, "learning_rate": 1.9762351462141835e-05, "loss": 0.3184, "step": 921 }, { "epoch": 0.49, "learning_rate": 1.976160770836341e-05, "loss": 0.1751, "step": 922 }, { "epoch": 0.49, "learning_rate": 1.976086280660234e-05, "loss": 0.1862, "step": 923 }, { "epoch": 0.49, "learning_rate": 1.976011675694621e-05, "loss": 0.2146, "step": 924 }, { "epoch": 0.49, "learning_rate": 1.975936955948277e-05, "loss": 0.2916, "step": 925 }, { "epoch": 0.49, "learning_rate": 1.975862121429989e-05, "loss": 0.1843, "step": 926 }, { "epoch": 0.49, "learning_rate": 1.9757871721485575e-05, "loss": 0.1343, "step": 927 }, { "epoch": 0.49, "learning_rate": 1.975712108112796e-05, "loss": 0.1698, "step": 928 }, { "epoch": 0.49, "learning_rate": 1.975636929331533e-05, "loss": 0.3007, "step": 929 }, { "epoch": 0.49, "learning_rate": 1.9755616358136096e-05, "loss": 0.3978, "step": 930 }, { "epoch": 0.49, "learning_rate": 1.97548622756788e-05, "loss": 0.2306, "step": 931 }, { "epoch": 0.49, "learning_rate": 1.9754107046032128e-05, "loss": 0.1514, "step": 932 }, { "epoch": 0.49, "learning_rate": 1.97533506692849e-05, "loss": 0.2095, "step": 933 }, { "epoch": 0.49, "learning_rate": 1.9752593145526053e-05, "loss": 0.1638, "step": 934 }, { "epoch": 0.49, "learning_rate": 1.9751834474844687e-05, "loss": 0.234, "step": 935 }, { "epoch": 0.5, "learning_rate": 1.9751074657330015e-05, "loss": 0.1651, "step": 936 }, { "epoch": 0.5, "learning_rate": 1.9750313693071402e-05, "loss": 0.1683, "step": 937 }, { "epoch": 0.5, "learning_rate": 1.9749551582158325e-05, "loss": 0.2788, "step": 938 }, { "epoch": 0.5, "learning_rate": 1.9748788324680424e-05, "loss": 0.265, "step": 939 }, { "epoch": 0.5, "learning_rate": 1.974802392072745e-05, "loss": 0.2385, "step": 940 }, { "epoch": 0.5, "learning_rate": 1.97472583703893e-05, "loss": 0.1556, "step": 941 }, { "epoch": 0.5, "learning_rate": 1.974649167375601e-05, "loss": 0.1458, "step": 942 }, { "epoch": 0.5, "learning_rate": 1.9745723830917738e-05, "loss": 0.2496, "step": 943 }, { "epoch": 0.5, "learning_rate": 1.9744954841964782e-05, "loss": 0.3032, "step": 944 }, { "epoch": 0.5, "learning_rate": 1.9744184706987582e-05, "loss": 0.2186, "step": 945 }, { "epoch": 0.5, "learning_rate": 1.974341342607671e-05, "loss": 0.3392, "step": 946 }, { "epoch": 0.5, "learning_rate": 1.9742640999322863e-05, "loss": 0.2763, "step": 947 }, { "epoch": 0.5, "learning_rate": 1.9741867426816885e-05, "loss": 0.3098, "step": 948 }, { "epoch": 0.5, "learning_rate": 1.974109270864974e-05, "loss": 0.1698, "step": 949 }, { "epoch": 0.5, "learning_rate": 1.9740316844912548e-05, "loss": 0.1586, "step": 950 }, { "epoch": 0.5, "learning_rate": 1.9739539835696546e-05, "loss": 0.1538, "step": 951 }, { "epoch": 0.5, "learning_rate": 1.9738761681093113e-05, "loss": 0.1364, "step": 952 }, { "epoch": 0.5, "learning_rate": 1.9737982381193765e-05, "loss": 0.2318, "step": 953 }, { "epoch": 0.5, "learning_rate": 1.973720193609014e-05, "loss": 0.3293, "step": 954 }, { "epoch": 0.51, "learning_rate": 1.9736420345874026e-05, "loss": 0.1767, "step": 955 }, { "epoch": 0.51, "learning_rate": 1.973563761063734e-05, "loss": 0.2253, "step": 956 }, { "epoch": 0.51, "learning_rate": 1.973485373047213e-05, "loss": 0.3146, "step": 957 }, { "epoch": 0.51, "learning_rate": 1.9734068705470584e-05, "loss": 0.2941, "step": 958 }, { "epoch": 0.51, "learning_rate": 1.9733282535725022e-05, "loss": 0.2359, "step": 959 }, { "epoch": 0.51, "learning_rate": 1.9732495221327897e-05, "loss": 0.2223, "step": 960 }, { "epoch": 0.51, "learning_rate": 1.9731706762371803e-05, "loss": 0.1922, "step": 961 }, { "epoch": 0.51, "learning_rate": 1.973091715894946e-05, "loss": 0.2616, "step": 962 }, { "epoch": 0.51, "learning_rate": 1.973012641115373e-05, "loss": 0.1962, "step": 963 }, { "epoch": 0.51, "learning_rate": 1.97293345190776e-05, "loss": 0.2626, "step": 964 }, { "epoch": 0.51, "learning_rate": 1.9728541482814206e-05, "loss": 0.1771, "step": 965 }, { "epoch": 0.51, "learning_rate": 1.9727747302456806e-05, "loss": 0.2018, "step": 966 }, { "epoch": 0.51, "learning_rate": 1.97269519780988e-05, "loss": 0.1227, "step": 967 }, { "epoch": 0.51, "learning_rate": 1.9726155509833716e-05, "loss": 0.2522, "step": 968 }, { "epoch": 0.51, "learning_rate": 1.9725357897755223e-05, "loss": 0.3823, "step": 969 }, { "epoch": 0.51, "learning_rate": 1.9724559141957117e-05, "loss": 0.214, "step": 970 }, { "epoch": 0.51, "learning_rate": 1.972375924253334e-05, "loss": 0.1412, "step": 971 }, { "epoch": 0.51, "learning_rate": 1.9722958199577956e-05, "loss": 0.2626, "step": 972 }, { "epoch": 0.52, "learning_rate": 1.972215601318517e-05, "loss": 0.1708, "step": 973 }, { "epoch": 0.52, "learning_rate": 1.9721352683449326e-05, "loss": 0.1271, "step": 974 }, { "epoch": 0.52, "learning_rate": 1.972054821046489e-05, "loss": 0.309, "step": 975 }, { "epoch": 0.52, "learning_rate": 1.971974259432647e-05, "loss": 0.2551, "step": 976 }, { "epoch": 0.52, "learning_rate": 1.9718935835128815e-05, "loss": 0.2904, "step": 977 }, { "epoch": 0.52, "learning_rate": 1.9718127932966792e-05, "loss": 0.2507, "step": 978 }, { "epoch": 0.52, "learning_rate": 1.9717318887935417e-05, "loss": 0.3428, "step": 979 }, { "epoch": 0.52, "learning_rate": 1.9716508700129836e-05, "loss": 0.2526, "step": 980 }, { "epoch": 0.52, "learning_rate": 1.9715697369645327e-05, "loss": 0.1575, "step": 981 }, { "epoch": 0.52, "learning_rate": 1.9714884896577303e-05, "loss": 0.2305, "step": 982 }, { "epoch": 0.52, "learning_rate": 1.9714071281021316e-05, "loss": 0.1874, "step": 983 }, { "epoch": 0.52, "learning_rate": 1.971325652307304e-05, "loss": 0.2085, "step": 984 }, { "epoch": 0.52, "learning_rate": 1.97124406228283e-05, "loss": 0.2325, "step": 985 }, { "epoch": 0.52, "learning_rate": 1.9711623580383044e-05, "loss": 0.2458, "step": 986 }, { "epoch": 0.52, "learning_rate": 1.9710805395833365e-05, "loss": 0.2143, "step": 987 }, { "epoch": 0.52, "learning_rate": 1.970998606927547e-05, "loss": 0.176, "step": 988 }, { "epoch": 0.52, "learning_rate": 1.9709165600805723e-05, "loss": 0.3346, "step": 989 }, { "epoch": 0.52, "learning_rate": 1.970834399052061e-05, "loss": 0.1862, "step": 990 }, { "epoch": 0.52, "learning_rate": 1.970752123851675e-05, "loss": 0.2819, "step": 991 }, { "epoch": 0.53, "learning_rate": 1.970669734489091e-05, "loss": 0.2975, "step": 992 }, { "epoch": 0.53, "learning_rate": 1.970587230973997e-05, "loss": 0.2153, "step": 993 }, { "epoch": 0.53, "learning_rate": 1.970504613316096e-05, "loss": 0.2243, "step": 994 }, { "epoch": 0.53, "learning_rate": 1.970421881525104e-05, "loss": 0.0933, "step": 995 }, { "epoch": 0.53, "learning_rate": 1.970339035610751e-05, "loss": 0.3152, "step": 996 }, { "epoch": 0.53, "learning_rate": 1.9702560755827788e-05, "loss": 0.2311, "step": 997 }, { "epoch": 0.53, "learning_rate": 1.9701730014509444e-05, "loss": 0.216, "step": 998 }, { "epoch": 0.53, "learning_rate": 1.970089813225017e-05, "loss": 0.1034, "step": 999 }, { "epoch": 0.53, "learning_rate": 1.97000651091478e-05, "loss": 0.2811, "step": 1000 }, { "epoch": 0.53, "learning_rate": 1.9699230945300297e-05, "loss": 0.132, "step": 1001 }, { "epoch": 0.53, "learning_rate": 1.9698395640805762e-05, "loss": 0.2958, "step": 1002 }, { "epoch": 0.53, "learning_rate": 1.9697559195762427e-05, "loss": 0.2023, "step": 1003 }, { "epoch": 0.53, "learning_rate": 1.969672161026866e-05, "loss": 0.298, "step": 1004 }, { "epoch": 0.53, "learning_rate": 1.969588288442296e-05, "loss": 0.1617, "step": 1005 }, { "epoch": 0.53, "learning_rate": 1.9695043018323967e-05, "loss": 0.1336, "step": 1006 }, { "epoch": 0.53, "learning_rate": 1.969420201207045e-05, "loss": 0.2256, "step": 1007 }, { "epoch": 0.53, "learning_rate": 1.9693359865761308e-05, "loss": 0.2441, "step": 1008 }, { "epoch": 0.53, "learning_rate": 1.9692516579495584e-05, "loss": 0.3728, "step": 1009 }, { "epoch": 0.53, "learning_rate": 1.969167215337245e-05, "loss": 0.2378, "step": 1010 }, { "epoch": 0.54, "learning_rate": 1.969082658749121e-05, "loss": 0.2162, "step": 1011 }, { "epoch": 0.54, "learning_rate": 1.9689979881951303e-05, "loss": 0.3822, "step": 1012 }, { "epoch": 0.54, "learning_rate": 1.9689132036852305e-05, "loss": 0.2104, "step": 1013 }, { "epoch": 0.54, "learning_rate": 1.9688283052293923e-05, "loss": 0.2683, "step": 1014 }, { "epoch": 0.54, "learning_rate": 1.9687432928376006e-05, "loss": 0.217, "step": 1015 }, { "epoch": 0.54, "learning_rate": 1.9686581665198518e-05, "loss": 0.1587, "step": 1016 }, { "epoch": 0.54, "learning_rate": 1.9685729262861577e-05, "loss": 0.2902, "step": 1017 }, { "epoch": 0.54, "learning_rate": 1.9684875721465426e-05, "loss": 0.3171, "step": 1018 }, { "epoch": 0.54, "learning_rate": 1.968402104111044e-05, "loss": 0.2493, "step": 1019 }, { "epoch": 0.54, "learning_rate": 1.9683165221897132e-05, "loss": 0.204, "step": 1020 }, { "epoch": 0.54, "learning_rate": 1.9682308263926154e-05, "loss": 0.1488, "step": 1021 }, { "epoch": 0.54, "learning_rate": 1.9681450167298278e-05, "loss": 0.2084, "step": 1022 }, { "epoch": 0.54, "learning_rate": 1.968059093211442e-05, "loss": 0.2233, "step": 1023 }, { "epoch": 0.54, "learning_rate": 1.967973055847563e-05, "loss": 0.1277, "step": 1024 }, { "epoch": 0.54, "learning_rate": 1.9678869046483085e-05, "loss": 0.3581, "step": 1025 }, { "epoch": 0.54, "learning_rate": 1.9678006396238106e-05, "loss": 0.1982, "step": 1026 }, { "epoch": 0.54, "learning_rate": 1.9677142607842136e-05, "loss": 0.3801, "step": 1027 }, { "epoch": 0.54, "learning_rate": 1.9676277681396763e-05, "loss": 0.2929, "step": 1028 }, { "epoch": 0.54, "learning_rate": 1.96754116170037e-05, "loss": 0.2391, "step": 1029 }, { "epoch": 0.55, "learning_rate": 1.9674544414764804e-05, "loss": 0.1938, "step": 1030 }, { "epoch": 0.55, "learning_rate": 1.9673676074782054e-05, "loss": 0.1528, "step": 1031 }, { "epoch": 0.55, "learning_rate": 1.967280659715757e-05, "loss": 0.2734, "step": 1032 }, { "epoch": 0.55, "learning_rate": 1.96719359819936e-05, "loss": 0.2014, "step": 1033 }, { "epoch": 0.55, "learning_rate": 1.9671064229392538e-05, "loss": 0.3444, "step": 1034 }, { "epoch": 0.55, "learning_rate": 1.96701913394569e-05, "loss": 0.301, "step": 1035 }, { "epoch": 0.55, "learning_rate": 1.9669317312289336e-05, "loss": 0.37, "step": 1036 }, { "epoch": 0.55, "learning_rate": 1.966844214799264e-05, "loss": 0.1441, "step": 1037 }, { "epoch": 0.55, "learning_rate": 1.9667565846669725e-05, "loss": 0.2402, "step": 1038 }, { "epoch": 0.55, "learning_rate": 1.966668840842365e-05, "loss": 0.273, "step": 1039 }, { "epoch": 0.55, "learning_rate": 1.9665809833357606e-05, "loss": 0.1766, "step": 1040 }, { "epoch": 0.55, "learning_rate": 1.966493012157491e-05, "loss": 0.2498, "step": 1041 }, { "epoch": 0.55, "learning_rate": 1.966404927317902e-05, "loss": 0.0752, "step": 1042 }, { "epoch": 0.55, "learning_rate": 1.9663167288273523e-05, "loss": 0.1603, "step": 1043 }, { "epoch": 0.55, "learning_rate": 1.9662284166962147e-05, "loss": 0.2523, "step": 1044 }, { "epoch": 0.55, "learning_rate": 1.9661399909348743e-05, "loss": 0.2296, "step": 1045 }, { "epoch": 0.55, "learning_rate": 1.9660514515537307e-05, "loss": 0.2516, "step": 1046 }, { "epoch": 0.55, "learning_rate": 1.965962798563196e-05, "loss": 0.1408, "step": 1047 }, { "epoch": 0.55, "learning_rate": 1.9658740319736958e-05, "loss": 0.2341, "step": 1048 }, { "epoch": 0.56, "learning_rate": 1.9657851517956693e-05, "loss": 0.1969, "step": 1049 }, { "epoch": 0.56, "learning_rate": 1.965696158039569e-05, "loss": 0.2395, "step": 1050 }, { "epoch": 0.56, "learning_rate": 1.9656070507158606e-05, "loss": 0.1543, "step": 1051 }, { "epoch": 0.56, "learning_rate": 1.9655178298350236e-05, "loss": 0.2465, "step": 1052 }, { "epoch": 0.56, "learning_rate": 1.9654284954075505e-05, "loss": 0.2708, "step": 1053 }, { "epoch": 0.56, "learning_rate": 1.9653390474439467e-05, "loss": 0.1824, "step": 1054 }, { "epoch": 0.56, "learning_rate": 1.9652494859547323e-05, "loss": 0.232, "step": 1055 }, { "epoch": 0.56, "learning_rate": 1.9651598109504394e-05, "loss": 0.1362, "step": 1056 }, { "epoch": 0.56, "learning_rate": 1.9650700224416136e-05, "loss": 0.1864, "step": 1057 }, { "epoch": 0.56, "learning_rate": 1.9649801204388144e-05, "loss": 0.3072, "step": 1058 }, { "epoch": 0.56, "learning_rate": 1.964890104952615e-05, "loss": 0.1758, "step": 1059 }, { "epoch": 0.56, "learning_rate": 1.9647999759936005e-05, "loss": 0.1737, "step": 1060 }, { "epoch": 0.56, "learning_rate": 1.964709733572371e-05, "loss": 0.2002, "step": 1061 }, { "epoch": 0.56, "learning_rate": 1.9646193776995383e-05, "loss": 0.368, "step": 1062 }, { "epoch": 0.56, "learning_rate": 1.9645289083857295e-05, "loss": 0.144, "step": 1063 }, { "epoch": 0.56, "learning_rate": 1.9644383256415834e-05, "loss": 0.4139, "step": 1064 }, { "epoch": 0.56, "learning_rate": 1.9643476294777525e-05, "loss": 0.1764, "step": 1065 }, { "epoch": 0.56, "learning_rate": 1.964256819904903e-05, "loss": 0.2327, "step": 1066 }, { "epoch": 0.56, "learning_rate": 1.9641658969337146e-05, "loss": 0.3948, "step": 1067 }, { "epoch": 0.57, "learning_rate": 1.9640748605748793e-05, "loss": 0.3572, "step": 1068 }, { "epoch": 0.57, "learning_rate": 1.9639837108391038e-05, "loss": 0.2703, "step": 1069 }, { "epoch": 0.57, "learning_rate": 1.963892447737107e-05, "loss": 0.1806, "step": 1070 }, { "epoch": 0.57, "learning_rate": 1.963801071279622e-05, "loss": 0.1526, "step": 1071 }, { "epoch": 0.57, "learning_rate": 1.9637095814773947e-05, "loss": 0.1524, "step": 1072 }, { "epoch": 0.57, "learning_rate": 1.9636179783411845e-05, "loss": 0.2414, "step": 1073 }, { "epoch": 0.57, "learning_rate": 1.9635262618817637e-05, "loss": 0.2099, "step": 1074 }, { "epoch": 0.57, "learning_rate": 1.963434432109919e-05, "loss": 0.1792, "step": 1075 }, { "epoch": 0.57, "learning_rate": 1.9633424890364493e-05, "loss": 0.1342, "step": 1076 }, { "epoch": 0.57, "learning_rate": 1.963250432672167e-05, "loss": 0.1416, "step": 1077 }, { "epoch": 0.57, "learning_rate": 1.9631582630278992e-05, "loss": 0.2, "step": 1078 }, { "epoch": 0.57, "learning_rate": 1.963065980114484e-05, "loss": 0.2333, "step": 1079 }, { "epoch": 0.57, "learning_rate": 1.9629735839427745e-05, "loss": 0.2384, "step": 1080 }, { "epoch": 0.57, "learning_rate": 1.9628810745236367e-05, "loss": 0.1867, "step": 1081 }, { "epoch": 0.57, "learning_rate": 1.9627884518679495e-05, "loss": 0.2061, "step": 1082 }, { "epoch": 0.57, "learning_rate": 1.9626957159866064e-05, "loss": 0.3485, "step": 1083 }, { "epoch": 0.57, "learning_rate": 1.9626028668905125e-05, "loss": 0.2785, "step": 1084 }, { "epoch": 0.57, "learning_rate": 1.9625099045905872e-05, "loss": 0.1563, "step": 1085 }, { "epoch": 0.57, "learning_rate": 1.9624168290977634e-05, "loss": 0.2906, "step": 1086 }, { "epoch": 0.58, "learning_rate": 1.9623236404229863e-05, "loss": 0.2105, "step": 1087 }, { "epoch": 0.58, "learning_rate": 1.9622303385772155e-05, "loss": 0.2326, "step": 1088 }, { "epoch": 0.58, "learning_rate": 1.9621369235714233e-05, "loss": 0.1755, "step": 1089 }, { "epoch": 0.58, "learning_rate": 1.9620433954165955e-05, "loss": 0.2255, "step": 1090 }, { "epoch": 0.58, "learning_rate": 1.9619497541237312e-05, "loss": 0.1886, "step": 1091 }, { "epoch": 0.58, "learning_rate": 1.961855999703843e-05, "loss": 0.2121, "step": 1092 }, { "epoch": 0.58, "learning_rate": 1.9617621321679567e-05, "loss": 0.1819, "step": 1093 }, { "epoch": 0.58, "learning_rate": 1.96166815152711e-05, "loss": 0.148, "step": 1094 }, { "epoch": 0.58, "learning_rate": 1.961574057792357e-05, "loss": 0.2027, "step": 1095 }, { "epoch": 0.58, "learning_rate": 1.961479850974762e-05, "loss": 0.1469, "step": 1096 }, { "epoch": 0.58, "learning_rate": 1.9613855310854047e-05, "loss": 0.2118, "step": 1097 }, { "epoch": 0.58, "learning_rate": 1.961291098135377e-05, "loss": 0.2613, "step": 1098 }, { "epoch": 0.58, "learning_rate": 1.9611965521357838e-05, "loss": 0.3377, "step": 1099 }, { "epoch": 0.58, "learning_rate": 1.9611018930977444e-05, "loss": 0.1808, "step": 1100 }, { "epoch": 0.58, "learning_rate": 1.9610071210323916e-05, "loss": 0.3001, "step": 1101 }, { "epoch": 0.58, "learning_rate": 1.9609122359508696e-05, "loss": 0.3536, "step": 1102 }, { "epoch": 0.58, "learning_rate": 1.9608172378643375e-05, "loss": 0.1815, "step": 1103 }, { "epoch": 0.58, "learning_rate": 1.960722126783967e-05, "loss": 0.4018, "step": 1104 }, { "epoch": 0.58, "learning_rate": 1.960626902720944e-05, "loss": 0.1831, "step": 1105 }, { "epoch": 0.59, "learning_rate": 1.9605315656864667e-05, "loss": 0.2771, "step": 1106 }, { "epoch": 0.59, "learning_rate": 1.9604361156917466e-05, "loss": 0.2214, "step": 1107 }, { "epoch": 0.59, "learning_rate": 1.9603405527480088e-05, "loss": 0.1722, "step": 1108 }, { "epoch": 0.59, "learning_rate": 1.9602448768664922e-05, "loss": 0.2414, "step": 1109 }, { "epoch": 0.59, "learning_rate": 1.960149088058448e-05, "loss": 0.1707, "step": 1110 }, { "epoch": 0.59, "learning_rate": 1.960053186335141e-05, "loss": 0.2213, "step": 1111 }, { "epoch": 0.59, "learning_rate": 1.95995717170785e-05, "loss": 0.0974, "step": 1112 }, { "epoch": 0.59, "learning_rate": 1.9598610441878664e-05, "loss": 0.2996, "step": 1113 }, { "epoch": 0.59, "learning_rate": 1.9597648037864942e-05, "loss": 0.2827, "step": 1114 }, { "epoch": 0.59, "learning_rate": 1.9596684505150522e-05, "loss": 0.1868, "step": 1115 }, { "epoch": 0.59, "learning_rate": 1.9595719843848718e-05, "loss": 0.2708, "step": 1116 }, { "epoch": 0.59, "learning_rate": 1.9594754054072972e-05, "loss": 0.2159, "step": 1117 }, { "epoch": 0.59, "learning_rate": 1.959378713593686e-05, "loss": 0.2345, "step": 1118 }, { "epoch": 0.59, "learning_rate": 1.9592819089554102e-05, "loss": 0.2306, "step": 1119 }, { "epoch": 0.59, "learning_rate": 1.959184991503853e-05, "loss": 0.2445, "step": 1120 }, { "epoch": 0.59, "learning_rate": 1.9590879612504135e-05, "loss": 0.2116, "step": 1121 }, { "epoch": 0.59, "learning_rate": 1.9589908182065015e-05, "loss": 0.086, "step": 1122 }, { "epoch": 0.59, "learning_rate": 1.9588935623835416e-05, "loss": 0.1988, "step": 1123 }, { "epoch": 0.59, "learning_rate": 1.9587961937929712e-05, "loss": 0.2034, "step": 1124 }, { "epoch": 0.6, "learning_rate": 1.9586987124462413e-05, "loss": 0.3266, "step": 1125 }, { "epoch": 0.6, "learning_rate": 1.9586011183548155e-05, "loss": 0.2543, "step": 1126 }, { "epoch": 0.6, "learning_rate": 1.9585034115301712e-05, "loss": 0.1635, "step": 1127 }, { "epoch": 0.6, "learning_rate": 1.958405591983799e-05, "loss": 0.1636, "step": 1128 }, { "epoch": 0.6, "learning_rate": 1.9583076597272025e-05, "loss": 0.1945, "step": 1129 }, { "epoch": 0.6, "learning_rate": 1.958209614771899e-05, "loss": 0.3579, "step": 1130 }, { "epoch": 0.6, "learning_rate": 1.958111457129418e-05, "loss": 0.2334, "step": 1131 }, { "epoch": 0.6, "learning_rate": 1.958013186811304e-05, "loss": 0.2555, "step": 1132 }, { "epoch": 0.6, "learning_rate": 1.957914803829113e-05, "loss": 0.2919, "step": 1133 }, { "epoch": 0.6, "learning_rate": 1.9578163081944153e-05, "loss": 0.1487, "step": 1134 }, { "epoch": 0.6, "learning_rate": 1.9577176999187943e-05, "loss": 0.2203, "step": 1135 }, { "epoch": 0.6, "learning_rate": 1.9576189790138466e-05, "loss": 0.181, "step": 1136 }, { "epoch": 0.6, "learning_rate": 1.9575201454911815e-05, "loss": 0.2186, "step": 1137 }, { "epoch": 0.6, "learning_rate": 1.9574211993624223e-05, "loss": 0.2858, "step": 1138 }, { "epoch": 0.6, "learning_rate": 1.9573221406392054e-05, "loss": 0.2332, "step": 1139 }, { "epoch": 0.6, "learning_rate": 1.9572229693331797e-05, "loss": 0.1675, "step": 1140 }, { "epoch": 0.6, "learning_rate": 1.9571236854560084e-05, "loss": 0.3713, "step": 1141 }, { "epoch": 0.6, "learning_rate": 1.9570242890193673e-05, "loss": 0.1899, "step": 1142 }, { "epoch": 0.61, "learning_rate": 1.956924780034946e-05, "loss": 0.2585, "step": 1143 }, { "epoch": 0.61, "learning_rate": 1.9568251585144464e-05, "loss": 0.184, "step": 1144 }, { "epoch": 0.61, "learning_rate": 1.9567254244695847e-05, "loss": 0.1987, "step": 1145 }, { "epoch": 0.61, "learning_rate": 1.9566255779120888e-05, "loss": 0.2882, "step": 1146 }, { "epoch": 0.61, "learning_rate": 1.956525618853702e-05, "loss": 0.2854, "step": 1147 }, { "epoch": 0.61, "learning_rate": 1.956425547306179e-05, "loss": 0.2279, "step": 1148 }, { "epoch": 0.61, "learning_rate": 1.9563253632812885e-05, "loss": 0.2558, "step": 1149 }, { "epoch": 0.61, "learning_rate": 1.9562250667908126e-05, "loss": 0.2544, "step": 1150 }, { "epoch": 0.61, "learning_rate": 1.9561246578465462e-05, "loss": 0.1571, "step": 1151 }, { "epoch": 0.61, "learning_rate": 1.9560241364602972e-05, "loss": 0.2617, "step": 1152 }, { "epoch": 0.61, "learning_rate": 1.9559235026438876e-05, "loss": 0.2714, "step": 1153 }, { "epoch": 0.61, "learning_rate": 1.9558227564091517e-05, "loss": 0.2637, "step": 1154 }, { "epoch": 0.61, "learning_rate": 1.955721897767938e-05, "loss": 0.2538, "step": 1155 }, { "epoch": 0.61, "learning_rate": 1.9556209267321073e-05, "loss": 0.2045, "step": 1156 }, { "epoch": 0.61, "learning_rate": 1.955519843313534e-05, "loss": 0.2941, "step": 1157 }, { "epoch": 0.61, "learning_rate": 1.9554186475241054e-05, "loss": 0.2176, "step": 1158 }, { "epoch": 0.61, "learning_rate": 1.955317339375723e-05, "loss": 0.1546, "step": 1159 }, { "epoch": 0.61, "learning_rate": 1.9552159188803e-05, "loss": 0.2807, "step": 1160 }, { "epoch": 0.61, "learning_rate": 1.9551143860497645e-05, "loss": 0.2889, "step": 1161 }, { "epoch": 0.62, "learning_rate": 1.955012740896056e-05, "loss": 0.2632, "step": 1162 }, { "epoch": 0.62, "learning_rate": 1.954910983431129e-05, "loss": 0.282, "step": 1163 }, { "epoch": 0.62, "learning_rate": 1.9548091136669497e-05, "loss": 0.227, "step": 1164 }, { "epoch": 0.62, "learning_rate": 1.9547071316154985e-05, "loss": 0.1438, "step": 1165 }, { "epoch": 0.62, "learning_rate": 1.954605037288769e-05, "loss": 0.3144, "step": 1166 }, { "epoch": 0.62, "learning_rate": 1.954502830698767e-05, "loss": 0.2189, "step": 1167 }, { "epoch": 0.62, "learning_rate": 1.9544005118575127e-05, "loss": 0.2347, "step": 1168 }, { "epoch": 0.62, "learning_rate": 1.954298080777039e-05, "loss": 0.2196, "step": 1169 }, { "epoch": 0.62, "learning_rate": 1.9541955374693912e-05, "loss": 0.1557, "step": 1170 }, { "epoch": 0.62, "learning_rate": 1.95409288194663e-05, "loss": 0.2376, "step": 1171 }, { "epoch": 0.62, "learning_rate": 1.9539901142208262e-05, "loss": 0.2249, "step": 1172 }, { "epoch": 0.62, "learning_rate": 1.9538872343040663e-05, "loss": 0.2425, "step": 1173 }, { "epoch": 0.62, "learning_rate": 1.9537842422084496e-05, "loss": 0.1308, "step": 1174 }, { "epoch": 0.62, "learning_rate": 1.9536811379460876e-05, "loss": 0.2313, "step": 1175 }, { "epoch": 0.62, "learning_rate": 1.9535779215291055e-05, "loss": 0.2905, "step": 1176 }, { "epoch": 0.62, "learning_rate": 1.9534745929696418e-05, "loss": 0.248, "step": 1177 }, { "epoch": 0.62, "learning_rate": 1.9533711522798485e-05, "loss": 0.2744, "step": 1178 }, { "epoch": 0.62, "learning_rate": 1.95326759947189e-05, "loss": 0.1781, "step": 1179 }, { "epoch": 0.62, "learning_rate": 1.9531639345579445e-05, "loss": 0.2285, "step": 1180 }, { "epoch": 0.63, "learning_rate": 1.953060157550203e-05, "loss": 0.1916, "step": 1181 }, { "epoch": 0.63, "learning_rate": 1.95295626846087e-05, "loss": 0.39, "step": 1182 }, { "epoch": 0.63, "learning_rate": 1.9528522673021634e-05, "loss": 0.2644, "step": 1183 }, { "epoch": 0.63, "learning_rate": 1.9527481540863128e-05, "loss": 0.3036, "step": 1184 }, { "epoch": 0.63, "learning_rate": 1.9526439288255636e-05, "loss": 0.1886, "step": 1185 }, { "epoch": 0.63, "learning_rate": 1.9525395915321713e-05, "loss": 0.2658, "step": 1186 }, { "epoch": 0.63, "learning_rate": 1.9524351422184072e-05, "loss": 0.3219, "step": 1187 }, { "epoch": 0.63, "learning_rate": 1.952330580896555e-05, "loss": 0.286, "step": 1188 }, { "epoch": 0.63, "learning_rate": 1.9522259075789102e-05, "loss": 0.1122, "step": 1189 }, { "epoch": 0.63, "learning_rate": 1.9521211222777836e-05, "loss": 0.2777, "step": 1190 }, { "epoch": 0.63, "learning_rate": 1.9520162250054974e-05, "loss": 0.138, "step": 1191 }, { "epoch": 0.63, "learning_rate": 1.951911215774388e-05, "loss": 0.3283, "step": 1192 }, { "epoch": 0.63, "learning_rate": 1.951806094596805e-05, "loss": 0.2576, "step": 1193 }, { "epoch": 0.63, "learning_rate": 1.9517008614851108e-05, "loss": 0.2784, "step": 1194 }, { "epoch": 0.63, "learning_rate": 1.95159551645168e-05, "loss": 0.2882, "step": 1195 }, { "epoch": 0.63, "learning_rate": 1.9514900595089026e-05, "loss": 0.231, "step": 1196 }, { "epoch": 0.63, "learning_rate": 1.95138449066918e-05, "loss": 0.2591, "step": 1197 }, { "epoch": 0.63, "learning_rate": 1.9512788099449272e-05, "loss": 0.2427, "step": 1198 }, { "epoch": 0.63, "learning_rate": 1.9511730173485724e-05, "loss": 0.2306, "step": 1199 }, { "epoch": 0.64, "learning_rate": 1.9510671128925574e-05, "loss": 0.2023, "step": 1200 }, { "epoch": 0.64, "learning_rate": 1.950961096589336e-05, "loss": 0.2781, "step": 1201 }, { "epoch": 0.64, "learning_rate": 1.950854968451377e-05, "loss": 0.1792, "step": 1202 }, { "epoch": 0.64, "learning_rate": 1.9507487284911604e-05, "loss": 0.1428, "step": 1203 }, { "epoch": 0.64, "learning_rate": 1.9506423767211808e-05, "loss": 0.2511, "step": 1204 }, { "epoch": 0.64, "learning_rate": 1.9505359131539445e-05, "loss": 0.2125, "step": 1205 }, { "epoch": 0.64, "learning_rate": 1.950429337801973e-05, "loss": 0.23, "step": 1206 }, { "epoch": 0.64, "learning_rate": 1.9503226506777987e-05, "loss": 0.3689, "step": 1207 }, { "epoch": 0.64, "learning_rate": 1.9502158517939688e-05, "loss": 0.115, "step": 1208 }, { "epoch": 0.64, "learning_rate": 1.9501089411630425e-05, "loss": 0.187, "step": 1209 }, { "epoch": 0.64, "learning_rate": 1.9500019187975935e-05, "loss": 0.2389, "step": 1210 }, { "epoch": 0.64, "learning_rate": 1.949894784710207e-05, "loss": 0.1993, "step": 1211 }, { "epoch": 0.64, "learning_rate": 1.9497875389134827e-05, "loss": 0.2884, "step": 1212 }, { "epoch": 0.64, "learning_rate": 1.9496801814200326e-05, "loss": 0.253, "step": 1213 }, { "epoch": 0.64, "learning_rate": 1.9495727122424824e-05, "loss": 0.1692, "step": 1214 }, { "epoch": 0.64, "learning_rate": 1.9494651313934704e-05, "loss": 0.2256, "step": 1215 }, { "epoch": 0.64, "learning_rate": 1.9493574388856483e-05, "loss": 0.2942, "step": 1216 }, { "epoch": 0.64, "learning_rate": 1.9492496347316813e-05, "loss": 0.1315, "step": 1217 }, { "epoch": 0.64, "learning_rate": 1.949141718944247e-05, "loss": 0.1859, "step": 1218 }, { "epoch": 0.65, "learning_rate": 1.9490336915360365e-05, "loss": 0.2195, "step": 1219 }, { "epoch": 0.65, "learning_rate": 1.9489255525197544e-05, "loss": 0.246, "step": 1220 }, { "epoch": 0.65, "learning_rate": 1.948817301908118e-05, "loss": 0.1878, "step": 1221 }, { "epoch": 0.65, "learning_rate": 1.948708939713857e-05, "loss": 0.3291, "step": 1222 }, { "epoch": 0.65, "learning_rate": 1.9486004659497156e-05, "loss": 0.334, "step": 1223 }, { "epoch": 0.65, "learning_rate": 1.9484918806284506e-05, "loss": 0.2814, "step": 1224 }, { "epoch": 0.65, "learning_rate": 1.9483831837628316e-05, "loss": 0.1316, "step": 1225 }, { "epoch": 0.65, "learning_rate": 1.9482743753656415e-05, "loss": 0.1793, "step": 1226 }, { "epoch": 0.65, "learning_rate": 1.9481654554496764e-05, "loss": 0.3554, "step": 1227 }, { "epoch": 0.65, "learning_rate": 1.9480564240277456e-05, "loss": 0.2298, "step": 1228 }, { "epoch": 0.65, "learning_rate": 1.9479472811126715e-05, "loss": 0.2168, "step": 1229 }, { "epoch": 0.65, "learning_rate": 1.9478380267172892e-05, "loss": 0.2202, "step": 1230 }, { "epoch": 0.65, "learning_rate": 1.9477286608544477e-05, "loss": 0.2331, "step": 1231 }, { "epoch": 0.65, "learning_rate": 1.947619183537008e-05, "loss": 0.2161, "step": 1232 }, { "epoch": 0.65, "learning_rate": 1.9475095947778452e-05, "loss": 0.1463, "step": 1233 }, { "epoch": 0.65, "learning_rate": 1.947399894589847e-05, "loss": 0.1232, "step": 1234 }, { "epoch": 0.65, "learning_rate": 1.9472900829859145e-05, "loss": 0.2919, "step": 1235 }, { "epoch": 0.65, "learning_rate": 1.947180159978962e-05, "loss": 0.1745, "step": 1236 }, { "epoch": 0.65, "learning_rate": 1.9470701255819152e-05, "loss": 0.2978, "step": 1237 }, { "epoch": 0.66, "learning_rate": 1.946959979807716e-05, "loss": 0.2326, "step": 1238 }, { "epoch": 0.66, "learning_rate": 1.9468497226693176e-05, "loss": 0.2565, "step": 1239 }, { "epoch": 0.66, "learning_rate": 1.9467393541796856e-05, "loss": 0.3535, "step": 1240 }, { "epoch": 0.66, "learning_rate": 1.9466288743517997e-05, "loss": 0.245, "step": 1241 }, { "epoch": 0.66, "learning_rate": 1.946518283198653e-05, "loss": 0.2595, "step": 1242 }, { "epoch": 0.66, "learning_rate": 1.9464075807332514e-05, "loss": 0.2275, "step": 1243 }, { "epoch": 0.66, "learning_rate": 1.9462967669686125e-05, "loss": 0.2255, "step": 1244 }, { "epoch": 0.66, "learning_rate": 1.9461858419177692e-05, "loss": 0.1869, "step": 1245 }, { "epoch": 0.66, "learning_rate": 1.9460748055937666e-05, "loss": 0.1723, "step": 1246 }, { "epoch": 0.66, "learning_rate": 1.9459636580096624e-05, "loss": 0.2814, "step": 1247 }, { "epoch": 0.66, "learning_rate": 1.9458523991785277e-05, "loss": 0.1526, "step": 1248 }, { "epoch": 0.66, "learning_rate": 1.9457410291134465e-05, "loss": 0.242, "step": 1249 }, { "epoch": 0.66, "learning_rate": 1.945629547827517e-05, "loss": 0.1691, "step": 1250 }, { "epoch": 0.66, "learning_rate": 1.9455179553338488e-05, "loss": 0.2316, "step": 1251 }, { "epoch": 0.66, "learning_rate": 1.9454062516455656e-05, "loss": 0.1382, "step": 1252 }, { "epoch": 0.66, "learning_rate": 1.945294436775804e-05, "loss": 0.402, "step": 1253 }, { "epoch": 0.66, "learning_rate": 1.9451825107377134e-05, "loss": 0.2767, "step": 1254 }, { "epoch": 0.66, "learning_rate": 1.9450704735444574e-05, "loss": 0.3264, "step": 1255 }, { "epoch": 0.66, "learning_rate": 1.9449583252092107e-05, "loss": 0.2093, "step": 1256 }, { "epoch": 0.67, "learning_rate": 1.9448460657451625e-05, "loss": 0.4007, "step": 1257 }, { "epoch": 0.67, "learning_rate": 1.9447336951655148e-05, "loss": 0.2115, "step": 1258 }, { "epoch": 0.67, "learning_rate": 1.9446212134834824e-05, "loss": 0.0645, "step": 1259 }, { "epoch": 0.67, "learning_rate": 1.9445086207122937e-05, "loss": 0.1686, "step": 1260 }, { "epoch": 0.67, "learning_rate": 1.9443959168651893e-05, "loss": 0.3036, "step": 1261 }, { "epoch": 0.67, "learning_rate": 1.944283101955424e-05, "loss": 0.136, "step": 1262 }, { "epoch": 0.67, "learning_rate": 1.944170175996265e-05, "loss": 0.1231, "step": 1263 }, { "epoch": 0.67, "learning_rate": 1.9440571390009916e-05, "loss": 0.3368, "step": 1264 }, { "epoch": 0.67, "learning_rate": 1.9439439909828984e-05, "loss": 0.2634, "step": 1265 }, { "epoch": 0.67, "learning_rate": 1.943830731955291e-05, "loss": 0.1939, "step": 1266 }, { "epoch": 0.67, "learning_rate": 1.9437173619314892e-05, "loss": 0.1925, "step": 1267 }, { "epoch": 0.67, "learning_rate": 1.9436038809248256e-05, "loss": 0.3684, "step": 1268 }, { "epoch": 0.67, "learning_rate": 1.9434902889486457e-05, "loss": 0.4294, "step": 1269 }, { "epoch": 0.67, "learning_rate": 1.9433765860163077e-05, "loss": 0.241, "step": 1270 }, { "epoch": 0.67, "learning_rate": 1.943262772141184e-05, "loss": 0.247, "step": 1271 }, { "epoch": 0.67, "learning_rate": 1.9431488473366588e-05, "loss": 0.1654, "step": 1272 }, { "epoch": 0.67, "learning_rate": 1.94303481161613e-05, "loss": 0.1558, "step": 1273 }, { "epoch": 0.67, "learning_rate": 1.9429206649930085e-05, "loss": 0.3138, "step": 1274 }, { "epoch": 0.67, "learning_rate": 1.9428064074807178e-05, "loss": 0.2506, "step": 1275 }, { "epoch": 0.68, "learning_rate": 1.9426920390926952e-05, "loss": 0.2667, "step": 1276 }, { "epoch": 0.68, "learning_rate": 1.9425775598423903e-05, "loss": 0.1473, "step": 1277 }, { "epoch": 0.68, "learning_rate": 1.9424629697432668e-05, "loss": 0.1367, "step": 1278 }, { "epoch": 0.68, "learning_rate": 1.9423482688087996e-05, "loss": 0.2106, "step": 1279 }, { "epoch": 0.68, "learning_rate": 1.9422334570524783e-05, "loss": 0.2063, "step": 1280 }, { "epoch": 0.68, "learning_rate": 1.9421185344878052e-05, "loss": 0.1977, "step": 1281 }, { "epoch": 0.68, "learning_rate": 1.9420035011282953e-05, "loss": 0.2218, "step": 1282 }, { "epoch": 0.68, "learning_rate": 1.9418883569874763e-05, "loss": 0.1182, "step": 1283 }, { "epoch": 0.68, "learning_rate": 1.9417731020788896e-05, "loss": 0.2592, "step": 1284 }, { "epoch": 0.68, "learning_rate": 1.9416577364160894e-05, "loss": 0.2345, "step": 1285 }, { "epoch": 0.68, "learning_rate": 1.941542260012643e-05, "loss": 0.2426, "step": 1286 }, { "epoch": 0.68, "learning_rate": 1.9414266728821305e-05, "loss": 0.1643, "step": 1287 }, { "epoch": 0.68, "learning_rate": 1.9413109750381457e-05, "loss": 0.2082, "step": 1288 }, { "epoch": 0.68, "learning_rate": 1.9411951664942936e-05, "loss": 0.2421, "step": 1289 }, { "epoch": 0.68, "learning_rate": 1.9410792472641947e-05, "loss": 0.24, "step": 1290 }, { "epoch": 0.68, "learning_rate": 1.940963217361481e-05, "loss": 0.2385, "step": 1291 }, { "epoch": 0.68, "learning_rate": 1.9408470767997977e-05, "loss": 0.2667, "step": 1292 }, { "epoch": 0.68, "learning_rate": 1.9407308255928032e-05, "loss": 0.2387, "step": 1293 }, { "epoch": 0.68, "learning_rate": 1.9406144637541686e-05, "loss": 0.259, "step": 1294 }, { "epoch": 0.69, "learning_rate": 1.9404979912975788e-05, "loss": 0.3143, "step": 1295 }, { "epoch": 0.69, "learning_rate": 1.9403814082367307e-05, "loss": 0.2028, "step": 1296 }, { "epoch": 0.69, "learning_rate": 1.9402647145853346e-05, "loss": 0.2931, "step": 1297 }, { "epoch": 0.69, "learning_rate": 1.9401479103571143e-05, "loss": 0.13, "step": 1298 }, { "epoch": 0.69, "learning_rate": 1.940030995565806e-05, "loss": 0.1537, "step": 1299 }, { "epoch": 0.69, "learning_rate": 1.9399139702251593e-05, "loss": 0.2832, "step": 1300 }, { "epoch": 0.69, "learning_rate": 1.9397968343489362e-05, "loss": 0.2694, "step": 1301 }, { "epoch": 0.69, "learning_rate": 1.9396795879509122e-05, "loss": 0.1629, "step": 1302 }, { "epoch": 0.69, "learning_rate": 1.9395622310448757e-05, "loss": 0.1833, "step": 1303 }, { "epoch": 0.69, "learning_rate": 1.9394447636446286e-05, "loss": 0.1858, "step": 1304 }, { "epoch": 0.69, "learning_rate": 1.9393271857639843e-05, "loss": 0.2814, "step": 1305 }, { "epoch": 0.69, "learning_rate": 1.939209497416771e-05, "loss": 0.3063, "step": 1306 }, { "epoch": 0.69, "learning_rate": 1.9390916986168285e-05, "loss": 0.1975, "step": 1307 }, { "epoch": 0.69, "learning_rate": 1.9389737893780106e-05, "loss": 0.2183, "step": 1308 }, { "epoch": 0.69, "learning_rate": 1.9388557697141836e-05, "loss": 0.1671, "step": 1309 }, { "epoch": 0.69, "learning_rate": 1.938737639639226e-05, "loss": 0.2644, "step": 1310 }, { "epoch": 0.69, "learning_rate": 1.9386193991670313e-05, "loss": 0.2739, "step": 1311 }, { "epoch": 0.69, "learning_rate": 1.9385010483115042e-05, "loss": 0.2603, "step": 1312 }, { "epoch": 0.69, "learning_rate": 1.938382587086563e-05, "loss": 0.1023, "step": 1313 }, { "epoch": 0.7, "learning_rate": 1.938264015506139e-05, "loss": 0.246, "step": 1314 }, { "epoch": 0.7, "learning_rate": 1.9381453335841766e-05, "loss": 0.2591, "step": 1315 }, { "epoch": 0.7, "learning_rate": 1.9380265413346328e-05, "loss": 0.2055, "step": 1316 }, { "epoch": 0.7, "learning_rate": 1.9379076387714776e-05, "loss": 0.2952, "step": 1317 }, { "epoch": 0.7, "learning_rate": 1.9377886259086943e-05, "loss": 0.2137, "step": 1318 }, { "epoch": 0.7, "learning_rate": 1.9376695027602795e-05, "loss": 0.2033, "step": 1319 }, { "epoch": 0.7, "learning_rate": 1.9375502693402417e-05, "loss": 0.284, "step": 1320 }, { "epoch": 0.7, "learning_rate": 1.9374309256626033e-05, "loss": 0.1042, "step": 1321 }, { "epoch": 0.7, "learning_rate": 1.9373114717413986e-05, "loss": 0.1126, "step": 1322 }, { "epoch": 0.7, "learning_rate": 1.9371919075906768e-05, "loss": 0.4085, "step": 1323 }, { "epoch": 0.7, "learning_rate": 1.937072233224498e-05, "loss": 0.1127, "step": 1324 }, { "epoch": 0.7, "learning_rate": 1.9369524486569363e-05, "loss": 0.2494, "step": 1325 }, { "epoch": 0.7, "learning_rate": 1.936832553902079e-05, "loss": 0.3111, "step": 1326 }, { "epoch": 0.7, "learning_rate": 1.9367125489740255e-05, "loss": 0.1759, "step": 1327 }, { "epoch": 0.7, "learning_rate": 1.9365924338868886e-05, "loss": 0.2598, "step": 1328 }, { "epoch": 0.7, "learning_rate": 1.9364722086547944e-05, "loss": 0.3083, "step": 1329 }, { "epoch": 0.7, "learning_rate": 1.936351873291881e-05, "loss": 0.2815, "step": 1330 }, { "epoch": 0.7, "learning_rate": 1.9362314278123e-05, "loss": 0.273, "step": 1331 }, { "epoch": 0.71, "learning_rate": 1.936110872230217e-05, "loss": 0.2435, "step": 1332 }, { "epoch": 0.71, "learning_rate": 1.935990206559809e-05, "loss": 0.1684, "step": 1333 }, { "epoch": 0.71, "learning_rate": 1.935869430815266e-05, "loss": 0.1997, "step": 1334 }, { "epoch": 0.71, "learning_rate": 1.9357485450107926e-05, "loss": 0.2982, "step": 1335 }, { "epoch": 0.71, "learning_rate": 1.9356275491606042e-05, "loss": 0.1853, "step": 1336 }, { "epoch": 0.71, "learning_rate": 1.9355064432789298e-05, "loss": 0.2232, "step": 1337 }, { "epoch": 0.71, "learning_rate": 1.935385227380013e-05, "loss": 0.1443, "step": 1338 }, { "epoch": 0.71, "learning_rate": 1.9352639014781082e-05, "loss": 0.187, "step": 1339 }, { "epoch": 0.71, "learning_rate": 1.9351424655874834e-05, "loss": 0.169, "step": 1340 }, { "epoch": 0.71, "learning_rate": 1.9350209197224203e-05, "loss": 0.2849, "step": 1341 }, { "epoch": 0.71, "learning_rate": 1.9348992638972123e-05, "loss": 0.2824, "step": 1342 }, { "epoch": 0.71, "learning_rate": 1.9347774981261666e-05, "loss": 0.1937, "step": 1343 }, { "epoch": 0.71, "learning_rate": 1.934655622423603e-05, "loss": 0.2285, "step": 1344 }, { "epoch": 0.71, "learning_rate": 1.9345336368038544e-05, "loss": 0.2055, "step": 1345 }, { "epoch": 0.71, "learning_rate": 1.934411541281266e-05, "loss": 0.2019, "step": 1346 }, { "epoch": 0.71, "learning_rate": 1.9342893358701978e-05, "loss": 0.2124, "step": 1347 }, { "epoch": 0.71, "learning_rate": 1.9341670205850202e-05, "loss": 0.0712, "step": 1348 }, { "epoch": 0.71, "learning_rate": 1.934044595440118e-05, "loss": 0.1095, "step": 1349 }, { "epoch": 0.71, "learning_rate": 1.9339220604498884e-05, "loss": 0.2891, "step": 1350 }, { "epoch": 0.72, "learning_rate": 1.9337994156287422e-05, "loss": 0.2315, "step": 1351 }, { "epoch": 0.72, "learning_rate": 1.933676660991102e-05, "loss": 0.1481, "step": 1352 }, { "epoch": 0.72, "learning_rate": 1.933553796551405e-05, "loss": 0.1738, "step": 1353 }, { "epoch": 0.72, "learning_rate": 1.9334308223240992e-05, "loss": 0.2271, "step": 1354 }, { "epoch": 0.72, "learning_rate": 1.9333077383236472e-05, "loss": 0.2408, "step": 1355 }, { "epoch": 0.72, "learning_rate": 1.933184544564524e-05, "loss": 0.0858, "step": 1356 }, { "epoch": 0.72, "learning_rate": 1.9330612410612168e-05, "loss": 0.2023, "step": 1357 }, { "epoch": 0.72, "learning_rate": 1.9329378278282268e-05, "loss": 0.237, "step": 1358 }, { "epoch": 0.72, "learning_rate": 1.9328143048800675e-05, "loss": 0.2599, "step": 1359 }, { "epoch": 0.72, "learning_rate": 1.9326906722312658e-05, "loss": 0.2816, "step": 1360 }, { "epoch": 0.72, "learning_rate": 1.93256692989636e-05, "loss": 0.129, "step": 1361 }, { "epoch": 0.72, "learning_rate": 1.9324430778899036e-05, "loss": 0.1766, "step": 1362 }, { "epoch": 0.72, "learning_rate": 1.9323191162264615e-05, "loss": 0.1471, "step": 1363 }, { "epoch": 0.72, "learning_rate": 1.9321950449206114e-05, "loss": 0.1175, "step": 1364 }, { "epoch": 0.72, "learning_rate": 1.932070863986945e-05, "loss": 0.3068, "step": 1365 }, { "epoch": 0.72, "learning_rate": 1.9319465734400653e-05, "loss": 0.1385, "step": 1366 }, { "epoch": 0.72, "learning_rate": 1.9318221732945898e-05, "loss": 0.2332, "step": 1367 }, { "epoch": 0.72, "learning_rate": 1.931697663565148e-05, "loss": 0.1461, "step": 1368 }, { "epoch": 0.72, "learning_rate": 1.9315730442663827e-05, "loss": 0.3662, "step": 1369 }, { "epoch": 0.73, "learning_rate": 1.931448315412949e-05, "loss": 0.1823, "step": 1370 }, { "epoch": 0.73, "learning_rate": 1.931323477019515e-05, "loss": 0.2806, "step": 1371 }, { "epoch": 0.73, "learning_rate": 1.9311985291007625e-05, "loss": 0.1355, "step": 1372 }, { "epoch": 0.73, "learning_rate": 1.9310734716713853e-05, "loss": 0.2709, "step": 1373 }, { "epoch": 0.73, "learning_rate": 1.9309483047460905e-05, "loss": 0.2423, "step": 1374 }, { "epoch": 0.73, "learning_rate": 1.9308230283395975e-05, "loss": 0.3683, "step": 1375 }, { "epoch": 0.73, "learning_rate": 1.93069764246664e-05, "loss": 0.1169, "step": 1376 }, { "epoch": 0.73, "learning_rate": 1.930572147141963e-05, "loss": 0.1897, "step": 1377 }, { "epoch": 0.73, "learning_rate": 1.9304465423803247e-05, "loss": 0.2888, "step": 1378 }, { "epoch": 0.73, "learning_rate": 1.930320828196497e-05, "loss": 0.2355, "step": 1379 }, { "epoch": 0.73, "learning_rate": 1.9301950046052635e-05, "loss": 0.2062, "step": 1380 }, { "epoch": 0.73, "learning_rate": 1.930069071621422e-05, "loss": 0.2082, "step": 1381 }, { "epoch": 0.73, "learning_rate": 1.929943029259782e-05, "loss": 0.1881, "step": 1382 }, { "epoch": 0.73, "learning_rate": 1.9298168775351663e-05, "loss": 0.273, "step": 1383 }, { "epoch": 0.73, "learning_rate": 1.9296906164624107e-05, "loss": 0.2492, "step": 1384 }, { "epoch": 0.73, "learning_rate": 1.929564246056364e-05, "loss": 0.1773, "step": 1385 }, { "epoch": 0.73, "learning_rate": 1.929437766331887e-05, "loss": 0.319, "step": 1386 }, { "epoch": 0.73, "learning_rate": 1.929311177303854e-05, "loss": 0.1396, "step": 1387 }, { "epoch": 0.73, "learning_rate": 1.9291844789871526e-05, "loss": 0.3044, "step": 1388 }, { "epoch": 0.74, "learning_rate": 1.9290576713966828e-05, "loss": 0.1323, "step": 1389 }, { "epoch": 0.74, "learning_rate": 1.9289307545473566e-05, "loss": 0.2148, "step": 1390 }, { "epoch": 0.74, "learning_rate": 1.9288037284541e-05, "loss": 0.2116, "step": 1391 }, { "epoch": 0.74, "learning_rate": 1.928676593131852e-05, "loss": 0.1855, "step": 1392 }, { "epoch": 0.74, "learning_rate": 1.9285493485955634e-05, "loss": 0.1948, "step": 1393 }, { "epoch": 0.74, "learning_rate": 1.928421994860199e-05, "loss": 0.1949, "step": 1394 }, { "epoch": 0.74, "learning_rate": 1.928294531940735e-05, "loss": 0.2034, "step": 1395 }, { "epoch": 0.74, "learning_rate": 1.9281669598521616e-05, "loss": 0.1506, "step": 1396 }, { "epoch": 0.74, "learning_rate": 1.9280392786094816e-05, "loss": 0.3205, "step": 1397 }, { "epoch": 0.74, "learning_rate": 1.9279114882277105e-05, "loss": 0.2807, "step": 1398 }, { "epoch": 0.74, "learning_rate": 1.927783588721877e-05, "loss": 0.1097, "step": 1399 }, { "epoch": 0.74, "learning_rate": 1.9276555801070217e-05, "loss": 0.2422, "step": 1400 }, { "epoch": 0.74, "learning_rate": 1.927527462398199e-05, "loss": 0.3132, "step": 1401 }, { "epoch": 0.74, "learning_rate": 1.9273992356104757e-05, "loss": 0.2219, "step": 1402 }, { "epoch": 0.74, "learning_rate": 1.9272708997589314e-05, "loss": 0.1431, "step": 1403 }, { "epoch": 0.74, "learning_rate": 1.927142454858659e-05, "loss": 0.1785, "step": 1404 }, { "epoch": 0.74, "learning_rate": 1.9270139009247635e-05, "loss": 0.3382, "step": 1405 }, { "epoch": 0.74, "learning_rate": 1.9268852379723633e-05, "loss": 0.1861, "step": 1406 }, { "epoch": 0.74, "learning_rate": 1.9267564660165888e-05, "loss": 0.1984, "step": 1407 }, { "epoch": 0.75, "learning_rate": 1.9266275850725845e-05, "loss": 0.2059, "step": 1408 }, { "epoch": 0.75, "learning_rate": 1.926498595155507e-05, "loss": 0.1309, "step": 1409 }, { "epoch": 0.75, "learning_rate": 1.9263694962805255e-05, "loss": 0.1391, "step": 1410 }, { "epoch": 0.75, "learning_rate": 1.9262402884628223e-05, "loss": 0.2607, "step": 1411 }, { "epoch": 0.75, "learning_rate": 1.9261109717175924e-05, "loss": 0.2104, "step": 1412 }, { "epoch": 0.75, "learning_rate": 1.925981546060044e-05, "loss": 0.3161, "step": 1413 }, { "epoch": 0.75, "learning_rate": 1.9258520115053973e-05, "loss": 0.209, "step": 1414 }, { "epoch": 0.75, "learning_rate": 1.9257223680688862e-05, "loss": 0.3029, "step": 1415 }, { "epoch": 0.75, "learning_rate": 1.925592615765757e-05, "loss": 0.1875, "step": 1416 }, { "epoch": 0.75, "learning_rate": 1.9254627546112683e-05, "loss": 0.3341, "step": 1417 }, { "epoch": 0.75, "learning_rate": 1.9253327846206925e-05, "loss": 0.2471, "step": 1418 }, { "epoch": 0.75, "learning_rate": 1.9252027058093143e-05, "loss": 0.25, "step": 1419 }, { "epoch": 0.75, "learning_rate": 1.925072518192431e-05, "loss": 0.153, "step": 1420 }, { "epoch": 0.75, "learning_rate": 1.924942221785353e-05, "loss": 0.2234, "step": 1421 }, { "epoch": 0.75, "learning_rate": 1.9248118166034035e-05, "loss": 0.362, "step": 1422 }, { "epoch": 0.75, "learning_rate": 1.924681302661918e-05, "loss": 0.2801, "step": 1423 }, { "epoch": 0.75, "learning_rate": 1.9245506799762458e-05, "loss": 0.2309, "step": 1424 }, { "epoch": 0.75, "learning_rate": 1.924419948561748e-05, "loss": 0.1593, "step": 1425 }, { "epoch": 0.75, "learning_rate": 1.9242891084337983e-05, "loss": 0.082, "step": 1426 }, { "epoch": 0.76, "learning_rate": 1.9241581596077848e-05, "loss": 0.2125, "step": 1427 }, { "epoch": 0.76, "learning_rate": 1.9240271020991062e-05, "loss": 0.2812, "step": 1428 }, { "epoch": 0.76, "learning_rate": 1.9238959359231763e-05, "loss": 0.1827, "step": 1429 }, { "epoch": 0.76, "learning_rate": 1.92376466109542e-05, "loss": 0.3418, "step": 1430 }, { "epoch": 0.76, "learning_rate": 1.9236332776312745e-05, "loss": 0.2756, "step": 1431 }, { "epoch": 0.76, "learning_rate": 1.9235017855461922e-05, "loss": 0.1568, "step": 1432 }, { "epoch": 0.76, "learning_rate": 1.9233701848556358e-05, "loss": 0.1266, "step": 1433 }, { "epoch": 0.76, "learning_rate": 1.9232384755750824e-05, "loss": 0.3294, "step": 1434 }, { "epoch": 0.76, "learning_rate": 1.923106657720021e-05, "loss": 0.2792, "step": 1435 }, { "epoch": 0.76, "learning_rate": 1.9229747313059534e-05, "loss": 0.3072, "step": 1436 }, { "epoch": 0.76, "learning_rate": 1.922842696348395e-05, "loss": 0.2719, "step": 1437 }, { "epoch": 0.76, "learning_rate": 1.9227105528628724e-05, "loss": 0.1845, "step": 1438 }, { "epoch": 0.76, "learning_rate": 1.922578300864927e-05, "loss": 0.2762, "step": 1439 }, { "epoch": 0.76, "learning_rate": 1.922445940370111e-05, "loss": 0.2368, "step": 1440 }, { "epoch": 0.76, "learning_rate": 1.9223134713939906e-05, "loss": 0.1316, "step": 1441 }, { "epoch": 0.76, "learning_rate": 1.9221808939521445e-05, "loss": 0.4091, "step": 1442 }, { "epoch": 0.76, "learning_rate": 1.9220482080601637e-05, "loss": 0.1858, "step": 1443 }, { "epoch": 0.76, "learning_rate": 1.9219154137336524e-05, "loss": 0.2939, "step": 1444 }, { "epoch": 0.76, "learning_rate": 1.9217825109882277e-05, "loss": 0.1666, "step": 1445 }, { "epoch": 0.77, "learning_rate": 1.9216494998395192e-05, "loss": 0.314, "step": 1446 }, { "epoch": 0.77, "learning_rate": 1.921516380303169e-05, "loss": 0.2499, "step": 1447 }, { "epoch": 0.77, "learning_rate": 1.9213831523948324e-05, "loss": 0.2007, "step": 1448 }, { "epoch": 0.77, "learning_rate": 1.921249816130177e-05, "loss": 0.2009, "step": 1449 }, { "epoch": 0.77, "learning_rate": 1.9211163715248836e-05, "loss": 0.2835, "step": 1450 }, { "epoch": 0.77, "learning_rate": 1.9209828185946453e-05, "loss": 0.1989, "step": 1451 }, { "epoch": 0.77, "learning_rate": 1.9208491573551683e-05, "loss": 0.214, "step": 1452 }, { "epoch": 0.77, "learning_rate": 1.920715387822172e-05, "loss": 0.0847, "step": 1453 }, { "epoch": 0.77, "learning_rate": 1.9205815100113866e-05, "loss": 0.2672, "step": 1454 }, { "epoch": 0.77, "learning_rate": 1.9204475239385572e-05, "loss": 0.1516, "step": 1455 }, { "epoch": 0.77, "learning_rate": 1.9203134296194413e-05, "loss": 0.2207, "step": 1456 }, { "epoch": 0.77, "learning_rate": 1.9201792270698077e-05, "loss": 0.1817, "step": 1457 }, { "epoch": 0.77, "learning_rate": 1.9200449163054392e-05, "loss": 0.1222, "step": 1458 }, { "epoch": 0.77, "learning_rate": 1.919910497342131e-05, "loss": 0.253, "step": 1459 }, { "epoch": 0.77, "learning_rate": 1.9197759701956914e-05, "loss": 0.2189, "step": 1460 }, { "epoch": 0.77, "learning_rate": 1.9196413348819403e-05, "loss": 0.191, "step": 1461 }, { "epoch": 0.77, "learning_rate": 1.9195065914167115e-05, "loss": 0.1421, "step": 1462 }, { "epoch": 0.77, "learning_rate": 1.9193717398158508e-05, "loss": 0.2542, "step": 1463 }, { "epoch": 0.77, "learning_rate": 1.9192367800952175e-05, "loss": 0.2626, "step": 1464 }, { "epoch": 0.78, "learning_rate": 1.9191017122706827e-05, "loss": 0.1547, "step": 1465 }, { "epoch": 0.78, "learning_rate": 1.9189665363581308e-05, "loss": 0.149, "step": 1466 }, { "epoch": 0.78, "learning_rate": 1.9188312523734586e-05, "loss": 0.1091, "step": 1467 }, { "epoch": 0.78, "learning_rate": 1.9186958603325755e-05, "loss": 0.2462, "step": 1468 }, { "epoch": 0.78, "learning_rate": 1.9185603602514044e-05, "loss": 0.3413, "step": 1469 }, { "epoch": 0.78, "learning_rate": 1.9184247521458802e-05, "loss": 0.2129, "step": 1470 }, { "epoch": 0.78, "learning_rate": 1.9182890360319503e-05, "loss": 0.3162, "step": 1471 }, { "epoch": 0.78, "learning_rate": 1.9181532119255754e-05, "loss": 0.319, "step": 1472 }, { "epoch": 0.78, "learning_rate": 1.9180172798427286e-05, "loss": 0.1363, "step": 1473 }, { "epoch": 0.78, "learning_rate": 1.917881239799396e-05, "loss": 0.2586, "step": 1474 }, { "epoch": 0.78, "learning_rate": 1.917745091811576e-05, "loss": 0.2862, "step": 1475 }, { "epoch": 0.78, "learning_rate": 1.9176088358952798e-05, "loss": 0.3382, "step": 1476 }, { "epoch": 0.78, "learning_rate": 1.9174724720665315e-05, "loss": 0.2177, "step": 1477 }, { "epoch": 0.78, "learning_rate": 1.9173360003413674e-05, "loss": 0.3068, "step": 1478 }, { "epoch": 0.78, "learning_rate": 1.9171994207358367e-05, "loss": 0.1364, "step": 1479 }, { "epoch": 0.78, "learning_rate": 1.9170627332660023e-05, "loss": 0.1571, "step": 1480 }, { "epoch": 0.78, "learning_rate": 1.916925937947938e-05, "loss": 0.2141, "step": 1481 }, { "epoch": 0.78, "learning_rate": 1.9167890347977316e-05, "loss": 0.3066, "step": 1482 }, { "epoch": 0.78, "learning_rate": 1.916652023831483e-05, "loss": 0.1431, "step": 1483 }, { "epoch": 0.79, "learning_rate": 1.916514905065305e-05, "loss": 0.1269, "step": 1484 }, { "epoch": 0.79, "learning_rate": 1.916377678515323e-05, "loss": 0.3288, "step": 1485 }, { "epoch": 0.79, "learning_rate": 1.916240344197675e-05, "loss": 0.2117, "step": 1486 }, { "epoch": 0.79, "learning_rate": 1.916102902128512e-05, "loss": 0.3893, "step": 1487 }, { "epoch": 0.79, "learning_rate": 1.915965352323997e-05, "loss": 0.1636, "step": 1488 }, { "epoch": 0.79, "learning_rate": 1.9158276948003066e-05, "loss": 0.1647, "step": 1489 }, { "epoch": 0.79, "learning_rate": 1.9156899295736292e-05, "loss": 0.3167, "step": 1490 }, { "epoch": 0.79, "learning_rate": 1.9155520566601668e-05, "loss": 0.1799, "step": 1491 }, { "epoch": 0.79, "learning_rate": 1.915414076076133e-05, "loss": 0.1713, "step": 1492 }, { "epoch": 0.79, "learning_rate": 1.9152759878377546e-05, "loss": 0.2054, "step": 1493 }, { "epoch": 0.79, "learning_rate": 1.915137791961271e-05, "loss": 0.2042, "step": 1494 }, { "epoch": 0.79, "learning_rate": 1.9149994884629343e-05, "loss": 0.2828, "step": 1495 }, { "epoch": 0.79, "learning_rate": 1.9148610773590094e-05, "loss": 0.2048, "step": 1496 }, { "epoch": 0.79, "learning_rate": 1.9147225586657737e-05, "loss": 0.1067, "step": 1497 }, { "epoch": 0.79, "learning_rate": 1.9145839323995173e-05, "loss": 0.2051, "step": 1498 }, { "epoch": 0.79, "learning_rate": 1.9144451985765426e-05, "loss": 0.1862, "step": 1499 }, { "epoch": 0.79, "learning_rate": 1.9143063572131652e-05, "loss": 0.1832, "step": 1500 }, { "epoch": 0.79, "learning_rate": 1.9141674083257127e-05, "loss": 0.2058, "step": 1501 }, { "epoch": 0.8, "learning_rate": 1.9140283519305262e-05, "loss": 0.2469, "step": 1502 }, { "epoch": 0.8, "learning_rate": 1.913889188043959e-05, "loss": 0.3074, "step": 1503 }, { "epoch": 0.8, "learning_rate": 1.9137499166823764e-05, "loss": 0.3983, "step": 1504 }, { "epoch": 0.8, "learning_rate": 1.9136105378621578e-05, "loss": 0.2346, "step": 1505 }, { "epoch": 0.8, "learning_rate": 1.9134710515996943e-05, "loss": 0.1511, "step": 1506 }, { "epoch": 0.8, "learning_rate": 1.9133314579113892e-05, "loss": 0.1233, "step": 1507 }, { "epoch": 0.8, "learning_rate": 1.9131917568136587e-05, "loss": 0.2915, "step": 1508 }, { "epoch": 0.8, "learning_rate": 1.913051948322933e-05, "loss": 0.1368, "step": 1509 }, { "epoch": 0.8, "learning_rate": 1.912912032455653e-05, "loss": 0.2165, "step": 1510 }, { "epoch": 0.8, "learning_rate": 1.9127720092282737e-05, "loss": 0.381, "step": 1511 }, { "epoch": 0.8, "learning_rate": 1.9126318786572612e-05, "loss": 0.1338, "step": 1512 }, { "epoch": 0.8, "learning_rate": 1.9124916407590957e-05, "loss": 0.2474, "step": 1513 }, { "epoch": 0.8, "learning_rate": 1.9123512955502696e-05, "loss": 0.2137, "step": 1514 }, { "epoch": 0.8, "learning_rate": 1.9122108430472874e-05, "loss": 0.2029, "step": 1515 }, { "epoch": 0.8, "learning_rate": 1.9120702832666666e-05, "loss": 0.2653, "step": 1516 }, { "epoch": 0.8, "learning_rate": 1.9119296162249372e-05, "loss": 0.2876, "step": 1517 }, { "epoch": 0.8, "learning_rate": 1.911788841938642e-05, "loss": 0.1649, "step": 1518 }, { "epoch": 0.8, "learning_rate": 1.9116479604243365e-05, "loss": 0.157, "step": 1519 }, { "epoch": 0.8, "learning_rate": 1.9115069716985886e-05, "loss": 0.1465, "step": 1520 }, { "epoch": 0.81, "learning_rate": 1.9113658757779786e-05, "loss": 0.1798, "step": 1521 }, { "epoch": 0.81, "learning_rate": 1.9112246726790995e-05, "loss": 0.2831, "step": 1522 }, { "epoch": 0.81, "learning_rate": 1.9110833624185576e-05, "loss": 0.1643, "step": 1523 }, { "epoch": 0.81, "learning_rate": 1.9109419450129706e-05, "loss": 0.2406, "step": 1524 }, { "epoch": 0.81, "learning_rate": 1.9108004204789695e-05, "loss": 0.2538, "step": 1525 }, { "epoch": 0.81, "learning_rate": 1.9106587888331986e-05, "loss": 0.2377, "step": 1526 }, { "epoch": 0.81, "learning_rate": 1.910517050092313e-05, "loss": 0.0962, "step": 1527 }, { "epoch": 0.81, "learning_rate": 1.9103752042729824e-05, "loss": 0.3055, "step": 1528 }, { "epoch": 0.81, "learning_rate": 1.9102332513918876e-05, "loss": 0.3131, "step": 1529 }, { "epoch": 0.81, "learning_rate": 1.9100911914657224e-05, "loss": 0.2834, "step": 1530 }, { "epoch": 0.81, "learning_rate": 1.909949024511194e-05, "loss": 0.2819, "step": 1531 }, { "epoch": 0.81, "learning_rate": 1.9098067505450203e-05, "loss": 0.1905, "step": 1532 }, { "epoch": 0.81, "learning_rate": 1.9096643695839342e-05, "loss": 0.1108, "step": 1533 }, { "epoch": 0.81, "learning_rate": 1.909521881644679e-05, "loss": 0.1783, "step": 1534 }, { "epoch": 0.81, "learning_rate": 1.909379286744012e-05, "loss": 0.3113, "step": 1535 }, { "epoch": 0.81, "learning_rate": 1.909236584898703e-05, "loss": 0.1862, "step": 1536 }, { "epoch": 0.81, "learning_rate": 1.9090937761255332e-05, "loss": 0.2951, "step": 1537 }, { "epoch": 0.81, "learning_rate": 1.9089508604412973e-05, "loss": 0.1758, "step": 1538 }, { "epoch": 0.81, "learning_rate": 1.908807837862803e-05, "loss": 0.1755, "step": 1539 }, { "epoch": 0.82, "learning_rate": 1.9086647084068697e-05, "loss": 0.1984, "step": 1540 }, { "epoch": 0.82, "learning_rate": 1.9085214720903297e-05, "loss": 0.2696, "step": 1541 }, { "epoch": 0.82, "learning_rate": 1.9083781289300278e-05, "loss": 0.2242, "step": 1542 }, { "epoch": 0.82, "learning_rate": 1.9082346789428213e-05, "loss": 0.1566, "step": 1543 }, { "epoch": 0.82, "learning_rate": 1.9080911221455804e-05, "loss": 0.2886, "step": 1544 }, { "epoch": 0.82, "learning_rate": 1.9079474585551872e-05, "loss": 0.1305, "step": 1545 }, { "epoch": 0.82, "learning_rate": 1.907803688188538e-05, "loss": 0.2469, "step": 1546 }, { "epoch": 0.82, "learning_rate": 1.9076598110625387e-05, "loss": 0.216, "step": 1547 }, { "epoch": 0.82, "learning_rate": 1.907515827194111e-05, "loss": 0.2334, "step": 1548 }, { "epoch": 0.82, "learning_rate": 1.9073717366001866e-05, "loss": 0.2136, "step": 1549 }, { "epoch": 0.82, "learning_rate": 1.9072275392977116e-05, "loss": 0.2292, "step": 1550 }, { "epoch": 0.82, "learning_rate": 1.9070832353036435e-05, "loss": 0.2705, "step": 1551 }, { "epoch": 0.82, "learning_rate": 1.906938824634953e-05, "loss": 0.1986, "step": 1552 }, { "epoch": 0.82, "learning_rate": 1.9067943073086223e-05, "loss": 0.2823, "step": 1553 }, { "epoch": 0.82, "learning_rate": 1.9066496833416476e-05, "loss": 0.1043, "step": 1554 }, { "epoch": 0.82, "learning_rate": 1.906504952751037e-05, "loss": 0.2261, "step": 1555 }, { "epoch": 0.82, "learning_rate": 1.9063601155538106e-05, "loss": 0.0852, "step": 1556 }, { "epoch": 0.82, "learning_rate": 1.9062151717670018e-05, "loss": 0.2044, "step": 1557 }, { "epoch": 0.82, "learning_rate": 1.9060701214076557e-05, "loss": 0.2471, "step": 1558 }, { "epoch": 0.83, "learning_rate": 1.9059249644928317e-05, "loss": 0.3217, "step": 1559 }, { "epoch": 0.83, "learning_rate": 1.9057797010395995e-05, "loss": 0.2904, "step": 1560 }, { "epoch": 0.83, "learning_rate": 1.9056343310650428e-05, "loss": 0.1021, "step": 1561 }, { "epoch": 0.83, "learning_rate": 1.905488854586257e-05, "loss": 0.2098, "step": 1562 }, { "epoch": 0.83, "learning_rate": 1.9053432716203505e-05, "loss": 0.2554, "step": 1563 }, { "epoch": 0.83, "learning_rate": 1.9051975821844445e-05, "loss": 0.2934, "step": 1564 }, { "epoch": 0.83, "learning_rate": 1.9050517862956716e-05, "loss": 0.1774, "step": 1565 }, { "epoch": 0.83, "learning_rate": 1.9049058839711786e-05, "loss": 0.2235, "step": 1566 }, { "epoch": 0.83, "learning_rate": 1.9047598752281235e-05, "loss": 0.2022, "step": 1567 }, { "epoch": 0.83, "learning_rate": 1.9046137600836765e-05, "loss": 0.2838, "step": 1568 }, { "epoch": 0.83, "learning_rate": 1.904467538555022e-05, "loss": 0.2526, "step": 1569 }, { "epoch": 0.83, "learning_rate": 1.9043212106593554e-05, "loss": 0.1994, "step": 1570 }, { "epoch": 0.83, "learning_rate": 1.904174776413885e-05, "loss": 0.2697, "step": 1571 }, { "epoch": 0.83, "learning_rate": 1.9040282358358324e-05, "loss": 0.4012, "step": 1572 }, { "epoch": 0.83, "learning_rate": 1.9038815889424303e-05, "loss": 0.19, "step": 1573 }, { "epoch": 0.83, "learning_rate": 1.9037348357509252e-05, "loss": 0.4491, "step": 1574 }, { "epoch": 0.83, "learning_rate": 1.903587976278575e-05, "loss": 0.1787, "step": 1575 }, { "epoch": 0.83, "learning_rate": 1.903441010542651e-05, "loss": 0.2187, "step": 1576 }, { "epoch": 0.83, "learning_rate": 1.903293938560437e-05, "loss": 0.2325, "step": 1577 }, { "epoch": 0.84, "learning_rate": 1.9031467603492278e-05, "loss": 0.1681, "step": 1578 }, { "epoch": 0.84, "learning_rate": 1.9029994759263325e-05, "loss": 0.1643, "step": 1579 }, { "epoch": 0.84, "learning_rate": 1.9028520853090724e-05, "loss": 0.2562, "step": 1580 }, { "epoch": 0.84, "learning_rate": 1.9027045885147803e-05, "loss": 0.3877, "step": 1581 }, { "epoch": 0.84, "learning_rate": 1.9025569855608026e-05, "loss": 0.1911, "step": 1582 }, { "epoch": 0.84, "learning_rate": 1.902409276464497e-05, "loss": 0.1454, "step": 1583 }, { "epoch": 0.84, "learning_rate": 1.902261461243235e-05, "loss": 0.1949, "step": 1584 }, { "epoch": 0.84, "learning_rate": 1.9021135399143994e-05, "loss": 0.2717, "step": 1585 }, { "epoch": 0.84, "learning_rate": 1.9019655124953864e-05, "loss": 0.1931, "step": 1586 }, { "epoch": 0.84, "learning_rate": 1.9018173790036046e-05, "loss": 0.2156, "step": 1587 }, { "epoch": 0.84, "learning_rate": 1.9016691394564737e-05, "loss": 0.1682, "step": 1588 }, { "epoch": 0.84, "learning_rate": 1.901520793871428e-05, "loss": 0.2864, "step": 1589 }, { "epoch": 0.84, "learning_rate": 1.9013723422659124e-05, "loss": 0.1526, "step": 1590 }, { "epoch": 0.84, "learning_rate": 1.9012237846573857e-05, "loss": 0.2217, "step": 1591 }, { "epoch": 0.84, "learning_rate": 1.9010751210633184e-05, "loss": 0.1967, "step": 1592 }, { "epoch": 0.84, "learning_rate": 1.9009263515011936e-05, "loss": 0.3284, "step": 1593 }, { "epoch": 0.84, "learning_rate": 1.9007774759885068e-05, "loss": 0.2344, "step": 1594 }, { "epoch": 0.84, "learning_rate": 1.9006284945427656e-05, "loss": 0.1824, "step": 1595 }, { "epoch": 0.84, "learning_rate": 1.900479407181491e-05, "loss": 0.2684, "step": 1596 }, { "epoch": 0.85, "learning_rate": 1.900330213922216e-05, "loss": 0.2889, "step": 1597 }, { "epoch": 0.85, "learning_rate": 1.900180914782486e-05, "loss": 0.2359, "step": 1598 }, { "epoch": 0.85, "learning_rate": 1.9000315097798583e-05, "loss": 0.2007, "step": 1599 }, { "epoch": 0.85, "learning_rate": 1.8998819989319036e-05, "loss": 0.2379, "step": 1600 }, { "epoch": 0.85, "learning_rate": 1.8997323822562048e-05, "loss": 0.2615, "step": 1601 }, { "epoch": 0.85, "learning_rate": 1.899582659770357e-05, "loss": 0.1554, "step": 1602 }, { "epoch": 0.85, "learning_rate": 1.899432831491968e-05, "loss": 0.238, "step": 1603 }, { "epoch": 0.85, "learning_rate": 1.899282897438657e-05, "loss": 0.2378, "step": 1604 }, { "epoch": 0.85, "learning_rate": 1.8991328576280576e-05, "loss": 0.2303, "step": 1605 }, { "epoch": 0.85, "learning_rate": 1.898982712077814e-05, "loss": 0.3307, "step": 1606 }, { "epoch": 0.85, "learning_rate": 1.898832460805584e-05, "loss": 0.2949, "step": 1607 }, { "epoch": 0.85, "learning_rate": 1.8986821038290372e-05, "loss": 0.1521, "step": 1608 }, { "epoch": 0.85, "learning_rate": 1.898531641165856e-05, "loss": 0.2921, "step": 1609 }, { "epoch": 0.85, "learning_rate": 1.8983810728337353e-05, "loss": 0.1899, "step": 1610 }, { "epoch": 0.85, "learning_rate": 1.8982303988503815e-05, "loss": 0.2143, "step": 1611 }, { "epoch": 0.85, "learning_rate": 1.898079619233515e-05, "loss": 0.208, "step": 1612 }, { "epoch": 0.85, "learning_rate": 1.897928734000867e-05, "loss": 0.2674, "step": 1613 }, { "epoch": 0.85, "learning_rate": 1.8977777431701824e-05, "loss": 0.1706, "step": 1614 }, { "epoch": 0.85, "learning_rate": 1.8976266467592175e-05, "loss": 0.1362, "step": 1615 }, { "epoch": 0.86, "learning_rate": 1.8974754447857423e-05, "loss": 0.2654, "step": 1616 }, { "epoch": 0.86, "learning_rate": 1.8973241372675373e-05, "loss": 0.1917, "step": 1617 }, { "epoch": 0.86, "learning_rate": 1.8971727242223976e-05, "loss": 0.2, "step": 1618 }, { "epoch": 0.86, "learning_rate": 1.897021205668129e-05, "loss": 0.2222, "step": 1619 }, { "epoch": 0.86, "learning_rate": 1.896869581622551e-05, "loss": 0.2897, "step": 1620 }, { "epoch": 0.86, "learning_rate": 1.896717852103494e-05, "loss": 0.1937, "step": 1621 }, { "epoch": 0.86, "learning_rate": 1.896566017128802e-05, "loss": 0.2557, "step": 1622 }, { "epoch": 0.86, "learning_rate": 1.8964140767163313e-05, "loss": 0.2436, "step": 1623 }, { "epoch": 0.86, "learning_rate": 1.8962620308839507e-05, "loss": 0.2878, "step": 1624 }, { "epoch": 0.86, "learning_rate": 1.8961098796495402e-05, "loss": 0.242, "step": 1625 }, { "epoch": 0.86, "learning_rate": 1.8959576230309934e-05, "loss": 0.2904, "step": 1626 }, { "epoch": 0.86, "learning_rate": 1.895805261046216e-05, "loss": 0.2657, "step": 1627 }, { "epoch": 0.86, "learning_rate": 1.895652793713126e-05, "loss": 0.187, "step": 1628 }, { "epoch": 0.86, "learning_rate": 1.8955002210496537e-05, "loss": 0.2951, "step": 1629 }, { "epoch": 0.86, "learning_rate": 1.895347543073743e-05, "loss": 0.1905, "step": 1630 }, { "epoch": 0.86, "learning_rate": 1.895194759803347e-05, "loss": 0.2403, "step": 1631 }, { "epoch": 0.86, "learning_rate": 1.8950418712564353e-05, "loss": 0.2843, "step": 1632 }, { "epoch": 0.86, "learning_rate": 1.8948888774509866e-05, "loss": 0.2434, "step": 1633 }, { "epoch": 0.86, "learning_rate": 1.894735778404994e-05, "loss": 0.2887, "step": 1634 }, { "epoch": 0.87, "learning_rate": 1.8945825741364617e-05, "loss": 0.1564, "step": 1635 }, { "epoch": 0.87, "learning_rate": 1.8944292646634072e-05, "loss": 0.1768, "step": 1636 }, { "epoch": 0.87, "learning_rate": 1.89427585000386e-05, "loss": 0.2596, "step": 1637 }, { "epoch": 0.87, "learning_rate": 1.8941223301758612e-05, "loss": 0.2368, "step": 1638 }, { "epoch": 0.87, "learning_rate": 1.893968705197466e-05, "loss": 0.1647, "step": 1639 }, { "epoch": 0.87, "learning_rate": 1.8938149750867404e-05, "loss": 0.1835, "step": 1640 }, { "epoch": 0.87, "learning_rate": 1.8936611398617636e-05, "loss": 0.2634, "step": 1641 }, { "epoch": 0.87, "learning_rate": 1.8935071995406264e-05, "loss": 0.2078, "step": 1642 }, { "epoch": 0.87, "learning_rate": 1.8933531541414328e-05, "loss": 0.2841, "step": 1643 }, { "epoch": 0.87, "learning_rate": 1.8931990036822992e-05, "loss": 0.3286, "step": 1644 }, { "epoch": 0.87, "learning_rate": 1.8930447481813536e-05, "loss": 0.1522, "step": 1645 }, { "epoch": 0.87, "learning_rate": 1.8928903876567366e-05, "loss": 0.1467, "step": 1646 }, { "epoch": 0.87, "learning_rate": 1.8927359221266015e-05, "loss": 0.1777, "step": 1647 }, { "epoch": 0.87, "learning_rate": 1.8925813516091134e-05, "loss": 0.2074, "step": 1648 }, { "epoch": 0.87, "learning_rate": 1.8924266761224506e-05, "loss": 0.2809, "step": 1649 }, { "epoch": 0.87, "learning_rate": 1.8922718956848028e-05, "loss": 0.2763, "step": 1650 }, { "epoch": 0.87, "learning_rate": 1.8921170103143726e-05, "loss": 0.0915, "step": 1651 }, { "epoch": 0.87, "learning_rate": 1.8919620200293746e-05, "loss": 0.3159, "step": 1652 }, { "epoch": 0.87, "learning_rate": 1.891806924848036e-05, "loss": 0.1789, "step": 1653 }, { "epoch": 0.88, "learning_rate": 1.8916517247885968e-05, "loss": 0.3104, "step": 1654 }, { "epoch": 0.88, "learning_rate": 1.891496419869308e-05, "loss": 0.1355, "step": 1655 }, { "epoch": 0.88, "learning_rate": 1.8913410101084344e-05, "loss": 0.2456, "step": 1656 }, { "epoch": 0.88, "learning_rate": 1.8911854955242523e-05, "loss": 0.1344, "step": 1657 }, { "epoch": 0.88, "learning_rate": 1.89102987613505e-05, "loss": 0.3448, "step": 1658 }, { "epoch": 0.88, "learning_rate": 1.8908741519591297e-05, "loss": 0.2098, "step": 1659 }, { "epoch": 0.88, "learning_rate": 1.8907183230148035e-05, "loss": 0.1523, "step": 1660 }, { "epoch": 0.88, "learning_rate": 1.8905623893203978e-05, "loss": 0.1222, "step": 1661 }, { "epoch": 0.88, "learning_rate": 1.890406350894251e-05, "loss": 0.1947, "step": 1662 }, { "epoch": 0.88, "learning_rate": 1.8902502077547133e-05, "loss": 0.1568, "step": 1663 }, { "epoch": 0.88, "learning_rate": 1.8900939599201472e-05, "loss": 0.2382, "step": 1664 }, { "epoch": 0.88, "learning_rate": 1.8899376074089276e-05, "loss": 0.1767, "step": 1665 }, { "epoch": 0.88, "learning_rate": 1.8897811502394425e-05, "loss": 0.1949, "step": 1666 }, { "epoch": 0.88, "learning_rate": 1.889624588430091e-05, "loss": 0.2776, "step": 1667 }, { "epoch": 0.88, "learning_rate": 1.8894679219992847e-05, "loss": 0.2298, "step": 1668 }, { "epoch": 0.88, "learning_rate": 1.889311150965449e-05, "loss": 0.2832, "step": 1669 }, { "epoch": 0.88, "learning_rate": 1.8891542753470195e-05, "loss": 0.3042, "step": 1670 }, { "epoch": 0.88, "learning_rate": 1.8889972951624454e-05, "loss": 0.2299, "step": 1671 }, { "epoch": 0.89, "learning_rate": 1.8888402104301878e-05, "loss": 0.3395, "step": 1672 }, { "epoch": 0.89, "learning_rate": 1.88868302116872e-05, "loss": 0.2277, "step": 1673 }, { "epoch": 0.89, "learning_rate": 1.888525727396528e-05, "loss": 0.2594, "step": 1674 }, { "epoch": 0.89, "learning_rate": 1.88836832913211e-05, "loss": 0.2661, "step": 1675 }, { "epoch": 0.89, "learning_rate": 1.8882108263939754e-05, "loss": 0.1913, "step": 1676 }, { "epoch": 0.89, "learning_rate": 1.888053219200648e-05, "loss": 0.3667, "step": 1677 }, { "epoch": 0.89, "learning_rate": 1.887895507570662e-05, "loss": 0.1759, "step": 1678 }, { "epoch": 0.89, "learning_rate": 1.8877376915225646e-05, "loss": 0.1525, "step": 1679 }, { "epoch": 0.89, "learning_rate": 1.8875797710749152e-05, "loss": 0.1729, "step": 1680 }, { "epoch": 0.89, "learning_rate": 1.887421746246286e-05, "loss": 0.3038, "step": 1681 }, { "epoch": 0.89, "learning_rate": 1.8872636170552606e-05, "loss": 0.2211, "step": 1682 }, { "epoch": 0.89, "learning_rate": 1.887105383520435e-05, "loss": 0.2747, "step": 1683 }, { "epoch": 0.89, "learning_rate": 1.886947045660419e-05, "loss": 0.3284, "step": 1684 }, { "epoch": 0.89, "learning_rate": 1.8867886034938318e-05, "loss": 0.2778, "step": 1685 }, { "epoch": 0.89, "learning_rate": 1.8866300570393074e-05, "loss": 0.2867, "step": 1686 }, { "epoch": 0.89, "learning_rate": 1.886471406315491e-05, "loss": 0.1729, "step": 1687 }, { "epoch": 0.89, "learning_rate": 1.88631265134104e-05, "loss": 0.4128, "step": 1688 }, { "epoch": 0.89, "learning_rate": 1.8861537921346248e-05, "loss": 0.2411, "step": 1689 }, { "epoch": 0.89, "learning_rate": 1.885994828714927e-05, "loss": 0.1094, "step": 1690 }, { "epoch": 0.9, "learning_rate": 1.885835761100641e-05, "loss": 0.2582, "step": 1691 }, { "epoch": 0.9, "learning_rate": 1.8856765893104736e-05, "loss": 0.3346, "step": 1692 }, { "epoch": 0.9, "learning_rate": 1.885517313363144e-05, "loss": 0.2822, "step": 1693 }, { "epoch": 0.9, "learning_rate": 1.8853579332773825e-05, "loss": 0.2713, "step": 1694 }, { "epoch": 0.9, "learning_rate": 1.8851984490719334e-05, "loss": 0.2561, "step": 1695 }, { "epoch": 0.9, "learning_rate": 1.885038860765552e-05, "loss": 0.2509, "step": 1696 }, { "epoch": 0.9, "learning_rate": 1.884879168377006e-05, "loss": 0.2507, "step": 1697 }, { "epoch": 0.9, "learning_rate": 1.884719371925076e-05, "loss": 0.215, "step": 1698 }, { "epoch": 0.9, "learning_rate": 1.8845594714285535e-05, "loss": 0.3245, "step": 1699 }, { "epoch": 0.9, "learning_rate": 1.8843994669062436e-05, "loss": 0.226, "step": 1700 }, { "epoch": 0.9, "learning_rate": 1.884239358376963e-05, "loss": 0.1931, "step": 1701 }, { "epoch": 0.9, "learning_rate": 1.8840791458595414e-05, "loss": 0.0939, "step": 1702 }, { "epoch": 0.9, "learning_rate": 1.883918829372819e-05, "loss": 0.2322, "step": 1703 }, { "epoch": 0.9, "learning_rate": 1.8837584089356498e-05, "loss": 0.1199, "step": 1704 }, { "epoch": 0.9, "learning_rate": 1.8835978845668998e-05, "loss": 0.3347, "step": 1705 }, { "epoch": 0.9, "learning_rate": 1.8834372562854465e-05, "loss": 0.1551, "step": 1706 }, { "epoch": 0.9, "learning_rate": 1.8832765241101807e-05, "loss": 0.1799, "step": 1707 }, { "epoch": 0.9, "learning_rate": 1.883115688060004e-05, "loss": 0.2437, "step": 1708 }, { "epoch": 0.9, "learning_rate": 1.8829547481538314e-05, "loss": 0.1061, "step": 1709 }, { "epoch": 0.91, "learning_rate": 1.88279370441059e-05, "loss": 0.1334, "step": 1710 }, { "epoch": 0.91, "learning_rate": 1.882632556849218e-05, "loss": 0.2521, "step": 1711 }, { "epoch": 0.91, "learning_rate": 1.8824713054886676e-05, "loss": 0.3507, "step": 1712 }, { "epoch": 0.91, "learning_rate": 1.882309950347902e-05, "loss": 0.1849, "step": 1713 }, { "epoch": 0.91, "learning_rate": 1.8821484914458964e-05, "loss": 0.1491, "step": 1714 }, { "epoch": 0.91, "learning_rate": 1.8819869288016387e-05, "loss": 0.2472, "step": 1715 }, { "epoch": 0.91, "learning_rate": 1.88182526243413e-05, "loss": 0.2569, "step": 1716 }, { "epoch": 0.91, "learning_rate": 1.881663492362381e-05, "loss": 0.1564, "step": 1717 }, { "epoch": 0.91, "learning_rate": 1.8815016186054174e-05, "loss": 0.2951, "step": 1718 }, { "epoch": 0.91, "learning_rate": 1.881339641182275e-05, "loss": 0.1771, "step": 1719 }, { "epoch": 0.91, "learning_rate": 1.881177560112003e-05, "loss": 0.1853, "step": 1720 }, { "epoch": 0.91, "learning_rate": 1.881015375413663e-05, "loss": 0.301, "step": 1721 }, { "epoch": 0.91, "learning_rate": 1.8808530871063272e-05, "loss": 0.2594, "step": 1722 }, { "epoch": 0.91, "learning_rate": 1.8806906952090814e-05, "loss": 0.1709, "step": 1723 }, { "epoch": 0.91, "learning_rate": 1.8805281997410232e-05, "loss": 0.1418, "step": 1724 }, { "epoch": 0.91, "learning_rate": 1.880365600721263e-05, "loss": 0.2315, "step": 1725 }, { "epoch": 0.91, "learning_rate": 1.8802028981689214e-05, "loss": 0.1543, "step": 1726 }, { "epoch": 0.91, "learning_rate": 1.8800400921031335e-05, "loss": 0.1459, "step": 1727 }, { "epoch": 0.91, "learning_rate": 1.879877182543045e-05, "loss": 0.19, "step": 1728 }, { "epoch": 0.92, "learning_rate": 1.879714169507815e-05, "loss": 0.2072, "step": 1729 }, { "epoch": 0.92, "learning_rate": 1.8795510530166135e-05, "loss": 0.3728, "step": 1730 }, { "epoch": 0.92, "learning_rate": 1.879387833088624e-05, "loss": 0.1944, "step": 1731 }, { "epoch": 0.92, "learning_rate": 1.879224509743041e-05, "loss": 0.3615, "step": 1732 }, { "epoch": 0.92, "learning_rate": 1.8790610829990715e-05, "loss": 0.1924, "step": 1733 }, { "epoch": 0.92, "learning_rate": 1.878897552875935e-05, "loss": 0.2361, "step": 1734 }, { "epoch": 0.92, "learning_rate": 1.878733919392863e-05, "loss": 0.2021, "step": 1735 }, { "epoch": 0.92, "learning_rate": 1.8785701825690988e-05, "loss": 0.2072, "step": 1736 }, { "epoch": 0.92, "learning_rate": 1.8784063424238983e-05, "loss": 0.322, "step": 1737 }, { "epoch": 0.92, "learning_rate": 1.8782423989765295e-05, "loss": 0.1845, "step": 1738 }, { "epoch": 0.92, "learning_rate": 1.8780783522462724e-05, "loss": 0.3244, "step": 1739 }, { "epoch": 0.92, "learning_rate": 1.877914202252419e-05, "loss": 0.2256, "step": 1740 }, { "epoch": 0.92, "learning_rate": 1.8777499490142742e-05, "loss": 0.162, "step": 1741 }, { "epoch": 0.92, "learning_rate": 1.8775855925511537e-05, "loss": 0.26, "step": 1742 }, { "epoch": 0.92, "learning_rate": 1.8774211328823866e-05, "loss": 0.2309, "step": 1743 }, { "epoch": 0.92, "learning_rate": 1.8772565700273137e-05, "loss": 0.211, "step": 1744 }, { "epoch": 0.92, "learning_rate": 1.8770919040052873e-05, "loss": 0.1671, "step": 1745 }, { "epoch": 0.92, "learning_rate": 1.8769271348356734e-05, "loss": 0.2785, "step": 1746 }, { "epoch": 0.92, "learning_rate": 1.876762262537848e-05, "loss": 0.2299, "step": 1747 }, { "epoch": 0.93, "learning_rate": 1.876597287131201e-05, "loss": 0.2601, "step": 1748 }, { "epoch": 0.93, "learning_rate": 1.8764322086351342e-05, "loss": 0.2133, "step": 1749 }, { "epoch": 0.93, "learning_rate": 1.8762670270690605e-05, "loss": 0.1704, "step": 1750 }, { "epoch": 0.93, "learning_rate": 1.876101742452406e-05, "loss": 0.2721, "step": 1751 }, { "epoch": 0.93, "learning_rate": 1.875936354804608e-05, "loss": 0.1719, "step": 1752 }, { "epoch": 0.93, "learning_rate": 1.8757708641451166e-05, "loss": 0.1677, "step": 1753 }, { "epoch": 0.93, "learning_rate": 1.8756052704933938e-05, "loss": 0.2157, "step": 1754 }, { "epoch": 0.93, "learning_rate": 1.875439573868914e-05, "loss": 0.1783, "step": 1755 }, { "epoch": 0.93, "learning_rate": 1.8752737742911628e-05, "loss": 0.244, "step": 1756 }, { "epoch": 0.93, "learning_rate": 1.875107871779639e-05, "loss": 0.2698, "step": 1757 }, { "epoch": 0.93, "learning_rate": 1.874941866353853e-05, "loss": 0.3474, "step": 1758 }, { "epoch": 0.93, "learning_rate": 1.8747757580333273e-05, "loss": 0.2528, "step": 1759 }, { "epoch": 0.93, "learning_rate": 1.874609546837596e-05, "loss": 0.1518, "step": 1760 }, { "epoch": 0.93, "learning_rate": 1.874443232786207e-05, "loss": 0.1469, "step": 1761 }, { "epoch": 0.93, "learning_rate": 1.874276815898718e-05, "loss": 0.2212, "step": 1762 }, { "epoch": 0.93, "learning_rate": 1.8741102961947005e-05, "loss": 0.276, "step": 1763 }, { "epoch": 0.93, "learning_rate": 1.8739436736937374e-05, "loss": 0.3411, "step": 1764 }, { "epoch": 0.93, "learning_rate": 1.873776948415424e-05, "loss": 0.3201, "step": 1765 }, { "epoch": 0.93, "learning_rate": 1.8736101203793672e-05, "loss": 0.3237, "step": 1766 }, { "epoch": 0.94, "learning_rate": 1.8734431896051865e-05, "loss": 0.2722, "step": 1767 }, { "epoch": 0.94, "learning_rate": 1.8732761561125126e-05, "loss": 0.2104, "step": 1768 }, { "epoch": 0.94, "learning_rate": 1.8731090199209898e-05, "loss": 0.1891, "step": 1769 }, { "epoch": 0.94, "learning_rate": 1.8729417810502734e-05, "loss": 0.2903, "step": 1770 }, { "epoch": 0.94, "learning_rate": 1.872774439520031e-05, "loss": 0.2212, "step": 1771 }, { "epoch": 0.94, "learning_rate": 1.8726069953499417e-05, "loss": 0.313, "step": 1772 }, { "epoch": 0.94, "learning_rate": 1.872439448559698e-05, "loss": 0.1781, "step": 1773 }, { "epoch": 0.94, "learning_rate": 1.8722717991690033e-05, "loss": 0.1735, "step": 1774 }, { "epoch": 0.94, "learning_rate": 1.8721040471975732e-05, "loss": 0.1983, "step": 1775 }, { "epoch": 0.94, "learning_rate": 1.8719361926651367e-05, "loss": 0.2261, "step": 1776 }, { "epoch": 0.94, "learning_rate": 1.8717682355914326e-05, "loss": 0.1883, "step": 1777 }, { "epoch": 0.94, "learning_rate": 1.8716001759962136e-05, "loss": 0.1723, "step": 1778 }, { "epoch": 0.94, "learning_rate": 1.8714320138992438e-05, "loss": 0.3138, "step": 1779 }, { "epoch": 0.94, "learning_rate": 1.871263749320299e-05, "loss": 0.1958, "step": 1780 }, { "epoch": 0.94, "learning_rate": 1.8710953822791676e-05, "loss": 0.1991, "step": 1781 }, { "epoch": 0.94, "learning_rate": 1.8709269127956503e-05, "loss": 0.2423, "step": 1782 }, { "epoch": 0.94, "learning_rate": 1.8707583408895584e-05, "loss": 0.2367, "step": 1783 }, { "epoch": 0.94, "learning_rate": 1.8705896665807174e-05, "loss": 0.2328, "step": 1784 }, { "epoch": 0.94, "learning_rate": 1.8704208898889627e-05, "loss": 0.152, "step": 1785 }, { "epoch": 0.95, "learning_rate": 1.8702520108341437e-05, "loss": 0.1998, "step": 1786 }, { "epoch": 0.95, "learning_rate": 1.87008302943612e-05, "loss": 0.2625, "step": 1787 }, { "epoch": 0.95, "learning_rate": 1.8699139457147648e-05, "loss": 0.298, "step": 1788 }, { "epoch": 0.95, "learning_rate": 1.8697447596899625e-05, "loss": 0.1762, "step": 1789 }, { "epoch": 0.95, "learning_rate": 1.8695754713816094e-05, "loss": 0.1775, "step": 1790 }, { "epoch": 0.95, "learning_rate": 1.8694060808096142e-05, "loss": 0.1747, "step": 1791 }, { "epoch": 0.95, "learning_rate": 1.8692365879938976e-05, "loss": 0.1772, "step": 1792 }, { "epoch": 0.95, "learning_rate": 1.8690669929543923e-05, "loss": 0.2504, "step": 1793 }, { "epoch": 0.95, "learning_rate": 1.868897295711043e-05, "loss": 0.2975, "step": 1794 }, { "epoch": 0.95, "learning_rate": 1.8687274962838064e-05, "loss": 0.2965, "step": 1795 }, { "epoch": 0.95, "learning_rate": 1.868557594692651e-05, "loss": 0.1508, "step": 1796 }, { "epoch": 0.95, "learning_rate": 1.8683875909575574e-05, "loss": 0.1433, "step": 1797 }, { "epoch": 0.95, "learning_rate": 1.8682174850985194e-05, "loss": 0.1954, "step": 1798 }, { "epoch": 0.95, "learning_rate": 1.8680472771355404e-05, "loss": 0.2653, "step": 1799 }, { "epoch": 0.95, "learning_rate": 1.867876967088638e-05, "loss": 0.2274, "step": 1800 }, { "epoch": 0.95, "learning_rate": 1.867706554977841e-05, "loss": 0.3312, "step": 1801 }, { "epoch": 0.95, "learning_rate": 1.8675360408231894e-05, "loss": 0.2156, "step": 1802 }, { "epoch": 0.95, "learning_rate": 1.8673654246447365e-05, "loss": 0.2173, "step": 1803 }, { "epoch": 0.95, "learning_rate": 1.867194706462547e-05, "loss": 0.2768, "step": 1804 }, { "epoch": 0.96, "learning_rate": 1.8670238862966982e-05, "loss": 0.2373, "step": 1805 }, { "epoch": 0.96, "learning_rate": 1.866852964167278e-05, "loss": 0.1864, "step": 1806 }, { "epoch": 0.96, "learning_rate": 1.8666819400943877e-05, "loss": 0.1394, "step": 1807 }, { "epoch": 0.96, "learning_rate": 1.8665108140981397e-05, "loss": 0.2081, "step": 1808 }, { "epoch": 0.96, "learning_rate": 1.866339586198659e-05, "loss": 0.2135, "step": 1809 }, { "epoch": 0.96, "learning_rate": 1.8661682564160818e-05, "loss": 0.2899, "step": 1810 }, { "epoch": 0.96, "learning_rate": 1.8659968247705577e-05, "loss": 0.2852, "step": 1811 }, { "epoch": 0.96, "learning_rate": 1.8658252912822466e-05, "loss": 0.3317, "step": 1812 }, { "epoch": 0.96, "learning_rate": 1.8656536559713214e-05, "loss": 0.2043, "step": 1813 }, { "epoch": 0.96, "learning_rate": 1.865481918857967e-05, "loss": 0.1514, "step": 1814 }, { "epoch": 0.96, "learning_rate": 1.8653100799623794e-05, "loss": 0.1011, "step": 1815 }, { "epoch": 0.96, "learning_rate": 1.865138139304768e-05, "loss": 0.1644, "step": 1816 }, { "epoch": 0.96, "learning_rate": 1.864966096905352e-05, "loss": 0.194, "step": 1817 }, { "epoch": 0.96, "learning_rate": 1.864793952784365e-05, "loss": 0.3367, "step": 1818 }, { "epoch": 0.96, "learning_rate": 1.8646217069620512e-05, "loss": 0.0921, "step": 1819 }, { "epoch": 0.96, "learning_rate": 1.864449359458667e-05, "loss": 0.1428, "step": 1820 }, { "epoch": 0.96, "learning_rate": 1.8642769102944812e-05, "loss": 0.2963, "step": 1821 }, { "epoch": 0.96, "learning_rate": 1.864104359489773e-05, "loss": 0.163, "step": 1822 }, { "epoch": 0.96, "learning_rate": 1.8639317070648358e-05, "loss": 0.2945, "step": 1823 }, { "epoch": 0.97, "learning_rate": 1.8637589530399735e-05, "loss": 0.1998, "step": 1824 }, { "epoch": 0.97, "learning_rate": 1.863586097435502e-05, "loss": 0.2589, "step": 1825 }, { "epoch": 0.97, "learning_rate": 1.8634131402717495e-05, "loss": 0.2396, "step": 1826 }, { "epoch": 0.97, "learning_rate": 1.8632400815690565e-05, "loss": 0.2426, "step": 1827 }, { "epoch": 0.97, "learning_rate": 1.8630669213477747e-05, "loss": 0.2813, "step": 1828 }, { "epoch": 0.97, "learning_rate": 1.862893659628268e-05, "loss": 0.285, "step": 1829 }, { "epoch": 0.97, "learning_rate": 1.862720296430912e-05, "loss": 0.1173, "step": 1830 }, { "epoch": 0.97, "learning_rate": 1.8625468317760953e-05, "loss": 0.2383, "step": 1831 }, { "epoch": 0.97, "learning_rate": 1.8623732656842172e-05, "loss": 0.1785, "step": 1832 }, { "epoch": 0.97, "learning_rate": 1.862199598175689e-05, "loss": 0.2496, "step": 1833 }, { "epoch": 0.97, "learning_rate": 1.8620258292709352e-05, "loss": 0.2855, "step": 1834 }, { "epoch": 0.97, "learning_rate": 1.861851958990391e-05, "loss": 0.1903, "step": 1835 }, { "epoch": 0.97, "learning_rate": 1.8616779873545035e-05, "loss": 0.17, "step": 1836 }, { "epoch": 0.97, "learning_rate": 1.8615039143837325e-05, "loss": 0.2519, "step": 1837 }, { "epoch": 0.97, "learning_rate": 1.8613297400985487e-05, "loss": 0.3833, "step": 1838 }, { "epoch": 0.97, "learning_rate": 1.861155464519436e-05, "loss": 0.2679, "step": 1839 }, { "epoch": 0.97, "learning_rate": 1.8609810876668892e-05, "loss": 0.3379, "step": 1840 }, { "epoch": 0.97, "learning_rate": 1.8608066095614156e-05, "loss": 0.1552, "step": 1841 }, { "epoch": 0.97, "learning_rate": 1.8606320302235335e-05, "loss": 0.2583, "step": 1842 }, { "epoch": 0.98, "learning_rate": 1.8604573496737743e-05, "loss": 0.1433, "step": 1843 }, { "epoch": 0.98, "learning_rate": 1.860282567932681e-05, "loss": 0.3246, "step": 1844 }, { "epoch": 0.98, "learning_rate": 1.8601076850208072e-05, "loss": 0.2452, "step": 1845 }, { "epoch": 0.98, "learning_rate": 1.8599327009587204e-05, "loss": 0.2605, "step": 1846 }, { "epoch": 0.98, "learning_rate": 1.8597576157669986e-05, "loss": 0.3274, "step": 1847 }, { "epoch": 0.98, "learning_rate": 1.8595824294662327e-05, "loss": 0.2001, "step": 1848 }, { "epoch": 0.98, "learning_rate": 1.859407142077024e-05, "loss": 0.3322, "step": 1849 }, { "epoch": 0.98, "learning_rate": 1.8592317536199877e-05, "loss": 0.1803, "step": 1850 }, { "epoch": 0.98, "learning_rate": 1.8590562641157487e-05, "loss": 0.1471, "step": 1851 }, { "epoch": 0.98, "learning_rate": 1.8588806735849455e-05, "loss": 0.1199, "step": 1852 }, { "epoch": 0.98, "learning_rate": 1.8587049820482277e-05, "loss": 0.1629, "step": 1853 }, { "epoch": 0.98, "learning_rate": 1.858529189526257e-05, "loss": 0.2188, "step": 1854 }, { "epoch": 0.98, "learning_rate": 1.8583532960397072e-05, "loss": 0.1808, "step": 1855 }, { "epoch": 0.98, "learning_rate": 1.858177301609263e-05, "loss": 0.2806, "step": 1856 }, { "epoch": 0.98, "learning_rate": 1.8580012062556224e-05, "loss": 0.2824, "step": 1857 }, { "epoch": 0.98, "learning_rate": 1.8578250099994938e-05, "loss": 0.2944, "step": 1858 }, { "epoch": 0.98, "learning_rate": 1.857648712861599e-05, "loss": 0.1727, "step": 1859 }, { "epoch": 0.98, "learning_rate": 1.8574723148626703e-05, "loss": 0.3105, "step": 1860 }, { "epoch": 0.99, "learning_rate": 1.8572958160234525e-05, "loss": 0.2051, "step": 1861 }, { "epoch": 0.99, "learning_rate": 1.8571192163647024e-05, "loss": 0.2912, "step": 1862 }, { "epoch": 0.99, "learning_rate": 1.8569425159071883e-05, "loss": 0.1787, "step": 1863 }, { "epoch": 0.99, "learning_rate": 1.8567657146716905e-05, "loss": 0.1755, "step": 1864 }, { "epoch": 0.99, "learning_rate": 1.856588812679001e-05, "loss": 0.2148, "step": 1865 }, { "epoch": 0.99, "learning_rate": 1.856411809949924e-05, "loss": 0.2812, "step": 1866 }, { "epoch": 0.99, "learning_rate": 1.856234706505275e-05, "loss": 0.1967, "step": 1867 }, { "epoch": 0.99, "learning_rate": 1.856057502365882e-05, "loss": 0.1558, "step": 1868 }, { "epoch": 0.99, "learning_rate": 1.8558801975525852e-05, "loss": 0.2483, "step": 1869 }, { "epoch": 0.99, "learning_rate": 1.8557027920862344e-05, "loss": 0.2592, "step": 1870 }, { "epoch": 0.99, "learning_rate": 1.8555252859876942e-05, "loss": 0.2629, "step": 1871 }, { "epoch": 0.99, "learning_rate": 1.8553476792778385e-05, "loss": 0.2562, "step": 1872 }, { "epoch": 0.99, "learning_rate": 1.8551699719775552e-05, "loss": 0.1992, "step": 1873 }, { "epoch": 0.99, "learning_rate": 1.8549921641077426e-05, "loss": 0.2186, "step": 1874 }, { "epoch": 0.99, "learning_rate": 1.8548142556893108e-05, "loss": 0.2461, "step": 1875 }, { "epoch": 0.99, "learning_rate": 1.854636246743183e-05, "loss": 0.136, "step": 1876 }, { "epoch": 0.99, "learning_rate": 1.8544581372902924e-05, "loss": 0.1922, "step": 1877 }, { "epoch": 0.99, "learning_rate": 1.8542799273515858e-05, "loss": 0.192, "step": 1878 }, { "epoch": 0.99, "learning_rate": 1.8541016169480205e-05, "loss": 0.1706, "step": 1879 }, { "epoch": 1.0, "learning_rate": 1.8539232061005665e-05, "loss": 0.1609, "step": 1880 }, { "epoch": 1.0, "learning_rate": 1.853744694830205e-05, "loss": 0.1716, "step": 1881 }, { "epoch": 1.0, "learning_rate": 1.8535660831579292e-05, "loss": 0.216, "step": 1882 }, { "epoch": 1.0, "learning_rate": 1.8533873711047444e-05, "loss": 0.3389, "step": 1883 }, { "epoch": 1.0, "learning_rate": 1.8532085586916672e-05, "loss": 0.204, "step": 1884 }, { "epoch": 1.0, "learning_rate": 1.8530296459397267e-05, "loss": 0.4067, "step": 1885 }, { "epoch": 1.0, "learning_rate": 1.8528506328699626e-05, "loss": 0.2446, "step": 1886 }, { "epoch": 1.0, "learning_rate": 1.852671519503428e-05, "loss": 0.3089, "step": 1887 }, { "epoch": 1.0, "learning_rate": 1.852492305861186e-05, "loss": 0.2159, "step": 1888 }, { "epoch": 1.0, "learning_rate": 1.8523129919643134e-05, "loss": 0.2777, "step": 1889 }, { "epoch": 1.0, "learning_rate": 1.8521335778338976e-05, "loss": 0.0902, "step": 1890 }, { "epoch": 1.0, "learning_rate": 1.8519540634910375e-05, "loss": 0.194, "step": 1891 }, { "epoch": 1.0, "learning_rate": 1.851774448956845e-05, "loss": 0.112, "step": 1892 }, { "epoch": 1.0, "learning_rate": 1.8515947342524425e-05, "loss": 0.193, "step": 1893 }, { "epoch": 1.0, "learning_rate": 1.8514149193989654e-05, "loss": 0.2418, "step": 1894 }, { "epoch": 1.0, "learning_rate": 1.8512350044175596e-05, "loss": 0.1016, "step": 1895 }, { "epoch": 1.0, "learning_rate": 1.8510549893293837e-05, "loss": 0.1744, "step": 1896 }, { "epoch": 1.0, "learning_rate": 1.850874874155608e-05, "loss": 0.1653, "step": 1897 }, { "epoch": 1.0, "learning_rate": 1.8506946589174142e-05, "loss": 0.1297, "step": 1898 }, { "epoch": 1.01, "learning_rate": 1.8505143436359957e-05, "loss": 0.1555, "step": 1899 }, { "epoch": 1.01, "learning_rate": 1.8503339283325585e-05, "loss": 0.1329, "step": 1900 }, { "epoch": 1.01, "learning_rate": 1.850153413028319e-05, "loss": 0.1371, "step": 1901 }, { "epoch": 1.01, "learning_rate": 1.8499727977445066e-05, "loss": 0.1862, "step": 1902 }, { "epoch": 1.01, "learning_rate": 1.8497920825023622e-05, "loss": 0.2633, "step": 1903 }, { "epoch": 1.01, "learning_rate": 1.8496112673231375e-05, "loss": 0.3586, "step": 1904 }, { "epoch": 1.01, "learning_rate": 1.8494303522280974e-05, "loss": 0.2513, "step": 1905 }, { "epoch": 1.01, "learning_rate": 1.8492493372385178e-05, "loss": 0.3168, "step": 1906 }, { "epoch": 1.01, "learning_rate": 1.8490682223756855e-05, "loss": 0.0725, "step": 1907 }, { "epoch": 1.01, "learning_rate": 1.848887007660901e-05, "loss": 0.139, "step": 1908 }, { "epoch": 1.01, "learning_rate": 1.848705693115475e-05, "loss": 0.2051, "step": 1909 }, { "epoch": 1.01, "learning_rate": 1.8485242787607306e-05, "loss": 0.2172, "step": 1910 }, { "epoch": 1.01, "learning_rate": 1.8483427646180023e-05, "loss": 0.1716, "step": 1911 }, { "epoch": 1.01, "learning_rate": 1.8481611507086367e-05, "loss": 0.237, "step": 1912 }, { "epoch": 1.01, "learning_rate": 1.8479794370539915e-05, "loss": 0.1711, "step": 1913 }, { "epoch": 1.01, "learning_rate": 1.8477976236754368e-05, "loss": 0.1887, "step": 1914 }, { "epoch": 1.01, "learning_rate": 1.8476157105943544e-05, "loss": 0.1351, "step": 1915 }, { "epoch": 1.01, "learning_rate": 1.847433697832137e-05, "loss": 0.0897, "step": 1916 }, { "epoch": 1.01, "learning_rate": 1.8472515854101902e-05, "loss": 0.203, "step": 1917 }, { "epoch": 1.02, "learning_rate": 1.8470693733499307e-05, "loss": 0.2131, "step": 1918 }, { "epoch": 1.02, "learning_rate": 1.8468870616727867e-05, "loss": 0.1638, "step": 1919 }, { "epoch": 1.02, "learning_rate": 1.8467046504001985e-05, "loss": 0.1451, "step": 1920 }, { "epoch": 1.02, "learning_rate": 1.8465221395536185e-05, "loss": 0.1829, "step": 1921 }, { "epoch": 1.02, "learning_rate": 1.846339529154509e-05, "loss": 0.2458, "step": 1922 }, { "epoch": 1.02, "learning_rate": 1.8461568192243467e-05, "loss": 0.1917, "step": 1923 }, { "epoch": 1.02, "learning_rate": 1.845974009784618e-05, "loss": 0.2585, "step": 1924 }, { "epoch": 1.02, "learning_rate": 1.8457911008568218e-05, "loss": 0.1831, "step": 1925 }, { "epoch": 1.02, "learning_rate": 1.845608092462468e-05, "loss": 0.1131, "step": 1926 }, { "epoch": 1.02, "learning_rate": 1.8454249846230794e-05, "loss": 0.1655, "step": 1927 }, { "epoch": 1.02, "learning_rate": 1.8452417773601895e-05, "loss": 0.1841, "step": 1928 }, { "epoch": 1.02, "learning_rate": 1.8450584706953436e-05, "loss": 0.119, "step": 1929 }, { "epoch": 1.02, "learning_rate": 1.8448750646500992e-05, "loss": 0.1868, "step": 1930 }, { "epoch": 1.02, "learning_rate": 1.8446915592460253e-05, "loss": 0.2091, "step": 1931 }, { "epoch": 1.02, "learning_rate": 1.844507954504702e-05, "loss": 0.1373, "step": 1932 }, { "epoch": 1.02, "learning_rate": 1.844324250447722e-05, "loss": 0.1537, "step": 1933 }, { "epoch": 1.02, "learning_rate": 1.8441404470966887e-05, "loss": 0.1863, "step": 1934 }, { "epoch": 1.02, "learning_rate": 1.8439565444732185e-05, "loss": 0.2371, "step": 1935 }, { "epoch": 1.02, "learning_rate": 1.843772542598938e-05, "loss": 0.1078, "step": 1936 }, { "epoch": 1.03, "learning_rate": 1.8435884414954865e-05, "loss": 0.2168, "step": 1937 }, { "epoch": 1.03, "learning_rate": 1.8434042411845144e-05, "loss": 0.223, "step": 1938 }, { "epoch": 1.03, "learning_rate": 1.843219941687684e-05, "loss": 0.1702, "step": 1939 }, { "epoch": 1.03, "learning_rate": 1.8430355430266692e-05, "loss": 0.1849, "step": 1940 }, { "epoch": 1.03, "learning_rate": 1.842851045223156e-05, "loss": 0.214, "step": 1941 }, { "epoch": 1.03, "learning_rate": 1.8426664482988414e-05, "loss": 0.1065, "step": 1942 }, { "epoch": 1.03, "learning_rate": 1.8424817522754342e-05, "loss": 0.1827, "step": 1943 }, { "epoch": 1.03, "learning_rate": 1.8422969571746553e-05, "loss": 0.1757, "step": 1944 }, { "epoch": 1.03, "learning_rate": 1.842112063018237e-05, "loss": 0.1571, "step": 1945 }, { "epoch": 1.03, "learning_rate": 1.8419270698279226e-05, "loss": 0.2437, "step": 1946 }, { "epoch": 1.03, "learning_rate": 1.841741977625468e-05, "loss": 0.0897, "step": 1947 }, { "epoch": 1.03, "learning_rate": 1.8415567864326406e-05, "loss": 0.1741, "step": 1948 }, { "epoch": 1.03, "learning_rate": 1.841371496271219e-05, "loss": 0.136, "step": 1949 }, { "epoch": 1.03, "learning_rate": 1.8411861071629936e-05, "loss": 0.16, "step": 1950 }, { "epoch": 1.03, "learning_rate": 1.8410006191297666e-05, "loss": 0.2035, "step": 1951 }, { "epoch": 1.03, "learning_rate": 1.8408150321933514e-05, "loss": 0.2677, "step": 1952 }, { "epoch": 1.03, "learning_rate": 1.840629346375574e-05, "loss": 0.2014, "step": 1953 }, { "epoch": 1.03, "learning_rate": 1.8404435616982707e-05, "loss": 0.173, "step": 1954 }, { "epoch": 1.03, "learning_rate": 1.8402576781832905e-05, "loss": 0.1804, "step": 1955 }, { "epoch": 1.04, "learning_rate": 1.8400716958524935e-05, "loss": 0.1551, "step": 1956 }, { "epoch": 1.04, "learning_rate": 1.839885614727752e-05, "loss": 0.1795, "step": 1957 }, { "epoch": 1.04, "learning_rate": 1.8396994348309486e-05, "loss": 0.2249, "step": 1958 }, { "epoch": 1.04, "learning_rate": 1.839513156183979e-05, "loss": 0.0991, "step": 1959 }, { "epoch": 1.04, "learning_rate": 1.83932677880875e-05, "loss": 0.2029, "step": 1960 }, { "epoch": 1.04, "learning_rate": 1.8391403027271794e-05, "loss": 0.1166, "step": 1961 }, { "epoch": 1.04, "learning_rate": 1.8389537279611978e-05, "loss": 0.1378, "step": 1962 }, { "epoch": 1.04, "learning_rate": 1.8387670545327463e-05, "loss": 0.1667, "step": 1963 }, { "epoch": 1.04, "learning_rate": 1.8385802824637777e-05, "loss": 0.1506, "step": 1964 }, { "epoch": 1.04, "learning_rate": 1.8383934117762576e-05, "loss": 0.1844, "step": 1965 }, { "epoch": 1.04, "learning_rate": 1.8382064424921616e-05, "loss": 0.2588, "step": 1966 }, { "epoch": 1.04, "learning_rate": 1.838019374633478e-05, "loss": 0.1866, "step": 1967 }, { "epoch": 1.04, "learning_rate": 1.837832208222206e-05, "loss": 0.1875, "step": 1968 }, { "epoch": 1.04, "learning_rate": 1.8376449432803573e-05, "loss": 0.2236, "step": 1969 }, { "epoch": 1.04, "learning_rate": 1.8374575798299537e-05, "loss": 0.1946, "step": 1970 }, { "epoch": 1.04, "learning_rate": 1.8372701178930303e-05, "loss": 0.1753, "step": 1971 }, { "epoch": 1.04, "learning_rate": 1.8370825574916324e-05, "loss": 0.2689, "step": 1972 }, { "epoch": 1.04, "learning_rate": 1.8368948986478176e-05, "loss": 0.219, "step": 1973 }, { "epoch": 1.04, "learning_rate": 1.8367071413836554e-05, "loss": 0.285, "step": 1974 }, { "epoch": 1.05, "learning_rate": 1.836519285721226e-05, "loss": 0.2227, "step": 1975 }, { "epoch": 1.05, "learning_rate": 1.836331331682621e-05, "loss": 0.1803, "step": 1976 }, { "epoch": 1.05, "learning_rate": 1.8361432792899452e-05, "loss": 0.1308, "step": 1977 }, { "epoch": 1.05, "learning_rate": 1.835955128565313e-05, "loss": 0.1952, "step": 1978 }, { "epoch": 1.05, "learning_rate": 1.835766879530852e-05, "loss": 0.203, "step": 1979 }, { "epoch": 1.05, "learning_rate": 1.8355785322087e-05, "loss": 0.25, "step": 1980 }, { "epoch": 1.05, "learning_rate": 1.8353900866210075e-05, "loss": 0.2088, "step": 1981 }, { "epoch": 1.05, "learning_rate": 1.835201542789936e-05, "loss": 0.1578, "step": 1982 }, { "epoch": 1.05, "learning_rate": 1.8350129007376578e-05, "loss": 0.0832, "step": 1983 }, { "epoch": 1.05, "learning_rate": 1.8348241604863585e-05, "loss": 0.147, "step": 1984 }, { "epoch": 1.05, "learning_rate": 1.8346353220582337e-05, "loss": 0.2001, "step": 1985 }, { "epoch": 1.05, "learning_rate": 1.8344463854754915e-05, "loss": 0.2965, "step": 1986 }, { "epoch": 1.05, "learning_rate": 1.834257350760351e-05, "loss": 0.2228, "step": 1987 }, { "epoch": 1.05, "learning_rate": 1.8340682179350434e-05, "loss": 0.1268, "step": 1988 }, { "epoch": 1.05, "learning_rate": 1.8338789870218107e-05, "loss": 0.1941, "step": 1989 }, { "epoch": 1.05, "learning_rate": 1.8336896580429062e-05, "loss": 0.2626, "step": 1990 }, { "epoch": 1.05, "learning_rate": 1.8335002310205964e-05, "loss": 0.1502, "step": 1991 }, { "epoch": 1.05, "learning_rate": 1.8333107059771576e-05, "loss": 0.1318, "step": 1992 }, { "epoch": 1.05, "learning_rate": 1.8331210829348792e-05, "loss": 0.1502, "step": 1993 }, { "epoch": 1.06, "learning_rate": 1.83293136191606e-05, "loss": 0.2946, "step": 1994 }, { "epoch": 1.06, "learning_rate": 1.832741542943012e-05, "loss": 0.1438, "step": 1995 }, { "epoch": 1.06, "learning_rate": 1.8325516260380584e-05, "loss": 0.0573, "step": 1996 }, { "epoch": 1.06, "learning_rate": 1.8323616112235337e-05, "loss": 0.181, "step": 1997 }, { "epoch": 1.06, "learning_rate": 1.832171498521784e-05, "loss": 0.2141, "step": 1998 }, { "epoch": 1.06, "learning_rate": 1.8319812879551667e-05, "loss": 0.2217, "step": 1999 }, { "epoch": 1.06, "learning_rate": 1.8317909795460514e-05, "loss": 0.2678, "step": 2000 }, { "epoch": 1.06, "learning_rate": 1.831600573316818e-05, "loss": 0.2364, "step": 2001 }, { "epoch": 1.06, "learning_rate": 1.8314100692898594e-05, "loss": 0.2072, "step": 2002 }, { "epoch": 1.06, "learning_rate": 1.8312194674875785e-05, "loss": 0.1116, "step": 2003 }, { "epoch": 1.06, "learning_rate": 1.8310287679323904e-05, "loss": 0.1449, "step": 2004 }, { "epoch": 1.06, "learning_rate": 1.8308379706467225e-05, "loss": 0.1402, "step": 2005 }, { "epoch": 1.06, "learning_rate": 1.8306470756530126e-05, "loss": 0.1407, "step": 2006 }, { "epoch": 1.06, "learning_rate": 1.83045608297371e-05, "loss": 0.1359, "step": 2007 }, { "epoch": 1.06, "learning_rate": 1.8302649926312755e-05, "loss": 0.1723, "step": 2008 }, { "epoch": 1.06, "learning_rate": 1.8300738046481824e-05, "loss": 0.1796, "step": 2009 }, { "epoch": 1.06, "learning_rate": 1.8298825190469145e-05, "loss": 0.2626, "step": 2010 }, { "epoch": 1.06, "learning_rate": 1.8296911358499672e-05, "loss": 0.187, "step": 2011 }, { "epoch": 1.06, "learning_rate": 1.8294996550798477e-05, "loss": 0.1653, "step": 2012 }, { "epoch": 1.07, "learning_rate": 1.829308076759074e-05, "loss": 0.0929, "step": 2013 }, { "epoch": 1.07, "learning_rate": 1.8291164009101766e-05, "loss": 0.1831, "step": 2014 }, { "epoch": 1.07, "learning_rate": 1.8289246275556972e-05, "loss": 0.1509, "step": 2015 }, { "epoch": 1.07, "learning_rate": 1.8287327567181878e-05, "loss": 0.0974, "step": 2016 }, { "epoch": 1.07, "learning_rate": 1.828540788420213e-05, "loss": 0.2032, "step": 2017 }, { "epoch": 1.07, "learning_rate": 1.828348722684349e-05, "loss": 0.1738, "step": 2018 }, { "epoch": 1.07, "learning_rate": 1.828156559533183e-05, "loss": 0.1415, "step": 2019 }, { "epoch": 1.07, "learning_rate": 1.8279642989893136e-05, "loss": 0.1919, "step": 2020 }, { "epoch": 1.07, "learning_rate": 1.8277719410753508e-05, "loss": 0.0944, "step": 2021 }, { "epoch": 1.07, "learning_rate": 1.8275794858139168e-05, "loss": 0.2332, "step": 2022 }, { "epoch": 1.07, "learning_rate": 1.827386933227644e-05, "loss": 0.2643, "step": 2023 }, { "epoch": 1.07, "learning_rate": 1.8271942833391775e-05, "loss": 0.1972, "step": 2024 }, { "epoch": 1.07, "learning_rate": 1.827001536171173e-05, "loss": 0.1653, "step": 2025 }, { "epoch": 1.07, "learning_rate": 1.8268086917462975e-05, "loss": 0.1807, "step": 2026 }, { "epoch": 1.07, "learning_rate": 1.8266157500872307e-05, "loss": 0.1321, "step": 2027 }, { "epoch": 1.07, "learning_rate": 1.8264227112166625e-05, "loss": 0.2042, "step": 2028 }, { "epoch": 1.07, "learning_rate": 1.8262295751572943e-05, "loss": 0.0952, "step": 2029 }, { "epoch": 1.07, "learning_rate": 1.82603634193184e-05, "loss": 0.2264, "step": 2030 }, { "epoch": 1.08, "learning_rate": 1.8258430115630236e-05, "loss": 0.1602, "step": 2031 }, { "epoch": 1.08, "learning_rate": 1.825649584073581e-05, "loss": 0.1703, "step": 2032 }, { "epoch": 1.08, "learning_rate": 1.82545605948626e-05, "loss": 0.2702, "step": 2033 }, { "epoch": 1.08, "learning_rate": 1.8252624378238193e-05, "loss": 0.2033, "step": 2034 }, { "epoch": 1.08, "learning_rate": 1.825068719109029e-05, "loss": 0.1481, "step": 2035 }, { "epoch": 1.08, "learning_rate": 1.824874903364671e-05, "loss": 0.1654, "step": 2036 }, { "epoch": 1.08, "learning_rate": 1.8246809906135387e-05, "loss": 0.1863, "step": 2037 }, { "epoch": 1.08, "learning_rate": 1.8244869808784355e-05, "loss": 0.0803, "step": 2038 }, { "epoch": 1.08, "learning_rate": 1.8242928741821783e-05, "loss": 0.1023, "step": 2039 }, { "epoch": 1.08, "learning_rate": 1.8240986705475943e-05, "loss": 0.1517, "step": 2040 }, { "epoch": 1.08, "learning_rate": 1.8239043699975214e-05, "loss": 0.1767, "step": 2041 }, { "epoch": 1.08, "learning_rate": 1.8237099725548107e-05, "loss": 0.1223, "step": 2042 }, { "epoch": 1.08, "learning_rate": 1.823515478242323e-05, "loss": 0.1413, "step": 2043 }, { "epoch": 1.08, "learning_rate": 1.8233208870829317e-05, "loss": 0.143, "step": 2044 }, { "epoch": 1.08, "learning_rate": 1.823126199099521e-05, "loss": 0.1716, "step": 2045 }, { "epoch": 1.08, "learning_rate": 1.822931414314986e-05, "loss": 0.2502, "step": 2046 }, { "epoch": 1.08, "learning_rate": 1.822736532752234e-05, "loss": 0.0807, "step": 2047 }, { "epoch": 1.08, "learning_rate": 1.822541554434184e-05, "loss": 0.269, "step": 2048 }, { "epoch": 1.08, "learning_rate": 1.822346479383765e-05, "loss": 0.2369, "step": 2049 }, { "epoch": 1.09, "learning_rate": 1.8221513076239183e-05, "loss": 0.2597, "step": 2050 }, { "epoch": 1.09, "learning_rate": 1.821956039177597e-05, "loss": 0.2655, "step": 2051 }, { "epoch": 1.09, "learning_rate": 1.8217606740677647e-05, "loss": 0.176, "step": 2052 }, { "epoch": 1.09, "learning_rate": 1.8215652123173964e-05, "loss": 0.1581, "step": 2053 }, { "epoch": 1.09, "learning_rate": 1.821369653949479e-05, "loss": 0.2742, "step": 2054 }, { "epoch": 1.09, "learning_rate": 1.821173998987011e-05, "loss": 0.1493, "step": 2055 }, { "epoch": 1.09, "learning_rate": 1.8209782474530013e-05, "loss": 0.1479, "step": 2056 }, { "epoch": 1.09, "learning_rate": 1.82078239937047e-05, "loss": 0.1768, "step": 2057 }, { "epoch": 1.09, "learning_rate": 1.8205864547624503e-05, "loss": 0.1859, "step": 2058 }, { "epoch": 1.09, "learning_rate": 1.820390413651985e-05, "loss": 0.221, "step": 2059 }, { "epoch": 1.09, "learning_rate": 1.8201942760621293e-05, "loss": 0.1693, "step": 2060 }, { "epoch": 1.09, "learning_rate": 1.819998042015949e-05, "loss": 0.176, "step": 2061 }, { "epoch": 1.09, "learning_rate": 1.8198017115365217e-05, "loss": 0.1607, "step": 2062 }, { "epoch": 1.09, "learning_rate": 1.819605284646936e-05, "loss": 0.1197, "step": 2063 }, { "epoch": 1.09, "learning_rate": 1.8194087613702923e-05, "loss": 0.1551, "step": 2064 }, { "epoch": 1.09, "learning_rate": 1.819212141729702e-05, "loss": 0.2326, "step": 2065 }, { "epoch": 1.09, "learning_rate": 1.8190154257482886e-05, "loss": 0.2267, "step": 2066 }, { "epoch": 1.09, "learning_rate": 1.818818613449185e-05, "loss": 0.1671, "step": 2067 }, { "epoch": 1.09, "learning_rate": 1.8186217048555376e-05, "loss": 0.1268, "step": 2068 }, { "epoch": 1.1, "learning_rate": 1.8184246999905024e-05, "loss": 0.2321, "step": 2069 }, { "epoch": 1.1, "learning_rate": 1.8182275988772485e-05, "loss": 0.2777, "step": 2070 }, { "epoch": 1.1, "learning_rate": 1.8180304015389545e-05, "loss": 0.1855, "step": 2071 }, { "epoch": 1.1, "learning_rate": 1.817833107998812e-05, "loss": 0.1632, "step": 2072 }, { "epoch": 1.1, "learning_rate": 1.8176357182800223e-05, "loss": 0.1946, "step": 2073 }, { "epoch": 1.1, "learning_rate": 1.8174382324057994e-05, "loss": 0.1594, "step": 2074 }, { "epoch": 1.1, "learning_rate": 1.8172406503993672e-05, "loss": 0.175, "step": 2075 }, { "epoch": 1.1, "learning_rate": 1.8170429722839625e-05, "loss": 0.1399, "step": 2076 }, { "epoch": 1.1, "learning_rate": 1.8168451980828323e-05, "loss": 0.1075, "step": 2077 }, { "epoch": 1.1, "learning_rate": 1.8166473278192355e-05, "loss": 0.1964, "step": 2078 }, { "epoch": 1.1, "learning_rate": 1.8164493615164413e-05, "loss": 0.2056, "step": 2079 }, { "epoch": 1.1, "learning_rate": 1.8162512991977314e-05, "loss": 0.172, "step": 2080 }, { "epoch": 1.1, "learning_rate": 1.816053140886398e-05, "loss": 0.2924, "step": 2081 }, { "epoch": 1.1, "learning_rate": 1.8158548866057453e-05, "loss": 0.2017, "step": 2082 }, { "epoch": 1.1, "learning_rate": 1.8156565363790882e-05, "loss": 0.2173, "step": 2083 }, { "epoch": 1.1, "learning_rate": 1.8154580902297526e-05, "loss": 0.149, "step": 2084 }, { "epoch": 1.1, "learning_rate": 1.8152595481810768e-05, "loss": 0.127, "step": 2085 }, { "epoch": 1.1, "learning_rate": 1.8150609102564096e-05, "loss": 0.1, "step": 2086 }, { "epoch": 1.1, "learning_rate": 1.8148621764791105e-05, "loss": 0.1971, "step": 2087 }, { "epoch": 1.11, "learning_rate": 1.8146633468725512e-05, "loss": 0.149, "step": 2088 }, { "epoch": 1.11, "learning_rate": 1.814464421460115e-05, "loss": 0.1086, "step": 2089 }, { "epoch": 1.11, "learning_rate": 1.8142654002651958e-05, "loss": 0.0939, "step": 2090 }, { "epoch": 1.11, "learning_rate": 1.814066283311198e-05, "loss": 0.1701, "step": 2091 }, { "epoch": 1.11, "learning_rate": 1.813867070621539e-05, "loss": 0.1988, "step": 2092 }, { "epoch": 1.11, "learning_rate": 1.8136677622196463e-05, "loss": 0.206, "step": 2093 }, { "epoch": 1.11, "learning_rate": 1.813468358128959e-05, "loss": 0.2649, "step": 2094 }, { "epoch": 1.11, "learning_rate": 1.813268858372927e-05, "loss": 0.16, "step": 2095 }, { "epoch": 1.11, "learning_rate": 1.813069262975012e-05, "loss": 0.1302, "step": 2096 }, { "epoch": 1.11, "learning_rate": 1.812869571958687e-05, "loss": 0.225, "step": 2097 }, { "epoch": 1.11, "learning_rate": 1.812669785347436e-05, "loss": 0.1604, "step": 2098 }, { "epoch": 1.11, "learning_rate": 1.8124699031647543e-05, "loss": 0.1136, "step": 2099 }, { "epoch": 1.11, "learning_rate": 1.812269925434148e-05, "loss": 0.3018, "step": 2100 }, { "epoch": 1.11, "learning_rate": 1.8120698521791352e-05, "loss": 0.257, "step": 2101 }, { "epoch": 1.11, "learning_rate": 1.8118696834232445e-05, "loss": 0.2902, "step": 2102 }, { "epoch": 1.11, "learning_rate": 1.811669419190017e-05, "loss": 0.1653, "step": 2103 }, { "epoch": 1.11, "learning_rate": 1.8114690595030028e-05, "loss": 0.2025, "step": 2104 }, { "epoch": 1.11, "learning_rate": 1.811268604385766e-05, "loss": 0.1768, "step": 2105 }, { "epoch": 1.11, "learning_rate": 1.8110680538618793e-05, "loss": 0.2385, "step": 2106 }, { "epoch": 1.12, "learning_rate": 1.8108674079549283e-05, "loss": 0.1503, "step": 2107 }, { "epoch": 1.12, "learning_rate": 1.8106666666885093e-05, "loss": 0.2407, "step": 2108 }, { "epoch": 1.12, "learning_rate": 1.8104658300862303e-05, "loss": 0.1855, "step": 2109 }, { "epoch": 1.12, "learning_rate": 1.810264898171709e-05, "loss": 0.1306, "step": 2110 }, { "epoch": 1.12, "learning_rate": 1.810063870968576e-05, "loss": 0.1764, "step": 2111 }, { "epoch": 1.12, "learning_rate": 1.8098627485004727e-05, "loss": 0.2033, "step": 2112 }, { "epoch": 1.12, "learning_rate": 1.809661530791051e-05, "loss": 0.1845, "step": 2113 }, { "epoch": 1.12, "learning_rate": 1.8094602178639743e-05, "loss": 0.2223, "step": 2114 }, { "epoch": 1.12, "learning_rate": 1.809258809742918e-05, "loss": 0.2363, "step": 2115 }, { "epoch": 1.12, "learning_rate": 1.8090573064515676e-05, "loss": 0.1788, "step": 2116 }, { "epoch": 1.12, "learning_rate": 1.8088557080136206e-05, "loss": 0.2267, "step": 2117 }, { "epoch": 1.12, "learning_rate": 1.8086540144527853e-05, "loss": 0.2372, "step": 2118 }, { "epoch": 1.12, "learning_rate": 1.8084522257927807e-05, "loss": 0.2119, "step": 2119 }, { "epoch": 1.12, "learning_rate": 1.808250342057338e-05, "loss": 0.1831, "step": 2120 }, { "epoch": 1.12, "learning_rate": 1.8080483632701988e-05, "loss": 0.2038, "step": 2121 }, { "epoch": 1.12, "learning_rate": 1.807846289455116e-05, "loss": 0.1497, "step": 2122 }, { "epoch": 1.12, "learning_rate": 1.8076441206358548e-05, "loss": 0.1681, "step": 2123 }, { "epoch": 1.12, "learning_rate": 1.8074418568361896e-05, "loss": 0.1599, "step": 2124 }, { "epoch": 1.12, "learning_rate": 1.8072394980799077e-05, "loss": 0.1468, "step": 2125 }, { "epoch": 1.13, "learning_rate": 1.8070370443908064e-05, "loss": 0.1773, "step": 2126 }, { "epoch": 1.13, "learning_rate": 1.8068344957926946e-05, "loss": 0.1754, "step": 2127 }, { "epoch": 1.13, "learning_rate": 1.8066318523093927e-05, "loss": 0.0886, "step": 2128 }, { "epoch": 1.13, "learning_rate": 1.8064291139647318e-05, "loss": 0.2006, "step": 2129 }, { "epoch": 1.13, "learning_rate": 1.8062262807825538e-05, "loss": 0.2101, "step": 2130 }, { "epoch": 1.13, "learning_rate": 1.8060233527867135e-05, "loss": 0.2149, "step": 2131 }, { "epoch": 1.13, "learning_rate": 1.8058203300010743e-05, "loss": 0.1726, "step": 2132 }, { "epoch": 1.13, "learning_rate": 1.8056172124495126e-05, "loss": 0.1902, "step": 2133 }, { "epoch": 1.13, "learning_rate": 1.8054140001559154e-05, "loss": 0.1184, "step": 2134 }, { "epoch": 1.13, "learning_rate": 1.8052106931441806e-05, "loss": 0.1582, "step": 2135 }, { "epoch": 1.13, "learning_rate": 1.805007291438218e-05, "loss": 0.2056, "step": 2136 }, { "epoch": 1.13, "learning_rate": 1.804803795061947e-05, "loss": 0.2455, "step": 2137 }, { "epoch": 1.13, "learning_rate": 1.8046002040393004e-05, "loss": 0.1312, "step": 2138 }, { "epoch": 1.13, "learning_rate": 1.8043965183942203e-05, "loss": 0.2, "step": 2139 }, { "epoch": 1.13, "learning_rate": 1.8041927381506602e-05, "loss": 0.1821, "step": 2140 }, { "epoch": 1.13, "learning_rate": 1.803988863332585e-05, "loss": 0.1089, "step": 2141 }, { "epoch": 1.13, "learning_rate": 1.8037848939639717e-05, "loss": 0.1252, "step": 2142 }, { "epoch": 1.13, "learning_rate": 1.8035808300688063e-05, "loss": 0.1521, "step": 2143 }, { "epoch": 1.13, "learning_rate": 1.8033766716710877e-05, "loss": 0.204, "step": 2144 }, { "epoch": 1.14, "learning_rate": 1.803172418794825e-05, "loss": 0.1804, "step": 2145 }, { "epoch": 1.14, "learning_rate": 1.8029680714640393e-05, "loss": 0.2375, "step": 2146 }, { "epoch": 1.14, "learning_rate": 1.802763629702762e-05, "loss": 0.2283, "step": 2147 }, { "epoch": 1.14, "learning_rate": 1.8025590935350347e-05, "loss": 0.1857, "step": 2148 }, { "epoch": 1.14, "learning_rate": 1.8023544629849127e-05, "loss": 0.2756, "step": 2149 }, { "epoch": 1.14, "learning_rate": 1.8021497380764602e-05, "loss": 0.1535, "step": 2150 }, { "epoch": 1.14, "learning_rate": 1.8019449188337535e-05, "loss": 0.166, "step": 2151 }, { "epoch": 1.14, "learning_rate": 1.8017400052808796e-05, "loss": 0.1827, "step": 2152 }, { "epoch": 1.14, "learning_rate": 1.8015349974419365e-05, "loss": 0.1239, "step": 2153 }, { "epoch": 1.14, "learning_rate": 1.8013298953410338e-05, "loss": 0.1773, "step": 2154 }, { "epoch": 1.14, "learning_rate": 1.801124699002292e-05, "loss": 0.1985, "step": 2155 }, { "epoch": 1.14, "learning_rate": 1.8009194084498422e-05, "loss": 0.3806, "step": 2156 }, { "epoch": 1.14, "learning_rate": 1.8007140237078266e-05, "loss": 0.208, "step": 2157 }, { "epoch": 1.14, "learning_rate": 1.8005085448003997e-05, "loss": 0.098, "step": 2158 }, { "epoch": 1.14, "learning_rate": 1.800302971751726e-05, "loss": 0.3039, "step": 2159 }, { "epoch": 1.14, "learning_rate": 1.8000973045859802e-05, "loss": 0.1355, "step": 2160 }, { "epoch": 1.14, "learning_rate": 1.7998915433273506e-05, "loss": 0.155, "step": 2161 }, { "epoch": 1.14, "learning_rate": 1.799685688000034e-05, "loss": 0.1943, "step": 2162 }, { "epoch": 1.14, "learning_rate": 1.7994797386282404e-05, "loss": 0.1288, "step": 2163 }, { "epoch": 1.15, "learning_rate": 1.7992736952361887e-05, "loss": 0.1431, "step": 2164 }, { "epoch": 1.15, "learning_rate": 1.799067557848111e-05, "loss": 0.1904, "step": 2165 }, { "epoch": 1.15, "learning_rate": 1.7988613264882485e-05, "loss": 0.2711, "step": 2166 }, { "epoch": 1.15, "learning_rate": 1.798655001180855e-05, "loss": 0.1352, "step": 2167 }, { "epoch": 1.15, "learning_rate": 1.7984485819501943e-05, "loss": 0.1858, "step": 2168 }, { "epoch": 1.15, "learning_rate": 1.798242068820542e-05, "loss": 0.1703, "step": 2169 }, { "epoch": 1.15, "learning_rate": 1.7980354618161845e-05, "loss": 0.2539, "step": 2170 }, { "epoch": 1.15, "learning_rate": 1.7978287609614188e-05, "loss": 0.1968, "step": 2171 }, { "epoch": 1.15, "learning_rate": 1.7976219662805535e-05, "loss": 0.1622, "step": 2172 }, { "epoch": 1.15, "learning_rate": 1.797415077797908e-05, "loss": 0.1633, "step": 2173 }, { "epoch": 1.15, "learning_rate": 1.7972080955378126e-05, "loss": 0.1359, "step": 2174 }, { "epoch": 1.15, "learning_rate": 1.797001019524609e-05, "loss": 0.1567, "step": 2175 }, { "epoch": 1.15, "learning_rate": 1.79679384978265e-05, "loss": 0.0846, "step": 2176 }, { "epoch": 1.15, "learning_rate": 1.7965865863362985e-05, "loss": 0.1711, "step": 2177 }, { "epoch": 1.15, "learning_rate": 1.7963792292099292e-05, "loss": 0.1405, "step": 2178 }, { "epoch": 1.15, "learning_rate": 1.796171778427928e-05, "loss": 0.1315, "step": 2179 }, { "epoch": 1.15, "learning_rate": 1.7959642340146915e-05, "loss": 0.1603, "step": 2180 }, { "epoch": 1.15, "learning_rate": 1.7957565959946268e-05, "loss": 0.205, "step": 2181 }, { "epoch": 1.15, "learning_rate": 1.795548864392153e-05, "loss": 0.2349, "step": 2182 }, { "epoch": 1.16, "learning_rate": 1.795341039231699e-05, "loss": 0.114, "step": 2183 }, { "epoch": 1.16, "learning_rate": 1.7951331205377063e-05, "loss": 0.1763, "step": 2184 }, { "epoch": 1.16, "learning_rate": 1.7949251083346264e-05, "loss": 0.1943, "step": 2185 }, { "epoch": 1.16, "learning_rate": 1.7947170026469215e-05, "loss": 0.2358, "step": 2186 }, { "epoch": 1.16, "learning_rate": 1.7945088034990652e-05, "loss": 0.3292, "step": 2187 }, { "epoch": 1.16, "learning_rate": 1.7943005109155423e-05, "loss": 0.1316, "step": 2188 }, { "epoch": 1.16, "learning_rate": 1.794092124920848e-05, "loss": 0.1434, "step": 2189 }, { "epoch": 1.16, "learning_rate": 1.7938836455394898e-05, "loss": 0.1778, "step": 2190 }, { "epoch": 1.16, "learning_rate": 1.7936750727959845e-05, "loss": 0.1668, "step": 2191 }, { "epoch": 1.16, "learning_rate": 1.7934664067148602e-05, "loss": 0.1964, "step": 2192 }, { "epoch": 1.16, "learning_rate": 1.7932576473206577e-05, "loss": 0.14, "step": 2193 }, { "epoch": 1.16, "learning_rate": 1.7930487946379265e-05, "loss": 0.154, "step": 2194 }, { "epoch": 1.16, "learning_rate": 1.7928398486912282e-05, "loss": 0.1785, "step": 2195 }, { "epoch": 1.16, "learning_rate": 1.7926308095051354e-05, "loss": 0.1585, "step": 2196 }, { "epoch": 1.16, "learning_rate": 1.7924216771042315e-05, "loss": 0.2215, "step": 2197 }, { "epoch": 1.16, "learning_rate": 1.7922124515131108e-05, "loss": 0.211, "step": 2198 }, { "epoch": 1.16, "learning_rate": 1.7920031327563784e-05, "loss": 0.181, "step": 2199 }, { "epoch": 1.16, "learning_rate": 1.791793720858651e-05, "loss": 0.1592, "step": 2200 }, { "epoch": 1.17, "learning_rate": 1.7915842158445557e-05, "loss": 0.2219, "step": 2201 }, { "epoch": 1.17, "learning_rate": 1.79137461773873e-05, "loss": 0.2127, "step": 2202 }, { "epoch": 1.17, "learning_rate": 1.7911649265658237e-05, "loss": 0.0847, "step": 2203 }, { "epoch": 1.17, "learning_rate": 1.7909551423504966e-05, "loss": 0.2177, "step": 2204 }, { "epoch": 1.17, "learning_rate": 1.79074526511742e-05, "loss": 0.1806, "step": 2205 }, { "epoch": 1.17, "learning_rate": 1.7905352948912757e-05, "loss": 0.18, "step": 2206 }, { "epoch": 1.17, "learning_rate": 1.7903252316967563e-05, "loss": 0.1299, "step": 2207 }, { "epoch": 1.17, "learning_rate": 1.790115075558566e-05, "loss": 0.1434, "step": 2208 }, { "epoch": 1.17, "learning_rate": 1.7899048265014194e-05, "loss": 0.2535, "step": 2209 }, { "epoch": 1.17, "learning_rate": 1.789694484550042e-05, "loss": 0.143, "step": 2210 }, { "epoch": 1.17, "learning_rate": 1.7894840497291705e-05, "loss": 0.1196, "step": 2211 }, { "epoch": 1.17, "learning_rate": 1.7892735220635526e-05, "loss": 0.1514, "step": 2212 }, { "epoch": 1.17, "learning_rate": 1.7890629015779465e-05, "loss": 0.1496, "step": 2213 }, { "epoch": 1.17, "learning_rate": 1.7888521882971218e-05, "loss": 0.1522, "step": 2214 }, { "epoch": 1.17, "learning_rate": 1.7886413822458584e-05, "loss": 0.122, "step": 2215 }, { "epoch": 1.17, "learning_rate": 1.7884304834489475e-05, "loss": 0.2081, "step": 2216 }, { "epoch": 1.17, "learning_rate": 1.788219491931192e-05, "loss": 0.2124, "step": 2217 }, { "epoch": 1.17, "learning_rate": 1.788008407717403e-05, "loss": 0.0788, "step": 2218 }, { "epoch": 1.17, "learning_rate": 1.7877972308324066e-05, "loss": 0.1693, "step": 2219 }, { "epoch": 1.18, "learning_rate": 1.7875859613010362e-05, "loss": 0.2664, "step": 2220 }, { "epoch": 1.18, "learning_rate": 1.787374599148138e-05, "loss": 0.1277, "step": 2221 }, { "epoch": 1.18, "learning_rate": 1.7871631443985684e-05, "loss": 0.1429, "step": 2222 }, { "epoch": 1.18, "learning_rate": 1.7869515970771952e-05, "loss": 0.2901, "step": 2223 }, { "epoch": 1.18, "learning_rate": 1.7867399572088958e-05, "loss": 0.193, "step": 2224 }, { "epoch": 1.18, "learning_rate": 1.7865282248185604e-05, "loss": 0.1406, "step": 2225 }, { "epoch": 1.18, "learning_rate": 1.7863163999310885e-05, "loss": 0.2272, "step": 2226 }, { "epoch": 1.18, "learning_rate": 1.7861044825713922e-05, "loss": 0.2116, "step": 2227 }, { "epoch": 1.18, "learning_rate": 1.785892472764392e-05, "loss": 0.1302, "step": 2228 }, { "epoch": 1.18, "learning_rate": 1.7856803705350207e-05, "loss": 0.3932, "step": 2229 }, { "epoch": 1.18, "learning_rate": 1.7854681759082228e-05, "loss": 0.1902, "step": 2230 }, { "epoch": 1.18, "learning_rate": 1.7852558889089523e-05, "loss": 0.1317, "step": 2231 }, { "epoch": 1.18, "learning_rate": 1.7850435095621746e-05, "loss": 0.1564, "step": 2232 }, { "epoch": 1.18, "learning_rate": 1.784831037892866e-05, "loss": 0.1889, "step": 2233 }, { "epoch": 1.18, "learning_rate": 1.7846184739260132e-05, "loss": 0.2245, "step": 2234 }, { "epoch": 1.18, "learning_rate": 1.7844058176866147e-05, "loss": 0.1232, "step": 2235 }, { "epoch": 1.18, "learning_rate": 1.784193069199678e-05, "loss": 0.2085, "step": 2236 }, { "epoch": 1.18, "learning_rate": 1.7839802284902243e-05, "loss": 0.1843, "step": 2237 }, { "epoch": 1.18, "learning_rate": 1.7837672955832836e-05, "loss": 0.1654, "step": 2238 }, { "epoch": 1.19, "learning_rate": 1.7835542705038963e-05, "loss": 0.2769, "step": 2239 }, { "epoch": 1.19, "learning_rate": 1.783341153277115e-05, "loss": 0.1892, "step": 2240 }, { "epoch": 1.19, "learning_rate": 1.7831279439280034e-05, "loss": 0.1221, "step": 2241 }, { "epoch": 1.19, "learning_rate": 1.7829146424816345e-05, "loss": 0.2193, "step": 2242 }, { "epoch": 1.19, "learning_rate": 1.7827012489630932e-05, "loss": 0.1415, "step": 2243 }, { "epoch": 1.19, "learning_rate": 1.7824877633974752e-05, "loss": 0.1943, "step": 2244 }, { "epoch": 1.19, "learning_rate": 1.7822741858098864e-05, "loss": 0.182, "step": 2245 }, { "epoch": 1.19, "learning_rate": 1.7820605162254437e-05, "loss": 0.1435, "step": 2246 }, { "epoch": 1.19, "learning_rate": 1.7818467546692756e-05, "loss": 0.1544, "step": 2247 }, { "epoch": 1.19, "learning_rate": 1.7816329011665202e-05, "loss": 0.1733, "step": 2248 }, { "epoch": 1.19, "learning_rate": 1.781418955742328e-05, "loss": 0.064, "step": 2249 }, { "epoch": 1.19, "learning_rate": 1.7812049184218586e-05, "loss": 0.2267, "step": 2250 }, { "epoch": 1.19, "learning_rate": 1.7809907892302836e-05, "loss": 0.1728, "step": 2251 }, { "epoch": 1.19, "learning_rate": 1.7807765681927845e-05, "loss": 0.1564, "step": 2252 }, { "epoch": 1.19, "learning_rate": 1.7805622553345545e-05, "loss": 0.2128, "step": 2253 }, { "epoch": 1.19, "learning_rate": 1.780347850680797e-05, "loss": 0.1611, "step": 2254 }, { "epoch": 1.19, "learning_rate": 1.7801333542567265e-05, "loss": 0.2356, "step": 2255 }, { "epoch": 1.19, "learning_rate": 1.779918766087568e-05, "loss": 0.2321, "step": 2256 }, { "epoch": 1.19, "learning_rate": 1.7797040861985574e-05, "loss": 0.1958, "step": 2257 }, { "epoch": 1.2, "learning_rate": 1.7794893146149418e-05, "loss": 0.2322, "step": 2258 }, { "epoch": 1.2, "learning_rate": 1.7792744513619783e-05, "loss": 0.1332, "step": 2259 }, { "epoch": 1.2, "learning_rate": 1.7790594964649353e-05, "loss": 0.2794, "step": 2260 }, { "epoch": 1.2, "learning_rate": 1.7788444499490922e-05, "loss": 0.1293, "step": 2261 }, { "epoch": 1.2, "learning_rate": 1.7786293118397385e-05, "loss": 0.1758, "step": 2262 }, { "epoch": 1.2, "learning_rate": 1.7784140821621753e-05, "loss": 0.1719, "step": 2263 }, { "epoch": 1.2, "learning_rate": 1.7781987609417134e-05, "loss": 0.164, "step": 2264 }, { "epoch": 1.2, "learning_rate": 1.7779833482036752e-05, "loss": 0.1472, "step": 2265 }, { "epoch": 1.2, "learning_rate": 1.777767843973394e-05, "loss": 0.2324, "step": 2266 }, { "epoch": 1.2, "learning_rate": 1.7775522482762125e-05, "loss": 0.1857, "step": 2267 }, { "epoch": 1.2, "learning_rate": 1.7773365611374865e-05, "loss": 0.2232, "step": 2268 }, { "epoch": 1.2, "learning_rate": 1.7771207825825803e-05, "loss": 0.1573, "step": 2269 }, { "epoch": 1.2, "learning_rate": 1.77690491263687e-05, "loss": 0.229, "step": 2270 }, { "epoch": 1.2, "learning_rate": 1.776688951325742e-05, "loss": 0.1635, "step": 2271 }, { "epoch": 1.2, "learning_rate": 1.7764728986745942e-05, "loss": 0.2717, "step": 2272 }, { "epoch": 1.2, "learning_rate": 1.7762567547088346e-05, "loss": 0.2342, "step": 2273 }, { "epoch": 1.2, "learning_rate": 1.7760405194538823e-05, "loss": 0.1172, "step": 2274 }, { "epoch": 1.2, "learning_rate": 1.7758241929351672e-05, "loss": 0.1857, "step": 2275 }, { "epoch": 1.2, "learning_rate": 1.775607775178129e-05, "loss": 0.2706, "step": 2276 }, { "epoch": 1.21, "learning_rate": 1.7753912662082188e-05, "loss": 0.1536, "step": 2277 }, { "epoch": 1.21, "learning_rate": 1.775174666050899e-05, "loss": 0.1695, "step": 2278 }, { "epoch": 1.21, "learning_rate": 1.774957974731642e-05, "loss": 0.1133, "step": 2279 }, { "epoch": 1.21, "learning_rate": 1.774741192275931e-05, "loss": 0.1285, "step": 2280 }, { "epoch": 1.21, "learning_rate": 1.77452431870926e-05, "loss": 0.2335, "step": 2281 }, { "epoch": 1.21, "learning_rate": 1.7743073540571337e-05, "loss": 0.1768, "step": 2282 }, { "epoch": 1.21, "learning_rate": 1.774090298345068e-05, "loss": 0.1828, "step": 2283 }, { "epoch": 1.21, "learning_rate": 1.773873151598588e-05, "loss": 0.1786, "step": 2284 }, { "epoch": 1.21, "learning_rate": 1.7736559138432316e-05, "loss": 0.1781, "step": 2285 }, { "epoch": 1.21, "learning_rate": 1.773438585104546e-05, "loss": 0.1475, "step": 2286 }, { "epoch": 1.21, "learning_rate": 1.7732211654080895e-05, "loss": 0.197, "step": 2287 }, { "epoch": 1.21, "learning_rate": 1.7730036547794306e-05, "loss": 0.2203, "step": 2288 }, { "epoch": 1.21, "learning_rate": 1.7727860532441493e-05, "loss": 0.1418, "step": 2289 }, { "epoch": 1.21, "learning_rate": 1.772568360827836e-05, "loss": 0.1504, "step": 2290 }, { "epoch": 1.21, "learning_rate": 1.7723505775560915e-05, "loss": 0.1516, "step": 2291 }, { "epoch": 1.21, "learning_rate": 1.7721327034545276e-05, "loss": 0.2065, "step": 2292 }, { "epoch": 1.21, "learning_rate": 1.771914738548767e-05, "loss": 0.157, "step": 2293 }, { "epoch": 1.21, "learning_rate": 1.7716966828644425e-05, "loss": 0.2035, "step": 2294 }, { "epoch": 1.21, "learning_rate": 1.7714785364271975e-05, "loss": 0.164, "step": 2295 }, { "epoch": 1.22, "learning_rate": 1.771260299262687e-05, "loss": 0.2343, "step": 2296 }, { "epoch": 1.22, "learning_rate": 1.7710419713965756e-05, "loss": 0.1504, "step": 2297 }, { "epoch": 1.22, "learning_rate": 1.7708235528545398e-05, "loss": 0.122, "step": 2298 }, { "epoch": 1.22, "learning_rate": 1.7706050436622648e-05, "loss": 0.3432, "step": 2299 }, { "epoch": 1.22, "learning_rate": 1.770386443845449e-05, "loss": 0.1754, "step": 2300 }, { "epoch": 1.22, "learning_rate": 1.770167753429799e-05, "loss": 0.218, "step": 2301 }, { "epoch": 1.22, "learning_rate": 1.7699489724410338e-05, "loss": 0.1653, "step": 2302 }, { "epoch": 1.22, "learning_rate": 1.7697301009048823e-05, "loss": 0.1571, "step": 2303 }, { "epoch": 1.22, "learning_rate": 1.7695111388470844e-05, "loss": 0.1977, "step": 2304 }, { "epoch": 1.22, "learning_rate": 1.7692920862933903e-05, "loss": 0.1892, "step": 2305 }, { "epoch": 1.22, "learning_rate": 1.769072943269561e-05, "loss": 0.1333, "step": 2306 }, { "epoch": 1.22, "learning_rate": 1.7688537098013675e-05, "loss": 0.2545, "step": 2307 }, { "epoch": 1.22, "learning_rate": 1.768634385914593e-05, "loss": 0.1945, "step": 2308 }, { "epoch": 1.22, "learning_rate": 1.7684149716350298e-05, "loss": 0.1178, "step": 2309 }, { "epoch": 1.22, "learning_rate": 1.7681954669884817e-05, "loss": 0.167, "step": 2310 }, { "epoch": 1.22, "learning_rate": 1.7679758720007626e-05, "loss": 0.1425, "step": 2311 }, { "epoch": 1.22, "learning_rate": 1.7677561866976973e-05, "loss": 0.1757, "step": 2312 }, { "epoch": 1.22, "learning_rate": 1.7675364111051215e-05, "loss": 0.2889, "step": 2313 }, { "epoch": 1.22, "learning_rate": 1.767316545248881e-05, "loss": 0.2299, "step": 2314 }, { "epoch": 1.23, "learning_rate": 1.767096589154832e-05, "loss": 0.1562, "step": 2315 }, { "epoch": 1.23, "learning_rate": 1.766876542848843e-05, "loss": 0.184, "step": 2316 }, { "epoch": 1.23, "learning_rate": 1.76665640635679e-05, "loss": 0.2646, "step": 2317 }, { "epoch": 1.23, "learning_rate": 1.766436179704563e-05, "loss": 0.0729, "step": 2318 }, { "epoch": 1.23, "learning_rate": 1.7662158629180607e-05, "loss": 0.1911, "step": 2319 }, { "epoch": 1.23, "learning_rate": 1.765995456023192e-05, "loss": 0.1687, "step": 2320 }, { "epoch": 1.23, "learning_rate": 1.765774959045878e-05, "loss": 0.1653, "step": 2321 }, { "epoch": 1.23, "learning_rate": 1.7655543720120488e-05, "loss": 0.1408, "step": 2322 }, { "epoch": 1.23, "learning_rate": 1.7653336949476467e-05, "loss": 0.1459, "step": 2323 }, { "epoch": 1.23, "learning_rate": 1.765112927878623e-05, "loss": 0.1687, "step": 2324 }, { "epoch": 1.23, "learning_rate": 1.7648920708309406e-05, "loss": 0.1337, "step": 2325 }, { "epoch": 1.23, "learning_rate": 1.764671123830573e-05, "loss": 0.1123, "step": 2326 }, { "epoch": 1.23, "learning_rate": 1.764450086903503e-05, "loss": 0.1359, "step": 2327 }, { "epoch": 1.23, "learning_rate": 1.7642289600757263e-05, "loss": 0.1477, "step": 2328 }, { "epoch": 1.23, "learning_rate": 1.7640077433732465e-05, "loss": 0.2341, "step": 2329 }, { "epoch": 1.23, "learning_rate": 1.76378643682208e-05, "loss": 0.2019, "step": 2330 }, { "epoch": 1.23, "learning_rate": 1.7635650404482523e-05, "loss": 0.1408, "step": 2331 }, { "epoch": 1.23, "learning_rate": 1.7633435542778003e-05, "loss": 0.1656, "step": 2332 }, { "epoch": 1.23, "learning_rate": 1.763121978336771e-05, "loss": 0.1905, "step": 2333 }, { "epoch": 1.24, "learning_rate": 1.7629003126512226e-05, "loss": 0.2631, "step": 2334 }, { "epoch": 1.24, "learning_rate": 1.762678557247223e-05, "loss": 0.2133, "step": 2335 }, { "epoch": 1.24, "learning_rate": 1.7624567121508506e-05, "loss": 0.199, "step": 2336 }, { "epoch": 1.24, "learning_rate": 1.7622347773881956e-05, "loss": 0.1339, "step": 2337 }, { "epoch": 1.24, "learning_rate": 1.762012752985357e-05, "loss": 0.1713, "step": 2338 }, { "epoch": 1.24, "learning_rate": 1.7617906389684466e-05, "loss": 0.111, "step": 2339 }, { "epoch": 1.24, "learning_rate": 1.7615684353635843e-05, "loss": 0.1462, "step": 2340 }, { "epoch": 1.24, "learning_rate": 1.761346142196902e-05, "loss": 0.1479, "step": 2341 }, { "epoch": 1.24, "learning_rate": 1.7611237594945415e-05, "loss": 0.1477, "step": 2342 }, { "epoch": 1.24, "learning_rate": 1.760901287282656e-05, "loss": 0.1626, "step": 2343 }, { "epoch": 1.24, "learning_rate": 1.7606787255874084e-05, "loss": 0.1408, "step": 2344 }, { "epoch": 1.24, "learning_rate": 1.7604560744349723e-05, "loss": 0.1584, "step": 2345 }, { "epoch": 1.24, "learning_rate": 1.7602333338515314e-05, "loss": 0.2435, "step": 2346 }, { "epoch": 1.24, "learning_rate": 1.7600105038632808e-05, "loss": 0.143, "step": 2347 }, { "epoch": 1.24, "learning_rate": 1.759787584496426e-05, "loss": 0.2704, "step": 2348 }, { "epoch": 1.24, "learning_rate": 1.7595645757771824e-05, "loss": 0.0878, "step": 2349 }, { "epoch": 1.24, "learning_rate": 1.7593414777317763e-05, "loss": 0.1889, "step": 2350 }, { "epoch": 1.24, "learning_rate": 1.7591182903864443e-05, "loss": 0.2507, "step": 2351 }, { "epoch": 1.24, "learning_rate": 1.758895013767434e-05, "loss": 0.1712, "step": 2352 }, { "epoch": 1.25, "learning_rate": 1.7586716479010026e-05, "loss": 0.1928, "step": 2353 }, { "epoch": 1.25, "learning_rate": 1.7584481928134186e-05, "loss": 0.1599, "step": 2354 }, { "epoch": 1.25, "learning_rate": 1.7582246485309613e-05, "loss": 0.2555, "step": 2355 }, { "epoch": 1.25, "learning_rate": 1.7580010150799185e-05, "loss": 0.1547, "step": 2356 }, { "epoch": 1.25, "learning_rate": 1.7577772924865913e-05, "loss": 0.213, "step": 2357 }, { "epoch": 1.25, "learning_rate": 1.757553480777289e-05, "loss": 0.141, "step": 2358 }, { "epoch": 1.25, "learning_rate": 1.7573295799783328e-05, "loss": 0.2518, "step": 2359 }, { "epoch": 1.25, "learning_rate": 1.7571055901160537e-05, "loss": 0.1283, "step": 2360 }, { "epoch": 1.25, "learning_rate": 1.756881511216793e-05, "loss": 0.2337, "step": 2361 }, { "epoch": 1.25, "learning_rate": 1.756657343306903e-05, "loss": 0.2162, "step": 2362 }, { "epoch": 1.25, "learning_rate": 1.7564330864127464e-05, "loss": 0.1629, "step": 2363 }, { "epoch": 1.25, "learning_rate": 1.756208740560696e-05, "loss": 0.1343, "step": 2364 }, { "epoch": 1.25, "learning_rate": 1.7559843057771352e-05, "loss": 0.1155, "step": 2365 }, { "epoch": 1.25, "learning_rate": 1.755759782088458e-05, "loss": 0.1077, "step": 2366 }, { "epoch": 1.25, "learning_rate": 1.7555351695210695e-05, "loss": 0.1079, "step": 2367 }, { "epoch": 1.25, "learning_rate": 1.7553104681013833e-05, "loss": 0.208, "step": 2368 }, { "epoch": 1.25, "learning_rate": 1.7550856778558256e-05, "loss": 0.2097, "step": 2369 }, { "epoch": 1.25, "learning_rate": 1.7548607988108317e-05, "loss": 0.2261, "step": 2370 }, { "epoch": 1.25, "learning_rate": 1.7546358309928478e-05, "loss": 0.0999, "step": 2371 }, { "epoch": 1.26, "learning_rate": 1.754410774428331e-05, "loss": 0.1204, "step": 2372 }, { "epoch": 1.26, "learning_rate": 1.754185629143748e-05, "loss": 0.1969, "step": 2373 }, { "epoch": 1.26, "learning_rate": 1.7539603951655762e-05, "loss": 0.1592, "step": 2374 }, { "epoch": 1.26, "learning_rate": 1.7537350725203035e-05, "loss": 0.2633, "step": 2375 }, { "epoch": 1.26, "learning_rate": 1.7535096612344285e-05, "loss": 0.1466, "step": 2376 }, { "epoch": 1.26, "learning_rate": 1.7532841613344595e-05, "loss": 0.2279, "step": 2377 }, { "epoch": 1.26, "learning_rate": 1.7530585728469165e-05, "loss": 0.1242, "step": 2378 }, { "epoch": 1.26, "learning_rate": 1.7528328957983282e-05, "loss": 0.2717, "step": 2379 }, { "epoch": 1.26, "learning_rate": 1.7526071302152355e-05, "loss": 0.1739, "step": 2380 }, { "epoch": 1.26, "learning_rate": 1.7523812761241882e-05, "loss": 0.1452, "step": 2381 }, { "epoch": 1.26, "learning_rate": 1.7521553335517475e-05, "loss": 0.2127, "step": 2382 }, { "epoch": 1.26, "learning_rate": 1.7519293025244842e-05, "loss": 0.1795, "step": 2383 }, { "epoch": 1.26, "learning_rate": 1.7517031830689806e-05, "loss": 0.2172, "step": 2384 }, { "epoch": 1.26, "learning_rate": 1.751476975211828e-05, "loss": 0.1749, "step": 2385 }, { "epoch": 1.26, "learning_rate": 1.7512506789796297e-05, "loss": 0.1743, "step": 2386 }, { "epoch": 1.26, "learning_rate": 1.751024294398998e-05, "loss": 0.0987, "step": 2387 }, { "epoch": 1.26, "learning_rate": 1.7507978214965563e-05, "loss": 0.1205, "step": 2388 }, { "epoch": 1.26, "learning_rate": 1.750571260298938e-05, "loss": 0.1375, "step": 2389 }, { "epoch": 1.27, "learning_rate": 1.7503446108327875e-05, "loss": 0.1866, "step": 2390 }, { "epoch": 1.27, "learning_rate": 1.7501178731247586e-05, "loss": 0.2714, "step": 2391 }, { "epoch": 1.27, "learning_rate": 1.7498910472015168e-05, "loss": 0.1563, "step": 2392 }, { "epoch": 1.27, "learning_rate": 1.7496641330897368e-05, "loss": 0.1197, "step": 2393 }, { "epoch": 1.27, "learning_rate": 1.7494371308161042e-05, "loss": 0.1989, "step": 2394 }, { "epoch": 1.27, "learning_rate": 1.7492100404073146e-05, "loss": 0.2288, "step": 2395 }, { "epoch": 1.27, "learning_rate": 1.748982861890075e-05, "loss": 0.1597, "step": 2396 }, { "epoch": 1.27, "learning_rate": 1.748755595291101e-05, "loss": 0.1432, "step": 2397 }, { "epoch": 1.27, "learning_rate": 1.7485282406371203e-05, "loss": 0.2337, "step": 2398 }, { "epoch": 1.27, "learning_rate": 1.74830079795487e-05, "loss": 0.1142, "step": 2399 }, { "epoch": 1.27, "learning_rate": 1.7480732672710976e-05, "loss": 0.143, "step": 2400 }, { "epoch": 1.27, "learning_rate": 1.7478456486125614e-05, "loss": 0.2129, "step": 2401 }, { "epoch": 1.27, "learning_rate": 1.7476179420060298e-05, "loss": 0.1027, "step": 2402 }, { "epoch": 1.27, "learning_rate": 1.747390147478281e-05, "loss": 0.1712, "step": 2403 }, { "epoch": 1.27, "learning_rate": 1.7471622650561044e-05, "loss": 0.259, "step": 2404 }, { "epoch": 1.27, "learning_rate": 1.7469342947662996e-05, "loss": 0.1755, "step": 2405 }, { "epoch": 1.27, "learning_rate": 1.7467062366356756e-05, "loss": 0.1795, "step": 2406 }, { "epoch": 1.27, "learning_rate": 1.7464780906910534e-05, "loss": 0.2355, "step": 2407 }, { "epoch": 1.27, "learning_rate": 1.7462498569592627e-05, "loss": 0.1278, "step": 2408 }, { "epoch": 1.28, "learning_rate": 1.7460215354671445e-05, "loss": 0.213, "step": 2409 }, { "epoch": 1.28, "learning_rate": 1.74579312624155e-05, "loss": 0.2698, "step": 2410 }, { "epoch": 1.28, "learning_rate": 1.74556462930934e-05, "loss": 0.2845, "step": 2411 }, { "epoch": 1.28, "learning_rate": 1.745336044697386e-05, "loss": 0.1451, "step": 2412 }, { "epoch": 1.28, "learning_rate": 1.7451073724325706e-05, "loss": 0.23, "step": 2413 }, { "epoch": 1.28, "learning_rate": 1.744878612541786e-05, "loss": 0.2159, "step": 2414 }, { "epoch": 1.28, "learning_rate": 1.7446497650519347e-05, "loss": 0.1432, "step": 2415 }, { "epoch": 1.28, "learning_rate": 1.7444208299899293e-05, "loss": 0.2855, "step": 2416 }, { "epoch": 1.28, "learning_rate": 1.744191807382693e-05, "loss": 0.0743, "step": 2417 }, { "epoch": 1.28, "learning_rate": 1.7439626972571598e-05, "loss": 0.1922, "step": 2418 }, { "epoch": 1.28, "learning_rate": 1.743733499640273e-05, "loss": 0.1723, "step": 2419 }, { "epoch": 1.28, "learning_rate": 1.7435042145589866e-05, "loss": 0.2313, "step": 2420 }, { "epoch": 1.28, "learning_rate": 1.7432748420402653e-05, "loss": 0.1953, "step": 2421 }, { "epoch": 1.28, "learning_rate": 1.7430453821110832e-05, "loss": 0.2202, "step": 2422 }, { "epoch": 1.28, "learning_rate": 1.7428158347984257e-05, "loss": 0.2771, "step": 2423 }, { "epoch": 1.28, "learning_rate": 1.7425862001292882e-05, "loss": 0.2567, "step": 2424 }, { "epoch": 1.28, "learning_rate": 1.7423564781306753e-05, "loss": 0.2053, "step": 2425 }, { "epoch": 1.28, "learning_rate": 1.742126668829603e-05, "loss": 0.162, "step": 2426 }, { "epoch": 1.28, "learning_rate": 1.741896772253098e-05, "loss": 0.1994, "step": 2427 }, { "epoch": 1.29, "learning_rate": 1.7416667884281958e-05, "loss": 0.1345, "step": 2428 }, { "epoch": 1.29, "learning_rate": 1.7414367173819428e-05, "loss": 0.1746, "step": 2429 }, { "epoch": 1.29, "learning_rate": 1.7412065591413965e-05, "loss": 0.206, "step": 2430 }, { "epoch": 1.29, "learning_rate": 1.740976313733623e-05, "loss": 0.2251, "step": 2431 }, { "epoch": 1.29, "learning_rate": 1.7407459811857007e-05, "loss": 0.1464, "step": 2432 }, { "epoch": 1.29, "learning_rate": 1.740515561524716e-05, "loss": 0.1216, "step": 2433 }, { "epoch": 1.29, "learning_rate": 1.7402850547777673e-05, "loss": 0.2175, "step": 2434 }, { "epoch": 1.29, "learning_rate": 1.7400544609719626e-05, "loss": 0.1727, "step": 2435 }, { "epoch": 1.29, "learning_rate": 1.7398237801344204e-05, "loss": 0.241, "step": 2436 }, { "epoch": 1.29, "learning_rate": 1.739593012292268e-05, "loss": 0.2485, "step": 2437 }, { "epoch": 1.29, "learning_rate": 1.7393621574726457e-05, "loss": 0.1055, "step": 2438 }, { "epoch": 1.29, "learning_rate": 1.739131215702701e-05, "loss": 0.1723, "step": 2439 }, { "epoch": 1.29, "learning_rate": 1.7389001870095942e-05, "loss": 0.2625, "step": 2440 }, { "epoch": 1.29, "learning_rate": 1.738669071420494e-05, "loss": 0.1739, "step": 2441 }, { "epoch": 1.29, "learning_rate": 1.73843786896258e-05, "loss": 0.164, "step": 2442 }, { "epoch": 1.29, "learning_rate": 1.738206579663043e-05, "loss": 0.2933, "step": 2443 }, { "epoch": 1.29, "learning_rate": 1.7379752035490817e-05, "loss": 0.168, "step": 2444 }, { "epoch": 1.29, "learning_rate": 1.7377437406479072e-05, "loss": 0.2146, "step": 2445 }, { "epoch": 1.29, "learning_rate": 1.7375121909867396e-05, "loss": 0.222, "step": 2446 }, { "epoch": 1.3, "learning_rate": 1.7372805545928097e-05, "loss": 0.283, "step": 2447 }, { "epoch": 1.3, "learning_rate": 1.737048831493358e-05, "loss": 0.3651, "step": 2448 }, { "epoch": 1.3, "learning_rate": 1.7368170217156363e-05, "loss": 0.2472, "step": 2449 }, { "epoch": 1.3, "learning_rate": 1.7365851252869054e-05, "loss": 0.1841, "step": 2450 }, { "epoch": 1.3, "learning_rate": 1.7363531422344365e-05, "loss": 0.1103, "step": 2451 }, { "epoch": 1.3, "learning_rate": 1.736121072585512e-05, "loss": 0.1857, "step": 2452 }, { "epoch": 1.3, "learning_rate": 1.7358889163674227e-05, "loss": 0.1447, "step": 2453 }, { "epoch": 1.3, "learning_rate": 1.735656673607471e-05, "loss": 0.2231, "step": 2454 }, { "epoch": 1.3, "learning_rate": 1.7354243443329694e-05, "loss": 0.1425, "step": 2455 }, { "epoch": 1.3, "learning_rate": 1.73519192857124e-05, "loss": 0.1579, "step": 2456 }, { "epoch": 1.3, "learning_rate": 1.7349594263496152e-05, "loss": 0.2199, "step": 2457 }, { "epoch": 1.3, "learning_rate": 1.734726837695438e-05, "loss": 0.2503, "step": 2458 }, { "epoch": 1.3, "learning_rate": 1.7344941626360606e-05, "loss": 0.2561, "step": 2459 }, { "epoch": 1.3, "learning_rate": 1.7342614011988468e-05, "loss": 0.3827, "step": 2460 }, { "epoch": 1.3, "learning_rate": 1.7340285534111695e-05, "loss": 0.2257, "step": 2461 }, { "epoch": 1.3, "learning_rate": 1.7337956193004113e-05, "loss": 0.1584, "step": 2462 }, { "epoch": 1.3, "learning_rate": 1.733562598893967e-05, "loss": 0.1664, "step": 2463 }, { "epoch": 1.3, "learning_rate": 1.733329492219239e-05, "loss": 0.1308, "step": 2464 }, { "epoch": 1.3, "learning_rate": 1.7330962993036416e-05, "loss": 0.2276, "step": 2465 }, { "epoch": 1.31, "learning_rate": 1.732863020174599e-05, "loss": 0.1884, "step": 2466 }, { "epoch": 1.31, "learning_rate": 1.7326296548595447e-05, "loss": 0.1293, "step": 2467 }, { "epoch": 1.31, "learning_rate": 1.7323962033859232e-05, "loss": 0.1915, "step": 2468 }, { "epoch": 1.31, "learning_rate": 1.7321626657811887e-05, "loss": 0.1311, "step": 2469 }, { "epoch": 1.31, "learning_rate": 1.7319290420728056e-05, "loss": 0.1551, "step": 2470 }, { "epoch": 1.31, "learning_rate": 1.7316953322882487e-05, "loss": 0.1887, "step": 2471 }, { "epoch": 1.31, "learning_rate": 1.7314615364550024e-05, "loss": 0.2203, "step": 2472 }, { "epoch": 1.31, "learning_rate": 1.7312276546005615e-05, "loss": 0.1471, "step": 2473 }, { "epoch": 1.31, "learning_rate": 1.7309936867524314e-05, "loss": 0.2768, "step": 2474 }, { "epoch": 1.31, "learning_rate": 1.7307596329381263e-05, "loss": 0.2719, "step": 2475 }, { "epoch": 1.31, "learning_rate": 1.730525493185172e-05, "loss": 0.0854, "step": 2476 }, { "epoch": 1.31, "learning_rate": 1.730291267521104e-05, "loss": 0.1523, "step": 2477 }, { "epoch": 1.31, "learning_rate": 1.730056955973467e-05, "loss": 0.2208, "step": 2478 }, { "epoch": 1.31, "learning_rate": 1.729822558569817e-05, "loss": 0.2001, "step": 2479 }, { "epoch": 1.31, "learning_rate": 1.729588075337719e-05, "loss": 0.2361, "step": 2480 }, { "epoch": 1.31, "learning_rate": 1.7293535063047487e-05, "loss": 0.1166, "step": 2481 }, { "epoch": 1.31, "learning_rate": 1.7291188514984924e-05, "loss": 0.1454, "step": 2482 }, { "epoch": 1.31, "learning_rate": 1.7288841109465454e-05, "loss": 0.0983, "step": 2483 }, { "epoch": 1.31, "learning_rate": 1.7286492846765142e-05, "loss": 0.2073, "step": 2484 }, { "epoch": 1.32, "learning_rate": 1.728414372716014e-05, "loss": 0.1268, "step": 2485 }, { "epoch": 1.32, "learning_rate": 1.7281793750926717e-05, "loss": 0.143, "step": 2486 }, { "epoch": 1.32, "learning_rate": 1.7279442918341227e-05, "loss": 0.2117, "step": 2487 }, { "epoch": 1.32, "learning_rate": 1.7277091229680136e-05, "loss": 0.0796, "step": 2488 }, { "epoch": 1.32, "learning_rate": 1.7274738685220005e-05, "loss": 0.1059, "step": 2489 }, { "epoch": 1.32, "learning_rate": 1.72723852852375e-05, "loss": 0.253, "step": 2490 }, { "epoch": 1.32, "learning_rate": 1.7270031030009385e-05, "loss": 0.2042, "step": 2491 }, { "epoch": 1.32, "learning_rate": 1.7267675919812524e-05, "loss": 0.1804, "step": 2492 }, { "epoch": 1.32, "learning_rate": 1.7265319954923878e-05, "loss": 0.1745, "step": 2493 }, { "epoch": 1.32, "learning_rate": 1.7262963135620518e-05, "loss": 0.1734, "step": 2494 }, { "epoch": 1.32, "learning_rate": 1.726060546217961e-05, "loss": 0.2806, "step": 2495 }, { "epoch": 1.32, "learning_rate": 1.725824693487842e-05, "loss": 0.0769, "step": 2496 }, { "epoch": 1.32, "learning_rate": 1.725588755399431e-05, "loss": 0.1786, "step": 2497 }, { "epoch": 1.32, "learning_rate": 1.7253527319804754e-05, "loss": 0.0852, "step": 2498 }, { "epoch": 1.32, "learning_rate": 1.7251166232587318e-05, "loss": 0.1722, "step": 2499 }, { "epoch": 1.32, "learning_rate": 1.724880429261967e-05, "loss": 0.2642, "step": 2500 }, { "epoch": 1.32, "learning_rate": 1.7246441500179573e-05, "loss": 0.1824, "step": 2501 }, { "epoch": 1.32, "learning_rate": 1.7244077855544904e-05, "loss": 0.1661, "step": 2502 }, { "epoch": 1.32, "learning_rate": 1.7241713358993634e-05, "loss": 0.2386, "step": 2503 }, { "epoch": 1.33, "learning_rate": 1.7239348010803818e-05, "loss": 0.1133, "step": 2504 }, { "epoch": 1.33, "learning_rate": 1.7236981811253638e-05, "loss": 0.1591, "step": 2505 }, { "epoch": 1.33, "learning_rate": 1.7234614760621357e-05, "loss": 0.1516, "step": 2506 }, { "epoch": 1.33, "learning_rate": 1.723224685918535e-05, "loss": 0.1034, "step": 2507 }, { "epoch": 1.33, "learning_rate": 1.7229878107224077e-05, "loss": 0.0995, "step": 2508 }, { "epoch": 1.33, "learning_rate": 1.722750850501612e-05, "loss": 0.1021, "step": 2509 }, { "epoch": 1.33, "learning_rate": 1.7225138052840137e-05, "loss": 0.2762, "step": 2510 }, { "epoch": 1.33, "learning_rate": 1.7222766750974906e-05, "loss": 0.1723, "step": 2511 }, { "epoch": 1.33, "learning_rate": 1.722039459969929e-05, "loss": 0.1678, "step": 2512 }, { "epoch": 1.33, "learning_rate": 1.7218021599292264e-05, "loss": 0.1895, "step": 2513 }, { "epoch": 1.33, "learning_rate": 1.7215647750032893e-05, "loss": 0.1747, "step": 2514 }, { "epoch": 1.33, "learning_rate": 1.7213273052200346e-05, "loss": 0.1607, "step": 2515 }, { "epoch": 1.33, "learning_rate": 1.7210897506073895e-05, "loss": 0.197, "step": 2516 }, { "epoch": 1.33, "learning_rate": 1.72085211119329e-05, "loss": 0.2523, "step": 2517 }, { "epoch": 1.33, "learning_rate": 1.7206143870056843e-05, "loss": 0.2375, "step": 2518 }, { "epoch": 1.33, "learning_rate": 1.720376578072528e-05, "loss": 0.2328, "step": 2519 }, { "epoch": 1.33, "learning_rate": 1.7201386844217887e-05, "loss": 0.1404, "step": 2520 }, { "epoch": 1.33, "learning_rate": 1.7199007060814426e-05, "loss": 0.225, "step": 2521 }, { "epoch": 1.33, "learning_rate": 1.7196626430794765e-05, "loss": 0.1215, "step": 2522 }, { "epoch": 1.34, "learning_rate": 1.719424495443887e-05, "loss": 0.1455, "step": 2523 }, { "epoch": 1.34, "learning_rate": 1.7191862632026807e-05, "loss": 0.1104, "step": 2524 }, { "epoch": 1.34, "learning_rate": 1.718947946383874e-05, "loss": 0.328, "step": 2525 }, { "epoch": 1.34, "learning_rate": 1.7187095450154938e-05, "loss": 0.2066, "step": 2526 }, { "epoch": 1.34, "learning_rate": 1.7184710591255766e-05, "loss": 0.2062, "step": 2527 }, { "epoch": 1.34, "learning_rate": 1.718232488742168e-05, "loss": 0.1906, "step": 2528 }, { "epoch": 1.34, "learning_rate": 1.717993833893325e-05, "loss": 0.2014, "step": 2529 }, { "epoch": 1.34, "learning_rate": 1.7177550946071132e-05, "loss": 0.1138, "step": 2530 }, { "epoch": 1.34, "learning_rate": 1.71751627091161e-05, "loss": 0.1464, "step": 2531 }, { "epoch": 1.34, "learning_rate": 1.7172773628349002e-05, "loss": 0.2253, "step": 2532 }, { "epoch": 1.34, "learning_rate": 1.7170383704050802e-05, "loss": 0.2292, "step": 2533 }, { "epoch": 1.34, "learning_rate": 1.716799293650256e-05, "loss": 0.1689, "step": 2534 }, { "epoch": 1.34, "learning_rate": 1.7165601325985437e-05, "loss": 0.1921, "step": 2535 }, { "epoch": 1.34, "learning_rate": 1.716320887278069e-05, "loss": 0.1228, "step": 2536 }, { "epoch": 1.34, "learning_rate": 1.7160815577169674e-05, "loss": 0.1069, "step": 2537 }, { "epoch": 1.34, "learning_rate": 1.7158421439433845e-05, "loss": 0.2052, "step": 2538 }, { "epoch": 1.34, "learning_rate": 1.7156026459854757e-05, "loss": 0.0874, "step": 2539 }, { "epoch": 1.34, "learning_rate": 1.7153630638714064e-05, "loss": 0.1381, "step": 2540 }, { "epoch": 1.34, "learning_rate": 1.7151233976293526e-05, "loss": 0.2154, "step": 2541 }, { "epoch": 1.35, "learning_rate": 1.7148836472874984e-05, "loss": 0.1172, "step": 2542 }, { "epoch": 1.35, "learning_rate": 1.7146438128740395e-05, "loss": 0.2245, "step": 2543 }, { "epoch": 1.35, "learning_rate": 1.7144038944171808e-05, "loss": 0.1928, "step": 2544 }, { "epoch": 1.35, "learning_rate": 1.7141638919451367e-05, "loss": 0.1948, "step": 2545 }, { "epoch": 1.35, "learning_rate": 1.7139238054861325e-05, "loss": 0.1474, "step": 2546 }, { "epoch": 1.35, "learning_rate": 1.713683635068403e-05, "loss": 0.1139, "step": 2547 }, { "epoch": 1.35, "learning_rate": 1.7134433807201918e-05, "loss": 0.212, "step": 2548 }, { "epoch": 1.35, "learning_rate": 1.7132030424697536e-05, "loss": 0.1962, "step": 2549 }, { "epoch": 1.35, "learning_rate": 1.712962620345353e-05, "loss": 0.1718, "step": 2550 }, { "epoch": 1.35, "learning_rate": 1.712722114375264e-05, "loss": 0.2016, "step": 2551 }, { "epoch": 1.35, "learning_rate": 1.7124815245877703e-05, "loss": 0.1264, "step": 2552 }, { "epoch": 1.35, "learning_rate": 1.7122408510111653e-05, "loss": 0.2133, "step": 2553 }, { "epoch": 1.35, "learning_rate": 1.712000093673753e-05, "loss": 0.2911, "step": 2554 }, { "epoch": 1.35, "learning_rate": 1.7117592526038477e-05, "loss": 0.187, "step": 2555 }, { "epoch": 1.35, "learning_rate": 1.7115183278297715e-05, "loss": 0.1843, "step": 2556 }, { "epoch": 1.35, "learning_rate": 1.711277319379858e-05, "loss": 0.1795, "step": 2557 }, { "epoch": 1.35, "learning_rate": 1.7110362272824505e-05, "loss": 0.1484, "step": 2558 }, { "epoch": 1.35, "learning_rate": 1.7107950515659014e-05, "loss": 0.2446, "step": 2559 }, { "epoch": 1.36, "learning_rate": 1.7105537922585744e-05, "loss": 0.1871, "step": 2560 }, { "epoch": 1.36, "learning_rate": 1.7103124493888406e-05, "loss": 0.1532, "step": 2561 }, { "epoch": 1.36, "learning_rate": 1.7100710229850835e-05, "loss": 0.3215, "step": 2562 }, { "epoch": 1.36, "learning_rate": 1.7098295130756945e-05, "loss": 0.1474, "step": 2563 }, { "epoch": 1.36, "learning_rate": 1.7095879196890765e-05, "loss": 0.215, "step": 2564 }, { "epoch": 1.36, "learning_rate": 1.7093462428536402e-05, "loss": 0.145, "step": 2565 }, { "epoch": 1.36, "learning_rate": 1.709104482597808e-05, "loss": 0.1067, "step": 2566 }, { "epoch": 1.36, "learning_rate": 1.708862638950011e-05, "loss": 0.0992, "step": 2567 }, { "epoch": 1.36, "learning_rate": 1.708620711938691e-05, "loss": 0.2177, "step": 2568 }, { "epoch": 1.36, "learning_rate": 1.7083787015922985e-05, "loss": 0.148, "step": 2569 }, { "epoch": 1.36, "learning_rate": 1.7081366079392946e-05, "loss": 0.2095, "step": 2570 }, { "epoch": 1.36, "learning_rate": 1.7078944310081497e-05, "loss": 0.1632, "step": 2571 }, { "epoch": 1.36, "learning_rate": 1.7076521708273442e-05, "loss": 0.1675, "step": 2572 }, { "epoch": 1.36, "learning_rate": 1.7074098274253684e-05, "loss": 0.1864, "step": 2573 }, { "epoch": 1.36, "learning_rate": 1.707167400830723e-05, "loss": 0.1501, "step": 2574 }, { "epoch": 1.36, "learning_rate": 1.706924891071917e-05, "loss": 0.1548, "step": 2575 }, { "epoch": 1.36, "learning_rate": 1.70668229817747e-05, "loss": 0.2251, "step": 2576 }, { "epoch": 1.36, "learning_rate": 1.7064396221759122e-05, "loss": 0.1583, "step": 2577 }, { "epoch": 1.36, "learning_rate": 1.7061968630957817e-05, "loss": 0.1096, "step": 2578 }, { "epoch": 1.37, "learning_rate": 1.7059540209656278e-05, "loss": 0.1304, "step": 2579 }, { "epoch": 1.37, "learning_rate": 1.7057110958140092e-05, "loss": 0.2354, "step": 2580 }, { "epoch": 1.37, "learning_rate": 1.7054680876694947e-05, "loss": 0.2302, "step": 2581 }, { "epoch": 1.37, "learning_rate": 1.7052249965606616e-05, "loss": 0.1162, "step": 2582 }, { "epoch": 1.37, "learning_rate": 1.7049818225160988e-05, "loss": 0.1527, "step": 2583 }, { "epoch": 1.37, "learning_rate": 1.7047385655644035e-05, "loss": 0.1681, "step": 2584 }, { "epoch": 1.37, "learning_rate": 1.7044952257341832e-05, "loss": 0.11, "step": 2585 }, { "epoch": 1.37, "learning_rate": 1.7042518030540553e-05, "loss": 0.1036, "step": 2586 }, { "epoch": 1.37, "learning_rate": 1.7040082975526467e-05, "loss": 0.267, "step": 2587 }, { "epoch": 1.37, "learning_rate": 1.7037647092585935e-05, "loss": 0.2747, "step": 2588 }, { "epoch": 1.37, "learning_rate": 1.7035210382005432e-05, "loss": 0.2948, "step": 2589 }, { "epoch": 1.37, "learning_rate": 1.7032772844071507e-05, "loss": 0.1635, "step": 2590 }, { "epoch": 1.37, "learning_rate": 1.7030334479070832e-05, "loss": 0.2145, "step": 2591 }, { "epoch": 1.37, "learning_rate": 1.702789528729015e-05, "loss": 0.1468, "step": 2592 }, { "epoch": 1.37, "learning_rate": 1.7025455269016328e-05, "loss": 0.1123, "step": 2593 }, { "epoch": 1.37, "learning_rate": 1.7023014424536306e-05, "loss": 0.1828, "step": 2594 }, { "epoch": 1.37, "learning_rate": 1.7020572754137134e-05, "loss": 0.2391, "step": 2595 }, { "epoch": 1.37, "learning_rate": 1.701813025810596e-05, "loss": 0.1395, "step": 2596 }, { "epoch": 1.37, "learning_rate": 1.7015686936730022e-05, "loss": 0.1079, "step": 2597 }, { "epoch": 1.38, "learning_rate": 1.701324279029666e-05, "loss": 0.2438, "step": 2598 }, { "epoch": 1.38, "learning_rate": 1.7010797819093316e-05, "loss": 0.1389, "step": 2599 }, { "epoch": 1.38, "learning_rate": 1.7008352023407515e-05, "loss": 0.1682, "step": 2600 }, { "epoch": 1.38, "learning_rate": 1.7005905403526892e-05, "loss": 0.1697, "step": 2601 }, { "epoch": 1.38, "learning_rate": 1.700345795973917e-05, "loss": 0.1949, "step": 2602 }, { "epoch": 1.38, "learning_rate": 1.7001009692332178e-05, "loss": 0.1538, "step": 2603 }, { "epoch": 1.38, "learning_rate": 1.699856060159383e-05, "loss": 0.2733, "step": 2604 }, { "epoch": 1.38, "learning_rate": 1.6996110687812145e-05, "loss": 0.1367, "step": 2605 }, { "epoch": 1.38, "learning_rate": 1.6993659951275243e-05, "loss": 0.1212, "step": 2606 }, { "epoch": 1.38, "learning_rate": 1.699120839227133e-05, "loss": 0.1431, "step": 2607 }, { "epoch": 1.38, "learning_rate": 1.6988756011088714e-05, "loss": 0.1977, "step": 2608 }, { "epoch": 1.38, "learning_rate": 1.6986302808015803e-05, "loss": 0.1584, "step": 2609 }, { "epoch": 1.38, "learning_rate": 1.698384878334109e-05, "loss": 0.1219, "step": 2610 }, { "epoch": 1.38, "learning_rate": 1.6981393937353182e-05, "loss": 0.1769, "step": 2611 }, { "epoch": 1.38, "learning_rate": 1.6978938270340772e-05, "loss": 0.1521, "step": 2612 }, { "epoch": 1.38, "learning_rate": 1.6976481782592643e-05, "loss": 0.1001, "step": 2613 }, { "epoch": 1.38, "learning_rate": 1.697402447439769e-05, "loss": 0.129, "step": 2614 }, { "epoch": 1.38, "learning_rate": 1.6971566346044898e-05, "loss": 0.2185, "step": 2615 }, { "epoch": 1.38, "learning_rate": 1.6969107397823338e-05, "loss": 0.1648, "step": 2616 }, { "epoch": 1.39, "learning_rate": 1.6966647630022198e-05, "loss": 0.0684, "step": 2617 }, { "epoch": 1.39, "learning_rate": 1.696418704293074e-05, "loss": 0.11, "step": 2618 }, { "epoch": 1.39, "learning_rate": 1.6961725636838346e-05, "loss": 0.1235, "step": 2619 }, { "epoch": 1.39, "learning_rate": 1.695926341203447e-05, "loss": 0.2034, "step": 2620 }, { "epoch": 1.39, "learning_rate": 1.6956800368808682e-05, "loss": 0.1189, "step": 2621 }, { "epoch": 1.39, "learning_rate": 1.695433650745064e-05, "loss": 0.2264, "step": 2622 }, { "epoch": 1.39, "learning_rate": 1.6951871828250087e-05, "loss": 0.2022, "step": 2623 }, { "epoch": 1.39, "learning_rate": 1.6949406331496888e-05, "loss": 0.1263, "step": 2624 }, { "epoch": 1.39, "learning_rate": 1.6946940017480986e-05, "loss": 0.1383, "step": 2625 }, { "epoch": 1.39, "learning_rate": 1.694447288649242e-05, "loss": 0.1645, "step": 2626 }, { "epoch": 1.39, "learning_rate": 1.6942004938821336e-05, "loss": 0.1783, "step": 2627 }, { "epoch": 1.39, "learning_rate": 1.693953617475796e-05, "loss": 0.1733, "step": 2628 }, { "epoch": 1.39, "learning_rate": 1.693706659459263e-05, "loss": 0.2201, "step": 2629 }, { "epoch": 1.39, "learning_rate": 1.693459619861577e-05, "loss": 0.1103, "step": 2630 }, { "epoch": 1.39, "learning_rate": 1.69321249871179e-05, "loss": 0.1375, "step": 2631 }, { "epoch": 1.39, "learning_rate": 1.692965296038965e-05, "loss": 0.1201, "step": 2632 }, { "epoch": 1.39, "learning_rate": 1.6927180118721725e-05, "loss": 0.1256, "step": 2633 }, { "epoch": 1.39, "learning_rate": 1.6924706462404936e-05, "loss": 0.1257, "step": 2634 }, { "epoch": 1.39, "learning_rate": 1.6922231991730193e-05, "loss": 0.19, "step": 2635 }, { "epoch": 1.4, "learning_rate": 1.6919756706988497e-05, "loss": 0.191, "step": 2636 }, { "epoch": 1.4, "learning_rate": 1.691728060847095e-05, "loss": 0.277, "step": 2637 }, { "epoch": 1.4, "learning_rate": 1.6914803696468735e-05, "loss": 0.1551, "step": 2638 }, { "epoch": 1.4, "learning_rate": 1.691232597127315e-05, "loss": 0.2132, "step": 2639 }, { "epoch": 1.4, "learning_rate": 1.6909847433175577e-05, "loss": 0.209, "step": 2640 }, { "epoch": 1.4, "learning_rate": 1.69073680824675e-05, "loss": 0.1445, "step": 2641 }, { "epoch": 1.4, "learning_rate": 1.690488791944049e-05, "loss": 0.2641, "step": 2642 }, { "epoch": 1.4, "learning_rate": 1.6902406944386224e-05, "loss": 0.178, "step": 2643 }, { "epoch": 1.4, "learning_rate": 1.6899925157596466e-05, "loss": 0.1747, "step": 2644 }, { "epoch": 1.4, "learning_rate": 1.6897442559363076e-05, "loss": 0.2519, "step": 2645 }, { "epoch": 1.4, "learning_rate": 1.6894959149978013e-05, "loss": 0.137, "step": 2646 }, { "epoch": 1.4, "learning_rate": 1.6892474929733334e-05, "loss": 0.2212, "step": 2647 }, { "epoch": 1.4, "learning_rate": 1.688998989892119e-05, "loss": 0.1539, "step": 2648 }, { "epoch": 1.4, "learning_rate": 1.6887504057833815e-05, "loss": 0.1403, "step": 2649 }, { "epoch": 1.4, "learning_rate": 1.688501740676356e-05, "loss": 0.2052, "step": 2650 }, { "epoch": 1.4, "learning_rate": 1.688252994600285e-05, "loss": 0.08, "step": 2651 }, { "epoch": 1.4, "learning_rate": 1.6880041675844217e-05, "loss": 0.2357, "step": 2652 }, { "epoch": 1.4, "learning_rate": 1.6877552596580295e-05, "loss": 0.2255, "step": 2653 }, { "epoch": 1.4, "learning_rate": 1.6875062708503795e-05, "loss": 0.192, "step": 2654 }, { "epoch": 1.41, "learning_rate": 1.6872572011907533e-05, "loss": 0.1335, "step": 2655 }, { "epoch": 1.41, "learning_rate": 1.687008050708442e-05, "loss": 0.1699, "step": 2656 }, { "epoch": 1.41, "learning_rate": 1.6867588194327466e-05, "loss": 0.1445, "step": 2657 }, { "epoch": 1.41, "learning_rate": 1.6865095073929763e-05, "loss": 0.202, "step": 2658 }, { "epoch": 1.41, "learning_rate": 1.6862601146184517e-05, "loss": 0.1557, "step": 2659 }, { "epoch": 1.41, "learning_rate": 1.6860106411385008e-05, "loss": 0.1673, "step": 2660 }, { "epoch": 1.41, "learning_rate": 1.685761086982463e-05, "loss": 0.1377, "step": 2661 }, { "epoch": 1.41, "learning_rate": 1.6855114521796855e-05, "loss": 0.1563, "step": 2662 }, { "epoch": 1.41, "learning_rate": 1.6852617367595265e-05, "loss": 0.1139, "step": 2663 }, { "epoch": 1.41, "learning_rate": 1.6850119407513526e-05, "loss": 0.2157, "step": 2664 }, { "epoch": 1.41, "learning_rate": 1.6847620641845405e-05, "loss": 0.1071, "step": 2665 }, { "epoch": 1.41, "learning_rate": 1.684512107088476e-05, "loss": 0.0998, "step": 2666 }, { "epoch": 1.41, "learning_rate": 1.6842620694925544e-05, "loss": 0.166, "step": 2667 }, { "epoch": 1.41, "learning_rate": 1.6840119514261807e-05, "loss": 0.1722, "step": 2668 }, { "epoch": 1.41, "learning_rate": 1.6837617529187692e-05, "loss": 0.2063, "step": 2669 }, { "epoch": 1.41, "learning_rate": 1.6835114739997438e-05, "loss": 0.1644, "step": 2670 }, { "epoch": 1.41, "learning_rate": 1.6832611146985372e-05, "loss": 0.1844, "step": 2671 }, { "epoch": 1.41, "learning_rate": 1.6830106750445933e-05, "loss": 0.0943, "step": 2672 }, { "epoch": 1.41, "learning_rate": 1.682760155067363e-05, "loss": 0.1164, "step": 2673 }, { "epoch": 1.42, "learning_rate": 1.682509554796309e-05, "loss": 0.1865, "step": 2674 }, { "epoch": 1.42, "learning_rate": 1.6822588742609012e-05, "loss": 0.2387, "step": 2675 }, { "epoch": 1.42, "learning_rate": 1.682008113490621e-05, "loss": 0.1695, "step": 2676 }, { "epoch": 1.42, "learning_rate": 1.6817572725149575e-05, "loss": 0.144, "step": 2677 }, { "epoch": 1.42, "learning_rate": 1.681506351363411e-05, "loss": 0.1609, "step": 2678 }, { "epoch": 1.42, "learning_rate": 1.6812553500654897e-05, "loss": 0.1916, "step": 2679 }, { "epoch": 1.42, "learning_rate": 1.6810042686507117e-05, "loss": 0.1749, "step": 2680 }, { "epoch": 1.42, "learning_rate": 1.680753107148605e-05, "loss": 0.1892, "step": 2681 }, { "epoch": 1.42, "learning_rate": 1.6805018655887066e-05, "loss": 0.1797, "step": 2682 }, { "epoch": 1.42, "learning_rate": 1.6802505440005624e-05, "loss": 0.2239, "step": 2683 }, { "epoch": 1.42, "learning_rate": 1.679999142413729e-05, "loss": 0.1503, "step": 2684 }, { "epoch": 1.42, "learning_rate": 1.6797476608577717e-05, "loss": 0.2506, "step": 2685 }, { "epoch": 1.42, "learning_rate": 1.6794960993622646e-05, "loss": 0.1425, "step": 2686 }, { "epoch": 1.42, "learning_rate": 1.679244457956792e-05, "loss": 0.2479, "step": 2687 }, { "epoch": 1.42, "learning_rate": 1.6789927366709474e-05, "loss": 0.1896, "step": 2688 }, { "epoch": 1.42, "learning_rate": 1.6787409355343334e-05, "loss": 0.2012, "step": 2689 }, { "epoch": 1.42, "learning_rate": 1.6784890545765634e-05, "loss": 0.2274, "step": 2690 }, { "epoch": 1.42, "learning_rate": 1.6782370938272577e-05, "loss": 0.1015, "step": 2691 }, { "epoch": 1.42, "learning_rate": 1.677985053316048e-05, "loss": 0.2399, "step": 2692 }, { "epoch": 1.43, "learning_rate": 1.677732933072574e-05, "loss": 0.2384, "step": 2693 }, { "epoch": 1.43, "learning_rate": 1.677480733126487e-05, "loss": 0.1642, "step": 2694 }, { "epoch": 1.43, "learning_rate": 1.677228453507445e-05, "loss": 0.1663, "step": 2695 }, { "epoch": 1.43, "learning_rate": 1.6769760942451165e-05, "loss": 0.2222, "step": 2696 }, { "epoch": 1.43, "learning_rate": 1.67672365536918e-05, "loss": 0.2464, "step": 2697 }, { "epoch": 1.43, "learning_rate": 1.6764711369093223e-05, "loss": 0.23, "step": 2698 }, { "epoch": 1.43, "learning_rate": 1.6762185388952402e-05, "loss": 0.1166, "step": 2699 }, { "epoch": 1.43, "learning_rate": 1.6759658613566398e-05, "loss": 0.2839, "step": 2700 }, { "epoch": 1.43, "learning_rate": 1.675713104323236e-05, "loss": 0.2236, "step": 2701 }, { "epoch": 1.43, "learning_rate": 1.6754602678247544e-05, "loss": 0.1822, "step": 2702 }, { "epoch": 1.43, "learning_rate": 1.6752073518909278e-05, "loss": 0.1901, "step": 2703 }, { "epoch": 1.43, "learning_rate": 1.6749543565515002e-05, "loss": 0.1942, "step": 2704 }, { "epoch": 1.43, "learning_rate": 1.6747012818362244e-05, "loss": 0.0907, "step": 2705 }, { "epoch": 1.43, "learning_rate": 1.674448127774863e-05, "loss": 0.0508, "step": 2706 }, { "epoch": 1.43, "learning_rate": 1.6741948943971858e-05, "loss": 0.1954, "step": 2707 }, { "epoch": 1.43, "learning_rate": 1.6739415817329743e-05, "loss": 0.143, "step": 2708 }, { "epoch": 1.43, "learning_rate": 1.6736881898120195e-05, "loss": 0.148, "step": 2709 }, { "epoch": 1.43, "learning_rate": 1.6734347186641194e-05, "loss": 0.17, "step": 2710 }, { "epoch": 1.43, "learning_rate": 1.6731811683190834e-05, "loss": 0.2122, "step": 2711 }, { "epoch": 1.44, "learning_rate": 1.672927538806729e-05, "loss": 0.166, "step": 2712 }, { "epoch": 1.44, "learning_rate": 1.6726738301568833e-05, "loss": 0.1359, "step": 2713 }, { "epoch": 1.44, "learning_rate": 1.6724200423993843e-05, "loss": 0.2292, "step": 2714 }, { "epoch": 1.44, "learning_rate": 1.672166175564076e-05, "loss": 0.0877, "step": 2715 }, { "epoch": 1.44, "learning_rate": 1.671912229680815e-05, "loss": 0.1715, "step": 2716 }, { "epoch": 1.44, "learning_rate": 1.671658204779465e-05, "loss": 0.1531, "step": 2717 }, { "epoch": 1.44, "learning_rate": 1.6714041008899004e-05, "loss": 0.2421, "step": 2718 }, { "epoch": 1.44, "learning_rate": 1.6711499180420035e-05, "loss": 0.1982, "step": 2719 }, { "epoch": 1.44, "learning_rate": 1.6708956562656674e-05, "loss": 0.2268, "step": 2720 }, { "epoch": 1.44, "learning_rate": 1.6706413155907936e-05, "loss": 0.1813, "step": 2721 }, { "epoch": 1.44, "learning_rate": 1.6703868960472926e-05, "loss": 0.1851, "step": 2722 }, { "epoch": 1.44, "learning_rate": 1.670132397665085e-05, "loss": 0.1479, "step": 2723 }, { "epoch": 1.44, "learning_rate": 1.6698778204741002e-05, "loss": 0.1585, "step": 2724 }, { "epoch": 1.44, "learning_rate": 1.6696231645042768e-05, "loss": 0.1382, "step": 2725 }, { "epoch": 1.44, "learning_rate": 1.669368429785563e-05, "loss": 0.207, "step": 2726 }, { "epoch": 1.44, "learning_rate": 1.6691136163479157e-05, "loss": 0.2532, "step": 2727 }, { "epoch": 1.44, "learning_rate": 1.6688587242213018e-05, "loss": 0.1598, "step": 2728 }, { "epoch": 1.44, "learning_rate": 1.6686037534356973e-05, "loss": 0.2216, "step": 2729 }, { "epoch": 1.45, "learning_rate": 1.6683487040210865e-05, "loss": 0.3255, "step": 2730 }, { "epoch": 1.45, "learning_rate": 1.6680935760074638e-05, "loss": 0.1133, "step": 2731 }, { "epoch": 1.45, "learning_rate": 1.6678383694248332e-05, "loss": 0.2096, "step": 2732 }, { "epoch": 1.45, "learning_rate": 1.667583084303207e-05, "loss": 0.2746, "step": 2733 }, { "epoch": 1.45, "learning_rate": 1.6673277206726077e-05, "loss": 0.0941, "step": 2734 }, { "epoch": 1.45, "learning_rate": 1.667072278563066e-05, "loss": 0.1336, "step": 2735 }, { "epoch": 1.45, "learning_rate": 1.6668167580046228e-05, "loss": 0.1594, "step": 2736 }, { "epoch": 1.45, "learning_rate": 1.6665611590273275e-05, "loss": 0.1538, "step": 2737 }, { "epoch": 1.45, "learning_rate": 1.666305481661239e-05, "loss": 0.2263, "step": 2738 }, { "epoch": 1.45, "learning_rate": 1.6660497259364252e-05, "loss": 0.143, "step": 2739 }, { "epoch": 1.45, "learning_rate": 1.6657938918829638e-05, "loss": 0.1937, "step": 2740 }, { "epoch": 1.45, "learning_rate": 1.6655379795309415e-05, "loss": 0.2255, "step": 2741 }, { "epoch": 1.45, "learning_rate": 1.6652819889104538e-05, "loss": 0.1851, "step": 2742 }, { "epoch": 1.45, "learning_rate": 1.6650259200516053e-05, "loss": 0.292, "step": 2743 }, { "epoch": 1.45, "learning_rate": 1.6647697729845104e-05, "loss": 0.2371, "step": 2744 }, { "epoch": 1.45, "learning_rate": 1.664513547739293e-05, "loss": 0.1599, "step": 2745 }, { "epoch": 1.45, "learning_rate": 1.6642572443460853e-05, "loss": 0.2002, "step": 2746 }, { "epoch": 1.45, "learning_rate": 1.6640008628350286e-05, "loss": 0.1554, "step": 2747 }, { "epoch": 1.45, "learning_rate": 1.6637444032362745e-05, "loss": 0.1878, "step": 2748 }, { "epoch": 1.46, "learning_rate": 1.6634878655799825e-05, "loss": 0.0857, "step": 2749 }, { "epoch": 1.46, "learning_rate": 1.6632312498963222e-05, "loss": 0.2091, "step": 2750 }, { "epoch": 1.46, "learning_rate": 1.662974556215472e-05, "loss": 0.1577, "step": 2751 }, { "epoch": 1.46, "learning_rate": 1.66271778456762e-05, "loss": 0.3094, "step": 2752 }, { "epoch": 1.46, "learning_rate": 1.6624609349829623e-05, "loss": 0.1421, "step": 2753 }, { "epoch": 1.46, "learning_rate": 1.6622040074917053e-05, "loss": 0.1365, "step": 2754 }, { "epoch": 1.46, "learning_rate": 1.661947002124064e-05, "loss": 0.1768, "step": 2755 }, { "epoch": 1.46, "learning_rate": 1.6616899189102627e-05, "loss": 0.1725, "step": 2756 }, { "epoch": 1.46, "learning_rate": 1.6614327578805346e-05, "loss": 0.0974, "step": 2757 }, { "epoch": 1.46, "learning_rate": 1.6611755190651232e-05, "loss": 0.1733, "step": 2758 }, { "epoch": 1.46, "learning_rate": 1.660918202494279e-05, "loss": 0.1162, "step": 2759 }, { "epoch": 1.46, "learning_rate": 1.6606608081982635e-05, "loss": 0.1897, "step": 2760 }, { "epoch": 1.46, "learning_rate": 1.660403336207347e-05, "loss": 0.146, "step": 2761 }, { "epoch": 1.46, "learning_rate": 1.6601457865518082e-05, "loss": 0.2005, "step": 2762 }, { "epoch": 1.46, "learning_rate": 1.6598881592619357e-05, "loss": 0.1787, "step": 2763 }, { "epoch": 1.46, "learning_rate": 1.6596304543680267e-05, "loss": 0.2806, "step": 2764 }, { "epoch": 1.46, "learning_rate": 1.659372671900388e-05, "loss": 0.2405, "step": 2765 }, { "epoch": 1.46, "learning_rate": 1.6591148118893355e-05, "loss": 0.0931, "step": 2766 }, { "epoch": 1.46, "learning_rate": 1.6588568743651934e-05, "loss": 0.1406, "step": 2767 }, { "epoch": 1.47, "learning_rate": 1.6585988593582957e-05, "loss": 0.1893, "step": 2768 }, { "epoch": 1.47, "learning_rate": 1.6583407668989858e-05, "loss": 0.1664, "step": 2769 }, { "epoch": 1.47, "learning_rate": 1.6580825970176156e-05, "loss": 0.1061, "step": 2770 }, { "epoch": 1.47, "learning_rate": 1.6578243497445466e-05, "loss": 0.1716, "step": 2771 }, { "epoch": 1.47, "learning_rate": 1.657566025110149e-05, "loss": 0.1546, "step": 2772 }, { "epoch": 1.47, "learning_rate": 1.657307623144802e-05, "loss": 0.1535, "step": 2773 }, { "epoch": 1.47, "learning_rate": 1.6570491438788944e-05, "loss": 0.1754, "step": 2774 }, { "epoch": 1.47, "learning_rate": 1.6567905873428237e-05, "loss": 0.2061, "step": 2775 }, { "epoch": 1.47, "learning_rate": 1.656531953566997e-05, "loss": 0.2575, "step": 2776 }, { "epoch": 1.47, "learning_rate": 1.656273242581829e-05, "loss": 0.1741, "step": 2777 }, { "epoch": 1.47, "learning_rate": 1.656014454417746e-05, "loss": 0.196, "step": 2778 }, { "epoch": 1.47, "learning_rate": 1.6557555891051812e-05, "loss": 0.1672, "step": 2779 }, { "epoch": 1.47, "learning_rate": 1.6554966466745773e-05, "loss": 0.2001, "step": 2780 }, { "epoch": 1.47, "learning_rate": 1.6552376271563874e-05, "loss": 0.2305, "step": 2781 }, { "epoch": 1.47, "learning_rate": 1.6549785305810723e-05, "loss": 0.0702, "step": 2782 }, { "epoch": 1.47, "learning_rate": 1.6547193569791014e-05, "loss": 0.0491, "step": 2783 }, { "epoch": 1.47, "learning_rate": 1.6544601063809547e-05, "loss": 0.2174, "step": 2784 }, { "epoch": 1.47, "learning_rate": 1.6542007788171206e-05, "loss": 0.166, "step": 2785 }, { "epoch": 1.47, "learning_rate": 1.6539413743180963e-05, "loss": 0.2125, "step": 2786 }, { "epoch": 1.48, "learning_rate": 1.653681892914388e-05, "loss": 0.1978, "step": 2787 }, { "epoch": 1.48, "learning_rate": 1.653422334636512e-05, "loss": 0.1796, "step": 2788 }, { "epoch": 1.48, "learning_rate": 1.6531626995149916e-05, "loss": 0.2519, "step": 2789 }, { "epoch": 1.48, "learning_rate": 1.6529029875803614e-05, "loss": 0.246, "step": 2790 }, { "epoch": 1.48, "learning_rate": 1.6526431988631638e-05, "loss": 0.1797, "step": 2791 }, { "epoch": 1.48, "learning_rate": 1.65238333339395e-05, "loss": 0.2405, "step": 2792 }, { "epoch": 1.48, "learning_rate": 1.6521233912032806e-05, "loss": 0.1415, "step": 2793 }, { "epoch": 1.48, "learning_rate": 1.6518633723217257e-05, "loss": 0.1804, "step": 2794 }, { "epoch": 1.48, "learning_rate": 1.6516032767798643e-05, "loss": 0.1608, "step": 2795 }, { "epoch": 1.48, "learning_rate": 1.651343104608283e-05, "loss": 0.1081, "step": 2796 }, { "epoch": 1.48, "learning_rate": 1.6510828558375795e-05, "loss": 0.1904, "step": 2797 }, { "epoch": 1.48, "learning_rate": 1.650822530498359e-05, "loss": 0.1834, "step": 2798 }, { "epoch": 1.48, "learning_rate": 1.650562128621237e-05, "loss": 0.1519, "step": 2799 }, { "epoch": 1.48, "learning_rate": 1.6503016502368358e-05, "loss": 0.2008, "step": 2800 }, { "epoch": 1.48, "learning_rate": 1.6500410953757895e-05, "loss": 0.1692, "step": 2801 }, { "epoch": 1.48, "learning_rate": 1.649780464068739e-05, "loss": 0.164, "step": 2802 }, { "epoch": 1.48, "learning_rate": 1.6495197563463358e-05, "loss": 0.1713, "step": 2803 }, { "epoch": 1.48, "learning_rate": 1.6492589722392386e-05, "loss": 0.2987, "step": 2804 }, { "epoch": 1.48, "learning_rate": 1.6489981117781166e-05, "loss": 0.2343, "step": 2805 }, { "epoch": 1.49, "learning_rate": 1.6487371749936474e-05, "loss": 0.2405, "step": 2806 }, { "epoch": 1.49, "learning_rate": 1.6484761619165182e-05, "loss": 0.198, "step": 2807 }, { "epoch": 1.49, "learning_rate": 1.6482150725774234e-05, "loss": 0.1937, "step": 2808 }, { "epoch": 1.49, "learning_rate": 1.6479539070070686e-05, "loss": 0.1401, "step": 2809 }, { "epoch": 1.49, "learning_rate": 1.6476926652361668e-05, "loss": 0.1749, "step": 2810 }, { "epoch": 1.49, "learning_rate": 1.647431347295441e-05, "loss": 0.1802, "step": 2811 }, { "epoch": 1.49, "learning_rate": 1.6471699532156226e-05, "loss": 0.1884, "step": 2812 }, { "epoch": 1.49, "learning_rate": 1.6469084830274513e-05, "loss": 0.1152, "step": 2813 }, { "epoch": 1.49, "learning_rate": 1.6466469367616772e-05, "loss": 0.1714, "step": 2814 }, { "epoch": 1.49, "learning_rate": 1.646385314449058e-05, "loss": 0.1882, "step": 2815 }, { "epoch": 1.49, "learning_rate": 1.6461236161203617e-05, "loss": 0.2279, "step": 2816 }, { "epoch": 1.49, "learning_rate": 1.645861841806364e-05, "loss": 0.1815, "step": 2817 }, { "epoch": 1.49, "learning_rate": 1.64559999153785e-05, "loss": 0.1569, "step": 2818 }, { "epoch": 1.49, "learning_rate": 1.645338065345614e-05, "loss": 0.2352, "step": 2819 }, { "epoch": 1.49, "learning_rate": 1.6450760632604587e-05, "loss": 0.238, "step": 2820 }, { "epoch": 1.49, "learning_rate": 1.6448139853131967e-05, "loss": 0.1497, "step": 2821 }, { "epoch": 1.49, "learning_rate": 1.644551831534648e-05, "loss": 0.1751, "step": 2822 }, { "epoch": 1.49, "learning_rate": 1.644289601955643e-05, "loss": 0.1256, "step": 2823 }, { "epoch": 1.49, "learning_rate": 1.6440272966070193e-05, "loss": 0.2051, "step": 2824 }, { "epoch": 1.5, "learning_rate": 1.643764915519626e-05, "loss": 0.2003, "step": 2825 }, { "epoch": 1.5, "learning_rate": 1.6435024587243185e-05, "loss": 0.2263, "step": 2826 }, { "epoch": 1.5, "learning_rate": 1.643239926251963e-05, "loss": 0.2146, "step": 2827 }, { "epoch": 1.5, "learning_rate": 1.6429773181334327e-05, "loss": 0.1235, "step": 2828 }, { "epoch": 1.5, "learning_rate": 1.642714634399612e-05, "loss": 0.2421, "step": 2829 }, { "epoch": 1.5, "learning_rate": 1.642451875081392e-05, "loss": 0.2093, "step": 2830 }, { "epoch": 1.5, "learning_rate": 1.6421890402096744e-05, "loss": 0.1914, "step": 2831 }, { "epoch": 1.5, "learning_rate": 1.6419261298153684e-05, "loss": 0.2239, "step": 2832 }, { "epoch": 1.5, "learning_rate": 1.6416631439293932e-05, "loss": 0.1157, "step": 2833 }, { "epoch": 1.5, "learning_rate": 1.641400082582676e-05, "loss": 0.1944, "step": 2834 }, { "epoch": 1.5, "learning_rate": 1.6411369458061542e-05, "loss": 0.1455, "step": 2835 }, { "epoch": 1.5, "learning_rate": 1.640873733630772e-05, "loss": 0.2451, "step": 2836 }, { "epoch": 1.5, "learning_rate": 1.6406104460874844e-05, "loss": 0.1614, "step": 2837 }, { "epoch": 1.5, "learning_rate": 1.640347083207254e-05, "loss": 0.2613, "step": 2838 }, { "epoch": 1.5, "learning_rate": 1.6400836450210533e-05, "loss": 0.1099, "step": 2839 }, { "epoch": 1.5, "learning_rate": 1.6398201315598626e-05, "loss": 0.1964, "step": 2840 }, { "epoch": 1.5, "learning_rate": 1.6395565428546715e-05, "loss": 0.173, "step": 2841 }, { "epoch": 1.5, "learning_rate": 1.639292878936479e-05, "loss": 0.1651, "step": 2842 }, { "epoch": 1.5, "learning_rate": 1.6390291398362922e-05, "loss": 0.1853, "step": 2843 }, { "epoch": 1.51, "learning_rate": 1.6387653255851276e-05, "loss": 0.1033, "step": 2844 }, { "epoch": 1.51, "learning_rate": 1.6385014362140095e-05, "loss": 0.1467, "step": 2845 }, { "epoch": 1.51, "learning_rate": 1.6382374717539722e-05, "loss": 0.3215, "step": 2846 }, { "epoch": 1.51, "learning_rate": 1.637973432236059e-05, "loss": 0.2242, "step": 2847 }, { "epoch": 1.51, "learning_rate": 1.63770931769132e-05, "loss": 0.146, "step": 2848 }, { "epoch": 1.51, "learning_rate": 1.6374451281508164e-05, "loss": 0.1419, "step": 2849 }, { "epoch": 1.51, "learning_rate": 1.6371808636456176e-05, "loss": 0.1418, "step": 2850 }, { "epoch": 1.51, "learning_rate": 1.6369165242068017e-05, "loss": 0.1615, "step": 2851 }, { "epoch": 1.51, "learning_rate": 1.6366521098654548e-05, "loss": 0.2147, "step": 2852 }, { "epoch": 1.51, "learning_rate": 1.6363876206526726e-05, "loss": 0.0648, "step": 2853 }, { "epoch": 1.51, "learning_rate": 1.6361230565995598e-05, "loss": 0.1182, "step": 2854 }, { "epoch": 1.51, "learning_rate": 1.6358584177372296e-05, "loss": 0.1278, "step": 2855 }, { "epoch": 1.51, "learning_rate": 1.635593704096804e-05, "loss": 0.2683, "step": 2856 }, { "epoch": 1.51, "learning_rate": 1.6353289157094137e-05, "loss": 0.1994, "step": 2857 }, { "epoch": 1.51, "learning_rate": 1.6350640526061984e-05, "loss": 0.1997, "step": 2858 }, { "epoch": 1.51, "learning_rate": 1.6347991148183063e-05, "loss": 0.1953, "step": 2859 }, { "epoch": 1.51, "learning_rate": 1.6345341023768947e-05, "loss": 0.1454, "step": 2860 }, { "epoch": 1.51, "learning_rate": 1.6342690153131293e-05, "loss": 0.1516, "step": 2861 }, { "epoch": 1.51, "learning_rate": 1.634003853658185e-05, "loss": 0.1135, "step": 2862 }, { "epoch": 1.52, "learning_rate": 1.6337386174432458e-05, "loss": 0.1702, "step": 2863 }, { "epoch": 1.52, "learning_rate": 1.633473306699503e-05, "loss": 0.3628, "step": 2864 }, { "epoch": 1.52, "learning_rate": 1.633207921458158e-05, "loss": 0.1521, "step": 2865 }, { "epoch": 1.52, "learning_rate": 1.632942461750421e-05, "loss": 0.2079, "step": 2866 }, { "epoch": 1.52, "learning_rate": 1.63267692760751e-05, "loss": 0.1519, "step": 2867 }, { "epoch": 1.52, "learning_rate": 1.6324113190606527e-05, "loss": 0.2237, "step": 2868 }, { "epoch": 1.52, "learning_rate": 1.6321456361410848e-05, "loss": 0.105, "step": 2869 }, { "epoch": 1.52, "learning_rate": 1.631879878880051e-05, "loss": 0.2393, "step": 2870 }, { "epoch": 1.52, "learning_rate": 1.6316140473088058e-05, "loss": 0.2038, "step": 2871 }, { "epoch": 1.52, "learning_rate": 1.6313481414586104e-05, "loss": 0.2376, "step": 2872 }, { "epoch": 1.52, "learning_rate": 1.6310821613607357e-05, "loss": 0.1628, "step": 2873 }, { "epoch": 1.52, "learning_rate": 1.6308161070464624e-05, "loss": 0.16, "step": 2874 }, { "epoch": 1.52, "learning_rate": 1.6305499785470785e-05, "loss": 0.1775, "step": 2875 }, { "epoch": 1.52, "learning_rate": 1.6302837758938807e-05, "loss": 0.1879, "step": 2876 }, { "epoch": 1.52, "learning_rate": 1.6300174991181755e-05, "loss": 0.2648, "step": 2877 }, { "epoch": 1.52, "learning_rate": 1.6297511482512774e-05, "loss": 0.1505, "step": 2878 }, { "epoch": 1.52, "learning_rate": 1.6294847233245098e-05, "loss": 0.2695, "step": 2879 }, { "epoch": 1.52, "learning_rate": 1.629218224369205e-05, "loss": 0.2859, "step": 2880 }, { "epoch": 1.52, "learning_rate": 1.6289516514167028e-05, "loss": 0.1393, "step": 2881 }, { "epoch": 1.53, "learning_rate": 1.6286850044983536e-05, "loss": 0.1557, "step": 2882 }, { "epoch": 1.53, "learning_rate": 1.6284182836455152e-05, "loss": 0.1334, "step": 2883 }, { "epoch": 1.53, "learning_rate": 1.6281514888895546e-05, "loss": 0.1852, "step": 2884 }, { "epoch": 1.53, "learning_rate": 1.6278846202618473e-05, "loss": 0.2156, "step": 2885 }, { "epoch": 1.53, "learning_rate": 1.6276176777937772e-05, "loss": 0.1962, "step": 2886 }, { "epoch": 1.53, "learning_rate": 1.6273506615167377e-05, "loss": 0.129, "step": 2887 }, { "epoch": 1.53, "learning_rate": 1.6270835714621304e-05, "loss": 0.1825, "step": 2888 }, { "epoch": 1.53, "learning_rate": 1.6268164076613652e-05, "loss": 0.0978, "step": 2889 }, { "epoch": 1.53, "learning_rate": 1.626549170145861e-05, "loss": 0.2062, "step": 2890 }, { "epoch": 1.53, "learning_rate": 1.6262818589470457e-05, "loss": 0.2156, "step": 2891 }, { "epoch": 1.53, "learning_rate": 1.6260144740963556e-05, "loss": 0.1082, "step": 2892 }, { "epoch": 1.53, "learning_rate": 1.6257470156252353e-05, "loss": 0.182, "step": 2893 }, { "epoch": 1.53, "learning_rate": 1.625479483565139e-05, "loss": 0.2353, "step": 2894 }, { "epoch": 1.53, "learning_rate": 1.6252118779475282e-05, "loss": 0.2062, "step": 2895 }, { "epoch": 1.53, "learning_rate": 1.6249441988038745e-05, "loss": 0.1562, "step": 2896 }, { "epoch": 1.53, "learning_rate": 1.624676446165657e-05, "loss": 0.1255, "step": 2897 }, { "epoch": 1.53, "learning_rate": 1.624408620064364e-05, "loss": 0.1566, "step": 2898 }, { "epoch": 1.53, "learning_rate": 1.6241407205314924e-05, "loss": 0.165, "step": 2899 }, { "epoch": 1.54, "learning_rate": 1.6238727475985477e-05, "loss": 0.2416, "step": 2900 }, { "epoch": 1.54, "learning_rate": 1.623604701297044e-05, "loss": 0.1423, "step": 2901 }, { "epoch": 1.54, "learning_rate": 1.6233365816585037e-05, "loss": 0.3346, "step": 2902 }, { "epoch": 1.54, "learning_rate": 1.623068388714459e-05, "loss": 0.1281, "step": 2903 }, { "epoch": 1.54, "learning_rate": 1.6228001224964484e-05, "loss": 0.0879, "step": 2904 }, { "epoch": 1.54, "learning_rate": 1.622531783036022e-05, "loss": 0.1838, "step": 2905 }, { "epoch": 1.54, "learning_rate": 1.6222633703647362e-05, "loss": 0.1704, "step": 2906 }, { "epoch": 1.54, "learning_rate": 1.6219948845141572e-05, "loss": 0.2215, "step": 2907 }, { "epoch": 1.54, "learning_rate": 1.621726325515859e-05, "loss": 0.2161, "step": 2908 }, { "epoch": 1.54, "learning_rate": 1.621457693401425e-05, "loss": 0.1249, "step": 2909 }, { "epoch": 1.54, "learning_rate": 1.6211889882024464e-05, "loss": 0.1751, "step": 2910 }, { "epoch": 1.54, "learning_rate": 1.6209202099505236e-05, "loss": 0.1705, "step": 2911 }, { "epoch": 1.54, "learning_rate": 1.620651358677266e-05, "loss": 0.1474, "step": 2912 }, { "epoch": 1.54, "learning_rate": 1.6203824344142898e-05, "loss": 0.1819, "step": 2913 }, { "epoch": 1.54, "learning_rate": 1.620113437193222e-05, "loss": 0.3657, "step": 2914 }, { "epoch": 1.54, "learning_rate": 1.6198443670456968e-05, "loss": 0.1757, "step": 2915 }, { "epoch": 1.54, "learning_rate": 1.6195752240033576e-05, "loss": 0.1829, "step": 2916 }, { "epoch": 1.54, "learning_rate": 1.6193060080978552e-05, "loss": 0.2077, "step": 2917 }, { "epoch": 1.54, "learning_rate": 1.619036719360851e-05, "loss": 0.1495, "step": 2918 }, { "epoch": 1.55, "learning_rate": 1.6187673578240134e-05, "loss": 0.199, "step": 2919 }, { "epoch": 1.55, "learning_rate": 1.6184979235190198e-05, "loss": 0.1922, "step": 2920 }, { "epoch": 1.55, "learning_rate": 1.6182284164775558e-05, "loss": 0.2029, "step": 2921 }, { "epoch": 1.55, "learning_rate": 1.6179588367313165e-05, "loss": 0.2268, "step": 2922 }, { "epoch": 1.55, "learning_rate": 1.6176891843120044e-05, "loss": 0.1154, "step": 2923 }, { "epoch": 1.55, "learning_rate": 1.6174194592513318e-05, "loss": 0.152, "step": 2924 }, { "epoch": 1.55, "learning_rate": 1.6171496615810183e-05, "loss": 0.2345, "step": 2925 }, { "epoch": 1.55, "learning_rate": 1.6168797913327928e-05, "loss": 0.1093, "step": 2926 }, { "epoch": 1.55, "learning_rate": 1.616609848538392e-05, "loss": 0.1968, "step": 2927 }, { "epoch": 1.55, "learning_rate": 1.6163398332295633e-05, "loss": 0.1793, "step": 2928 }, { "epoch": 1.55, "learning_rate": 1.616069745438059e-05, "loss": 0.1893, "step": 2929 }, { "epoch": 1.55, "learning_rate": 1.6157995851956428e-05, "loss": 0.1795, "step": 2930 }, { "epoch": 1.55, "learning_rate": 1.615529352534086e-05, "loss": 0.1679, "step": 2931 }, { "epoch": 1.55, "learning_rate": 1.6152590474851682e-05, "loss": 0.1207, "step": 2932 }, { "epoch": 1.55, "learning_rate": 1.6149886700806784e-05, "loss": 0.2157, "step": 2933 }, { "epoch": 1.55, "learning_rate": 1.614718220352413e-05, "loss": 0.1163, "step": 2934 }, { "epoch": 1.55, "learning_rate": 1.6144476983321774e-05, "loss": 0.168, "step": 2935 }, { "epoch": 1.55, "learning_rate": 1.6141771040517854e-05, "loss": 0.2037, "step": 2936 }, { "epoch": 1.55, "learning_rate": 1.6139064375430594e-05, "loss": 0.1842, "step": 2937 }, { "epoch": 1.56, "learning_rate": 1.6136356988378307e-05, "loss": 0.1546, "step": 2938 }, { "epoch": 1.56, "learning_rate": 1.613364887967938e-05, "loss": 0.2323, "step": 2939 }, { "epoch": 1.56, "learning_rate": 1.6130940049652297e-05, "loss": 0.1375, "step": 2940 }, { "epoch": 1.56, "learning_rate": 1.612823049861562e-05, "loss": 0.1705, "step": 2941 }, { "epoch": 1.56, "learning_rate": 1.6125520226887988e-05, "loss": 0.1641, "step": 2942 }, { "epoch": 1.56, "learning_rate": 1.6122809234788147e-05, "loss": 0.1689, "step": 2943 }, { "epoch": 1.56, "learning_rate": 1.612009752263491e-05, "loss": 0.1459, "step": 2944 }, { "epoch": 1.56, "learning_rate": 1.611738509074718e-05, "loss": 0.1261, "step": 2945 }, { "epoch": 1.56, "learning_rate": 1.6114671939443938e-05, "loss": 0.2785, "step": 2946 }, { "epoch": 1.56, "learning_rate": 1.6111958069044264e-05, "loss": 0.0779, "step": 2947 }, { "epoch": 1.56, "learning_rate": 1.6109243479867308e-05, "loss": 0.1729, "step": 2948 }, { "epoch": 1.56, "learning_rate": 1.610652817223231e-05, "loss": 0.1292, "step": 2949 }, { "epoch": 1.56, "learning_rate": 1.6103812146458603e-05, "loss": 0.2215, "step": 2950 }, { "epoch": 1.56, "learning_rate": 1.6101095402865587e-05, "loss": 0.2002, "step": 2951 }, { "epoch": 1.56, "learning_rate": 1.609837794177276e-05, "loss": 0.174, "step": 2952 }, { "epoch": 1.56, "learning_rate": 1.6095659763499703e-05, "loss": 0.1325, "step": 2953 }, { "epoch": 1.56, "learning_rate": 1.6092940868366078e-05, "loss": 0.1433, "step": 2954 }, { "epoch": 1.56, "learning_rate": 1.609022125669162e-05, "loss": 0.3045, "step": 2955 }, { "epoch": 1.56, "learning_rate": 1.6087500928796185e-05, "loss": 0.1847, "step": 2956 }, { "epoch": 1.57, "learning_rate": 1.6084779884999662e-05, "loss": 0.1392, "step": 2957 }, { "epoch": 1.57, "learning_rate": 1.608205812562207e-05, "loss": 0.1748, "step": 2958 }, { "epoch": 1.57, "learning_rate": 1.6079335650983478e-05, "loss": 0.1391, "step": 2959 }, { "epoch": 1.57, "learning_rate": 1.6076612461404065e-05, "loss": 0.2219, "step": 2960 }, { "epoch": 1.57, "learning_rate": 1.607388855720408e-05, "loss": 0.1112, "step": 2961 }, { "epoch": 1.57, "learning_rate": 1.607116393870386e-05, "loss": 0.1824, "step": 2962 }, { "epoch": 1.57, "learning_rate": 1.6068438606223816e-05, "loss": 0.2309, "step": 2963 }, { "epoch": 1.57, "learning_rate": 1.6065712560084467e-05, "loss": 0.1636, "step": 2964 }, { "epoch": 1.57, "learning_rate": 1.6062985800606384e-05, "loss": 0.1814, "step": 2965 }, { "epoch": 1.57, "learning_rate": 1.6060258328110255e-05, "loss": 0.1063, "step": 2966 }, { "epoch": 1.57, "learning_rate": 1.6057530142916825e-05, "loss": 0.2481, "step": 2967 }, { "epoch": 1.57, "learning_rate": 1.6054801245346942e-05, "loss": 0.1218, "step": 2968 }, { "epoch": 1.57, "learning_rate": 1.6052071635721523e-05, "loss": 0.2633, "step": 2969 }, { "epoch": 1.57, "learning_rate": 1.604934131436157e-05, "loss": 0.1865, "step": 2970 }, { "epoch": 1.57, "learning_rate": 1.604661028158819e-05, "loss": 0.2076, "step": 2971 }, { "epoch": 1.57, "learning_rate": 1.6043878537722543e-05, "loss": 0.1817, "step": 2972 }, { "epoch": 1.57, "learning_rate": 1.604114608308589e-05, "loss": 0.2601, "step": 2973 }, { "epoch": 1.57, "learning_rate": 1.603841291799957e-05, "loss": 0.2231, "step": 2974 }, { "epoch": 1.57, "learning_rate": 1.6035679042785022e-05, "loss": 0.0655, "step": 2975 }, { "epoch": 1.58, "learning_rate": 1.603294445776374e-05, "loss": 0.2121, "step": 2976 }, { "epoch": 1.58, "learning_rate": 1.6030209163257322e-05, "loss": 0.16, "step": 2977 }, { "epoch": 1.58, "learning_rate": 1.6027473159587443e-05, "loss": 0.1254, "step": 2978 }, { "epoch": 1.58, "learning_rate": 1.602473644707586e-05, "loss": 0.215, "step": 2979 }, { "epoch": 1.58, "learning_rate": 1.6021999026044417e-05, "loss": 0.1935, "step": 2980 }, { "epoch": 1.58, "learning_rate": 1.601926089681504e-05, "loss": 0.1708, "step": 2981 }, { "epoch": 1.58, "learning_rate": 1.6016522059709736e-05, "loss": 0.2656, "step": 2982 }, { "epoch": 1.58, "learning_rate": 1.60137825150506e-05, "loss": 0.1617, "step": 2983 }, { "epoch": 1.58, "learning_rate": 1.6011042263159806e-05, "loss": 0.1926, "step": 2984 }, { "epoch": 1.58, "learning_rate": 1.600830130435961e-05, "loss": 0.2843, "step": 2985 }, { "epoch": 1.58, "learning_rate": 1.6005559638972357e-05, "loss": 0.1574, "step": 2986 }, { "epoch": 1.58, "learning_rate": 1.6002817267320473e-05, "loss": 0.1968, "step": 2987 }, { "epoch": 1.58, "learning_rate": 1.6000074189726468e-05, "loss": 0.219, "step": 2988 }, { "epoch": 1.58, "learning_rate": 1.5997330406512923e-05, "loss": 0.139, "step": 2989 }, { "epoch": 1.58, "learning_rate": 1.5994585918002517e-05, "loss": 0.2691, "step": 2990 }, { "epoch": 1.58, "learning_rate": 1.5991840724518008e-05, "loss": 0.1262, "step": 2991 }, { "epoch": 1.58, "learning_rate": 1.5989094826382236e-05, "loss": 0.1433, "step": 2992 }, { "epoch": 1.58, "learning_rate": 1.5986348223918124e-05, "loss": 0.1383, "step": 2993 }, { "epoch": 1.58, "learning_rate": 1.5983600917448674e-05, "loss": 0.0942, "step": 2994 }, { "epoch": 1.59, "learning_rate": 1.5980852907296977e-05, "loss": 0.2143, "step": 2995 }, { "epoch": 1.59, "learning_rate": 1.5978104193786203e-05, "loss": 0.1263, "step": 2996 }, { "epoch": 1.59, "learning_rate": 1.5975354777239608e-05, "loss": 0.0716, "step": 2997 }, { "epoch": 1.59, "learning_rate": 1.5972604657980524e-05, "loss": 0.1745, "step": 2998 }, { "epoch": 1.59, "learning_rate": 1.5969853836332372e-05, "loss": 0.1531, "step": 2999 }, { "epoch": 1.59, "learning_rate": 1.5967102312618655e-05, "loss": 0.1805, "step": 3000 }, { "epoch": 1.59, "learning_rate": 1.5964350087162955e-05, "loss": 0.212, "step": 3001 }, { "epoch": 1.59, "learning_rate": 1.596159716028894e-05, "loss": 0.2704, "step": 3002 }, { "epoch": 1.59, "learning_rate": 1.5958843532320358e-05, "loss": 0.2174, "step": 3003 }, { "epoch": 1.59, "learning_rate": 1.595608920358104e-05, "loss": 0.1802, "step": 3004 }, { "epoch": 1.59, "learning_rate": 1.5953334174394904e-05, "loss": 0.2708, "step": 3005 }, { "epoch": 1.59, "learning_rate": 1.5950578445085943e-05, "loss": 0.1616, "step": 3006 }, { "epoch": 1.59, "learning_rate": 1.5947822015978233e-05, "loss": 0.2228, "step": 3007 }, { "epoch": 1.59, "learning_rate": 1.5945064887395947e-05, "loss": 0.1053, "step": 3008 }, { "epoch": 1.59, "learning_rate": 1.5942307059663314e-05, "loss": 0.1541, "step": 3009 }, { "epoch": 1.59, "learning_rate": 1.5939548533104665e-05, "loss": 0.1654, "step": 3010 }, { "epoch": 1.59, "learning_rate": 1.593678930804441e-05, "loss": 0.1514, "step": 3011 }, { "epoch": 1.59, "learning_rate": 1.5934029384807034e-05, "loss": 0.1028, "step": 3012 }, { "epoch": 1.59, "learning_rate": 1.5931268763717113e-05, "loss": 0.2239, "step": 3013 }, { "epoch": 1.6, "learning_rate": 1.59285074450993e-05, "loss": 0.2127, "step": 3014 }, { "epoch": 1.6, "learning_rate": 1.592574542927833e-05, "loss": 0.1807, "step": 3015 }, { "epoch": 1.6, "learning_rate": 1.5922982716579025e-05, "loss": 0.1967, "step": 3016 }, { "epoch": 1.6, "learning_rate": 1.592021930732628e-05, "loss": 0.224, "step": 3017 }, { "epoch": 1.6, "learning_rate": 1.5917455201845076e-05, "loss": 0.2766, "step": 3018 }, { "epoch": 1.6, "learning_rate": 1.5914690400460485e-05, "loss": 0.1825, "step": 3019 }, { "epoch": 1.6, "learning_rate": 1.5911924903497646e-05, "loss": 0.0989, "step": 3020 }, { "epoch": 1.6, "learning_rate": 1.5909158711281784e-05, "loss": 0.2392, "step": 3021 }, { "epoch": 1.6, "learning_rate": 1.5906391824138215e-05, "loss": 0.1576, "step": 3022 }, { "epoch": 1.6, "learning_rate": 1.5903624242392322e-05, "loss": 0.1733, "step": 3023 }, { "epoch": 1.6, "learning_rate": 1.590085596636959e-05, "loss": 0.1873, "step": 3024 }, { "epoch": 1.6, "learning_rate": 1.589808699639556e-05, "loss": 0.0986, "step": 3025 }, { "epoch": 1.6, "learning_rate": 1.5895317332795874e-05, "loss": 0.1095, "step": 3026 }, { "epoch": 1.6, "learning_rate": 1.5892546975896247e-05, "loss": 0.2412, "step": 3027 }, { "epoch": 1.6, "learning_rate": 1.5889775926022486e-05, "loss": 0.2713, "step": 3028 }, { "epoch": 1.6, "learning_rate": 1.5887004183500462e-05, "loss": 0.1866, "step": 3029 }, { "epoch": 1.6, "learning_rate": 1.588423174865614e-05, "loss": 0.1407, "step": 3030 }, { "epoch": 1.6, "learning_rate": 1.588145862181556e-05, "loss": 0.1675, "step": 3031 }, { "epoch": 1.6, "learning_rate": 1.5878684803304855e-05, "loss": 0.1132, "step": 3032 }, { "epoch": 1.61, "learning_rate": 1.5875910293450223e-05, "loss": 0.1194, "step": 3033 }, { "epoch": 1.61, "learning_rate": 1.5873135092577958e-05, "loss": 0.2578, "step": 3034 }, { "epoch": 1.61, "learning_rate": 1.5870359201014424e-05, "loss": 0.1422, "step": 3035 }, { "epoch": 1.61, "learning_rate": 1.586758261908607e-05, "loss": 0.2134, "step": 3036 }, { "epoch": 1.61, "learning_rate": 1.5864805347119433e-05, "loss": 0.2211, "step": 3037 }, { "epoch": 1.61, "learning_rate": 1.5862027385441118e-05, "loss": 0.1579, "step": 3038 }, { "epoch": 1.61, "learning_rate": 1.585924873437782e-05, "loss": 0.2075, "step": 3039 }, { "epoch": 1.61, "learning_rate": 1.5856469394256322e-05, "loss": 0.173, "step": 3040 }, { "epoch": 1.61, "learning_rate": 1.5853689365403468e-05, "loss": 0.2384, "step": 3041 }, { "epoch": 1.61, "learning_rate": 1.5850908648146196e-05, "loss": 0.1545, "step": 3042 }, { "epoch": 1.61, "learning_rate": 1.5848127242811534e-05, "loss": 0.1248, "step": 3043 }, { "epoch": 1.61, "learning_rate": 1.5845345149726567e-05, "loss": 0.1373, "step": 3044 }, { "epoch": 1.61, "learning_rate": 1.584256236921848e-05, "loss": 0.151, "step": 3045 }, { "epoch": 1.61, "learning_rate": 1.583977890161453e-05, "loss": 0.203, "step": 3046 }, { "epoch": 1.61, "learning_rate": 1.5836994747242066e-05, "loss": 0.1262, "step": 3047 }, { "epoch": 1.61, "learning_rate": 1.58342099064285e-05, "loss": 0.1458, "step": 3048 }, { "epoch": 1.61, "learning_rate": 1.583142437950134e-05, "loss": 0.0942, "step": 3049 }, { "epoch": 1.61, "learning_rate": 1.5828638166788167e-05, "loss": 0.2785, "step": 3050 }, { "epoch": 1.61, "learning_rate": 1.5825851268616644e-05, "loss": 0.2117, "step": 3051 }, { "epoch": 1.62, "learning_rate": 1.5823063685314517e-05, "loss": 0.2512, "step": 3052 }, { "epoch": 1.62, "learning_rate": 1.582027541720961e-05, "loss": 0.1146, "step": 3053 }, { "epoch": 1.62, "learning_rate": 1.5817486464629827e-05, "loss": 0.1459, "step": 3054 }, { "epoch": 1.62, "learning_rate": 1.5814696827903157e-05, "loss": 0.2023, "step": 3055 }, { "epoch": 1.62, "learning_rate": 1.581190650735766e-05, "loss": 0.1898, "step": 3056 }, { "epoch": 1.62, "learning_rate": 1.5809115503321494e-05, "loss": 0.2379, "step": 3057 }, { "epoch": 1.62, "learning_rate": 1.5806323816122876e-05, "loss": 0.1943, "step": 3058 }, { "epoch": 1.62, "learning_rate": 1.5803531446090115e-05, "loss": 0.1379, "step": 3059 }, { "epoch": 1.62, "learning_rate": 1.5800738393551602e-05, "loss": 0.1503, "step": 3060 }, { "epoch": 1.62, "learning_rate": 1.5797944658835806e-05, "loss": 0.1867, "step": 3061 }, { "epoch": 1.62, "learning_rate": 1.579515024227127e-05, "loss": 0.1568, "step": 3062 }, { "epoch": 1.62, "learning_rate": 1.5792355144186626e-05, "loss": 0.2702, "step": 3063 }, { "epoch": 1.62, "learning_rate": 1.5789559364910582e-05, "loss": 0.1516, "step": 3064 }, { "epoch": 1.62, "learning_rate": 1.5786762904771926e-05, "loss": 0.1887, "step": 3065 }, { "epoch": 1.62, "learning_rate": 1.5783965764099525e-05, "loss": 0.1742, "step": 3066 }, { "epoch": 1.62, "learning_rate": 1.5781167943222334e-05, "loss": 0.1239, "step": 3067 }, { "epoch": 1.62, "learning_rate": 1.5778369442469374e-05, "loss": 0.2062, "step": 3068 }, { "epoch": 1.62, "learning_rate": 1.5775570262169758e-05, "loss": 0.2061, "step": 3069 }, { "epoch": 1.62, "learning_rate": 1.5772770402652677e-05, "loss": 0.2215, "step": 3070 }, { "epoch": 1.63, "learning_rate": 1.5769969864247393e-05, "loss": 0.1368, "step": 3071 }, { "epoch": 1.63, "learning_rate": 1.5767168647283262e-05, "loss": 0.096, "step": 3072 }, { "epoch": 1.63, "learning_rate": 1.5764366752089707e-05, "loss": 0.1833, "step": 3073 }, { "epoch": 1.63, "learning_rate": 1.5761564178996232e-05, "loss": 0.2221, "step": 3074 }, { "epoch": 1.63, "learning_rate": 1.5758760928332436e-05, "loss": 0.1297, "step": 3075 }, { "epoch": 1.63, "learning_rate": 1.5755957000427974e-05, "loss": 0.1255, "step": 3076 }, { "epoch": 1.63, "learning_rate": 1.5753152395612605e-05, "loss": 0.2137, "step": 3077 }, { "epoch": 1.63, "learning_rate": 1.5750347114216148e-05, "loss": 0.2279, "step": 3078 }, { "epoch": 1.63, "learning_rate": 1.5747541156568508e-05, "loss": 0.1624, "step": 3079 }, { "epoch": 1.63, "learning_rate": 1.5744734522999675e-05, "loss": 0.1407, "step": 3080 }, { "epoch": 1.63, "learning_rate": 1.5741927213839715e-05, "loss": 0.2086, "step": 3081 }, { "epoch": 1.63, "learning_rate": 1.5739119229418767e-05, "loss": 0.1983, "step": 3082 }, { "epoch": 1.63, "learning_rate": 1.573631057006706e-05, "loss": 0.1509, "step": 3083 }, { "epoch": 1.63, "learning_rate": 1.5733501236114893e-05, "loss": 0.1333, "step": 3084 }, { "epoch": 1.63, "learning_rate": 1.5730691227892653e-05, "loss": 0.245, "step": 3085 }, { "epoch": 1.63, "learning_rate": 1.57278805457308e-05, "loss": 0.1577, "step": 3086 }, { "epoch": 1.63, "learning_rate": 1.5725069189959876e-05, "loss": 0.1213, "step": 3087 }, { "epoch": 1.63, "learning_rate": 1.5722257160910503e-05, "loss": 0.2671, "step": 3088 }, { "epoch": 1.64, "learning_rate": 1.571944445891338e-05, "loss": 0.1768, "step": 3089 }, { "epoch": 1.64, "learning_rate": 1.571663108429928e-05, "loss": 0.1726, "step": 3090 }, { "epoch": 1.64, "learning_rate": 1.5713817037399068e-05, "loss": 0.2305, "step": 3091 }, { "epoch": 1.64, "learning_rate": 1.5711002318543675e-05, "loss": 0.1776, "step": 3092 }, { "epoch": 1.64, "learning_rate": 1.5708186928064125e-05, "loss": 0.2529, "step": 3093 }, { "epoch": 1.64, "learning_rate": 1.5705370866291503e-05, "loss": 0.2303, "step": 3094 }, { "epoch": 1.64, "learning_rate": 1.5702554133556996e-05, "loss": 0.2052, "step": 3095 }, { "epoch": 1.64, "learning_rate": 1.5699736730191847e-05, "loss": 0.1045, "step": 3096 }, { "epoch": 1.64, "learning_rate": 1.569691865652739e-05, "loss": 0.1474, "step": 3097 }, { "epoch": 1.64, "learning_rate": 1.569409991289503e-05, "loss": 0.168, "step": 3098 }, { "epoch": 1.64, "learning_rate": 1.5691280499626265e-05, "loss": 0.2185, "step": 3099 }, { "epoch": 1.64, "learning_rate": 1.5688460417052665e-05, "loss": 0.2503, "step": 3100 }, { "epoch": 1.64, "learning_rate": 1.568563966550587e-05, "loss": 0.1141, "step": 3101 }, { "epoch": 1.64, "learning_rate": 1.5682818245317605e-05, "loss": 0.2762, "step": 3102 }, { "epoch": 1.64, "learning_rate": 1.5679996156819678e-05, "loss": 0.2092, "step": 3103 }, { "epoch": 1.64, "learning_rate": 1.5677173400343965e-05, "loss": 0.2975, "step": 3104 }, { "epoch": 1.64, "learning_rate": 1.5674349976222437e-05, "loss": 0.0772, "step": 3105 }, { "epoch": 1.64, "learning_rate": 1.567152588478713e-05, "loss": 0.1454, "step": 3106 }, { "epoch": 1.64, "learning_rate": 1.5668701126370157e-05, "loss": 0.1795, "step": 3107 }, { "epoch": 1.65, "learning_rate": 1.566587570130372e-05, "loss": 0.2001, "step": 3108 }, { "epoch": 1.65, "learning_rate": 1.5663049609920097e-05, "loss": 0.2306, "step": 3109 }, { "epoch": 1.65, "learning_rate": 1.5660222852551634e-05, "loss": 0.1806, "step": 3110 }, { "epoch": 1.65, "learning_rate": 1.565739542953077e-05, "loss": 0.1622, "step": 3111 }, { "epoch": 1.65, "learning_rate": 1.5654567341190005e-05, "loss": 0.2669, "step": 3112 }, { "epoch": 1.65, "learning_rate": 1.565173858786194e-05, "loss": 0.1396, "step": 3113 }, { "epoch": 1.65, "learning_rate": 1.564890916987923e-05, "loss": 0.1866, "step": 3114 }, { "epoch": 1.65, "learning_rate": 1.564607908757463e-05, "loss": 0.1298, "step": 3115 }, { "epoch": 1.65, "learning_rate": 1.5643248341280953e-05, "loss": 0.112, "step": 3116 }, { "epoch": 1.65, "learning_rate": 1.5640416931331107e-05, "loss": 0.2257, "step": 3117 }, { "epoch": 1.65, "learning_rate": 1.5637584858058067e-05, "loss": 0.1868, "step": 3118 }, { "epoch": 1.65, "learning_rate": 1.563475212179489e-05, "loss": 0.2194, "step": 3119 }, { "epoch": 1.65, "learning_rate": 1.5631918722874713e-05, "loss": 0.1905, "step": 3120 }, { "epoch": 1.65, "learning_rate": 1.5629084661630753e-05, "loss": 0.1023, "step": 3121 }, { "epoch": 1.65, "learning_rate": 1.5626249938396286e-05, "loss": 0.1859, "step": 3122 }, { "epoch": 1.65, "learning_rate": 1.5623414553504696e-05, "loss": 0.1789, "step": 3123 }, { "epoch": 1.65, "learning_rate": 1.5620578507289422e-05, "loss": 0.1715, "step": 3124 }, { "epoch": 1.65, "learning_rate": 1.561774180008399e-05, "loss": 0.144, "step": 3125 }, { "epoch": 1.65, "learning_rate": 1.5614904432222003e-05, "loss": 0.1443, "step": 3126 }, { "epoch": 1.66, "learning_rate": 1.5612066404037138e-05, "loss": 0.2536, "step": 3127 }, { "epoch": 1.66, "learning_rate": 1.5609227715863153e-05, "loss": 0.2243, "step": 3128 }, { "epoch": 1.66, "learning_rate": 1.5606388368033885e-05, "loss": 0.2344, "step": 3129 }, { "epoch": 1.66, "learning_rate": 1.5603548360883244e-05, "loss": 0.1773, "step": 3130 }, { "epoch": 1.66, "learning_rate": 1.560070769474522e-05, "loss": 0.2067, "step": 3131 }, { "epoch": 1.66, "learning_rate": 1.5597866369953876e-05, "loss": 0.1741, "step": 3132 }, { "epoch": 1.66, "learning_rate": 1.559502438684337e-05, "loss": 0.1591, "step": 3133 }, { "epoch": 1.66, "learning_rate": 1.5592181745747913e-05, "loss": 0.2195, "step": 3134 }, { "epoch": 1.66, "learning_rate": 1.5589338447001808e-05, "loss": 0.1421, "step": 3135 }, { "epoch": 1.66, "learning_rate": 1.558649449093943e-05, "loss": 0.1768, "step": 3136 }, { "epoch": 1.66, "learning_rate": 1.5583649877895238e-05, "loss": 0.2269, "step": 3137 }, { "epoch": 1.66, "learning_rate": 1.558080460820376e-05, "loss": 0.3, "step": 3138 }, { "epoch": 1.66, "learning_rate": 1.557795868219961e-05, "loss": 0.1689, "step": 3139 }, { "epoch": 1.66, "learning_rate": 1.557511210021746e-05, "loss": 0.232, "step": 3140 }, { "epoch": 1.66, "learning_rate": 1.5572264862592094e-05, "loss": 0.1483, "step": 3141 }, { "epoch": 1.66, "learning_rate": 1.5569416969658338e-05, "loss": 0.1618, "step": 3142 }, { "epoch": 1.66, "learning_rate": 1.5566568421751108e-05, "loss": 0.2214, "step": 3143 }, { "epoch": 1.66, "learning_rate": 1.5563719219205408e-05, "loss": 0.0703, "step": 3144 }, { "epoch": 1.66, "learning_rate": 1.5560869362356304e-05, "loss": 0.1535, "step": 3145 }, { "epoch": 1.67, "learning_rate": 1.5558018851538944e-05, "loss": 0.3297, "step": 3146 }, { "epoch": 1.67, "learning_rate": 1.5555167687088552e-05, "loss": 0.2386, "step": 3147 }, { "epoch": 1.67, "learning_rate": 1.5552315869340432e-05, "loss": 0.1091, "step": 3148 }, { "epoch": 1.67, "learning_rate": 1.5549463398629968e-05, "loss": 0.1381, "step": 3149 }, { "epoch": 1.67, "learning_rate": 1.5546610275292605e-05, "loss": 0.2098, "step": 3150 }, { "epoch": 1.67, "learning_rate": 1.5543756499663883e-05, "loss": 0.1267, "step": 3151 }, { "epoch": 1.67, "learning_rate": 1.5540902072079408e-05, "loss": 0.1171, "step": 3152 }, { "epoch": 1.67, "learning_rate": 1.553804699287487e-05, "loss": 0.1925, "step": 3153 }, { "epoch": 1.67, "learning_rate": 1.5535191262386025e-05, "loss": 0.1275, "step": 3154 }, { "epoch": 1.67, "learning_rate": 1.5532334880948716e-05, "loss": 0.1892, "step": 3155 }, { "epoch": 1.67, "learning_rate": 1.552947784889886e-05, "loss": 0.111, "step": 3156 }, { "epoch": 1.67, "learning_rate": 1.5526620166572447e-05, "loss": 0.2721, "step": 3157 }, { "epoch": 1.67, "learning_rate": 1.5523761834305544e-05, "loss": 0.2111, "step": 3158 }, { "epoch": 1.67, "learning_rate": 1.55209028524343e-05, "loss": 0.165, "step": 3159 }, { "epoch": 1.67, "learning_rate": 1.551804322129493e-05, "loss": 0.1286, "step": 3160 }, { "epoch": 1.67, "learning_rate": 1.5515182941223744e-05, "loss": 0.1368, "step": 3161 }, { "epoch": 1.67, "learning_rate": 1.55123220125571e-05, "loss": 0.2238, "step": 3162 }, { "epoch": 1.67, "learning_rate": 1.550946043563146e-05, "loss": 0.1979, "step": 3163 }, { "epoch": 1.67, "learning_rate": 1.550659821078335e-05, "loss": 0.2212, "step": 3164 }, { "epoch": 1.68, "learning_rate": 1.5503735338349365e-05, "loss": 0.0744, "step": 3165 }, { "epoch": 1.68, "learning_rate": 1.5500871818666193e-05, "loss": 0.1313, "step": 3166 }, { "epoch": 1.68, "learning_rate": 1.549800765207058e-05, "loss": 0.129, "step": 3167 }, { "epoch": 1.68, "learning_rate": 1.5495142838899362e-05, "loss": 0.1925, "step": 3168 }, { "epoch": 1.68, "learning_rate": 1.5492277379489454e-05, "loss": 0.2167, "step": 3169 }, { "epoch": 1.68, "learning_rate": 1.5489411274177827e-05, "loss": 0.1175, "step": 3170 }, { "epoch": 1.68, "learning_rate": 1.5486544523301544e-05, "loss": 0.2163, "step": 3171 }, { "epoch": 1.68, "learning_rate": 1.5483677127197738e-05, "loss": 0.1252, "step": 3172 }, { "epoch": 1.68, "learning_rate": 1.5480809086203624e-05, "loss": 0.1482, "step": 3173 }, { "epoch": 1.68, "learning_rate": 1.5477940400656493e-05, "loss": 0.1992, "step": 3174 }, { "epoch": 1.68, "learning_rate": 1.5475071070893694e-05, "loss": 0.214, "step": 3175 }, { "epoch": 1.68, "learning_rate": 1.547220109725268e-05, "loss": 0.2918, "step": 3176 }, { "epoch": 1.68, "learning_rate": 1.5469330480070956e-05, "loss": 0.2031, "step": 3177 }, { "epoch": 1.68, "learning_rate": 1.5466459219686112e-05, "loss": 0.1583, "step": 3178 }, { "epoch": 1.68, "learning_rate": 1.546358731643582e-05, "loss": 0.1064, "step": 3179 }, { "epoch": 1.68, "learning_rate": 1.5460714770657813e-05, "loss": 0.1853, "step": 3180 }, { "epoch": 1.68, "learning_rate": 1.545784158268991e-05, "loss": 0.1761, "step": 3181 }, { "epoch": 1.68, "learning_rate": 1.5454967752870005e-05, "loss": 0.2337, "step": 3182 }, { "epoch": 1.68, "learning_rate": 1.545209328153607e-05, "loss": 0.1912, "step": 3183 }, { "epoch": 1.69, "learning_rate": 1.5449218169026136e-05, "loss": 0.161, "step": 3184 }, { "epoch": 1.69, "learning_rate": 1.5446342415678332e-05, "loss": 0.1377, "step": 3185 }, { "epoch": 1.69, "learning_rate": 1.5443466021830844e-05, "loss": 0.1195, "step": 3186 }, { "epoch": 1.69, "learning_rate": 1.544058898782195e-05, "loss": 0.0833, "step": 3187 }, { "epoch": 1.69, "learning_rate": 1.5437711313989984e-05, "loss": 0.217, "step": 3188 }, { "epoch": 1.69, "learning_rate": 1.5434833000673378e-05, "loss": 0.1142, "step": 3189 }, { "epoch": 1.69, "learning_rate": 1.5431954048210616e-05, "loss": 0.1449, "step": 3190 }, { "epoch": 1.69, "learning_rate": 1.542907445694027e-05, "loss": 0.2002, "step": 3191 }, { "epoch": 1.69, "learning_rate": 1.542619422720099e-05, "loss": 0.1317, "step": 3192 }, { "epoch": 1.69, "learning_rate": 1.5423313359331492e-05, "loss": 0.1789, "step": 3193 }, { "epoch": 1.69, "learning_rate": 1.5420431853670576e-05, "loss": 0.19, "step": 3194 }, { "epoch": 1.69, "learning_rate": 1.54175497105571e-05, "loss": 0.2769, "step": 3195 }, { "epoch": 1.69, "learning_rate": 1.5414666930330025e-05, "loss": 0.1241, "step": 3196 }, { "epoch": 1.69, "learning_rate": 1.5411783513328364e-05, "loss": 0.1152, "step": 3197 }, { "epoch": 1.69, "learning_rate": 1.540889945989121e-05, "loss": 0.1402, "step": 3198 }, { "epoch": 1.69, "learning_rate": 1.540601477035774e-05, "loss": 0.1502, "step": 3199 }, { "epoch": 1.69, "learning_rate": 1.5403129445067185e-05, "loss": 0.1968, "step": 3200 }, { "epoch": 1.69, "learning_rate": 1.5400243484358877e-05, "loss": 0.1243, "step": 3201 }, { "epoch": 1.69, "learning_rate": 1.539735688857221e-05, "loss": 0.2804, "step": 3202 }, { "epoch": 1.7, "learning_rate": 1.5394469658046647e-05, "loss": 0.1764, "step": 3203 }, { "epoch": 1.7, "learning_rate": 1.5391581793121728e-05, "loss": 0.2627, "step": 3204 }, { "epoch": 1.7, "learning_rate": 1.538869329413708e-05, "loss": 0.1858, "step": 3205 }, { "epoch": 1.7, "learning_rate": 1.53858041614324e-05, "loss": 0.1119, "step": 3206 }, { "epoch": 1.7, "learning_rate": 1.538291439534744e-05, "loss": 0.162, "step": 3207 }, { "epoch": 1.7, "learning_rate": 1.538002399622205e-05, "loss": 0.1118, "step": 3208 }, { "epoch": 1.7, "learning_rate": 1.5377132964396145e-05, "loss": 0.2721, "step": 3209 }, { "epoch": 1.7, "learning_rate": 1.537424130020972e-05, "loss": 0.1876, "step": 3210 }, { "epoch": 1.7, "learning_rate": 1.5371349004002834e-05, "loss": 0.128, "step": 3211 }, { "epoch": 1.7, "learning_rate": 1.536845607611563e-05, "loss": 0.207, "step": 3212 }, { "epoch": 1.7, "learning_rate": 1.5365562516888317e-05, "loss": 0.1756, "step": 3213 }, { "epoch": 1.7, "learning_rate": 1.5362668326661184e-05, "loss": 0.213, "step": 3214 }, { "epoch": 1.7, "learning_rate": 1.53597735057746e-05, "loss": 0.085, "step": 3215 }, { "epoch": 1.7, "learning_rate": 1.5356878054568995e-05, "loss": 0.104, "step": 3216 }, { "epoch": 1.7, "learning_rate": 1.5353981973384876e-05, "loss": 0.1682, "step": 3217 }, { "epoch": 1.7, "learning_rate": 1.535108526256283e-05, "loss": 0.1591, "step": 3218 }, { "epoch": 1.7, "learning_rate": 1.534818792244352e-05, "loss": 0.1368, "step": 3219 }, { "epoch": 1.7, "learning_rate": 1.5345289953367678e-05, "loss": 0.1885, "step": 3220 }, { "epoch": 1.7, "learning_rate": 1.5342391355676103e-05, "loss": 0.1407, "step": 3221 }, { "epoch": 1.71, "learning_rate": 1.5339492129709685e-05, "loss": 0.2028, "step": 3222 }, { "epoch": 1.71, "learning_rate": 1.5336592275809367e-05, "loss": 0.1137, "step": 3223 }, { "epoch": 1.71, "learning_rate": 1.5333691794316185e-05, "loss": 0.1627, "step": 3224 }, { "epoch": 1.71, "learning_rate": 1.533079068557124e-05, "loss": 0.1374, "step": 3225 }, { "epoch": 1.71, "learning_rate": 1.5327888949915707e-05, "loss": 0.2151, "step": 3226 }, { "epoch": 1.71, "learning_rate": 1.5324986587690837e-05, "loss": 0.1682, "step": 3227 }, { "epoch": 1.71, "learning_rate": 1.5322083599237948e-05, "loss": 0.1961, "step": 3228 }, { "epoch": 1.71, "learning_rate": 1.531917998489844e-05, "loss": 0.2376, "step": 3229 }, { "epoch": 1.71, "learning_rate": 1.531627574501378e-05, "loss": 0.148, "step": 3230 }, { "epoch": 1.71, "learning_rate": 1.5313370879925516e-05, "loss": 0.0987, "step": 3231 }, { "epoch": 1.71, "learning_rate": 1.5310465389975266e-05, "loss": 0.2552, "step": 3232 }, { "epoch": 1.71, "learning_rate": 1.530755927550472e-05, "loss": 0.218, "step": 3233 }, { "epoch": 1.71, "learning_rate": 1.530465253685564e-05, "loss": 0.2007, "step": 3234 }, { "epoch": 1.71, "learning_rate": 1.5301745174369863e-05, "loss": 0.2016, "step": 3235 }, { "epoch": 1.71, "learning_rate": 1.5298837188389307e-05, "loss": 0.1768, "step": 3236 }, { "epoch": 1.71, "learning_rate": 1.5295928579255946e-05, "loss": 0.1568, "step": 3237 }, { "epoch": 1.71, "learning_rate": 1.529301934731185e-05, "loss": 0.1622, "step": 3238 }, { "epoch": 1.71, "learning_rate": 1.5290109492899137e-05, "loss": 0.1738, "step": 3239 }, { "epoch": 1.71, "learning_rate": 1.5287199016360017e-05, "loss": 0.2326, "step": 3240 }, { "epoch": 1.72, "learning_rate": 1.5284287918036768e-05, "loss": 0.2087, "step": 3241 }, { "epoch": 1.72, "learning_rate": 1.528137619827174e-05, "loss": 0.2227, "step": 3242 }, { "epoch": 1.72, "learning_rate": 1.527846385740736e-05, "loss": 0.1992, "step": 3243 }, { "epoch": 1.72, "learning_rate": 1.5275550895786117e-05, "loss": 0.1458, "step": 3244 }, { "epoch": 1.72, "learning_rate": 1.5272637313750583e-05, "loss": 0.2237, "step": 3245 }, { "epoch": 1.72, "learning_rate": 1.5269723111643405e-05, "loss": 0.2969, "step": 3246 }, { "epoch": 1.72, "learning_rate": 1.5266808289807294e-05, "loss": 0.1991, "step": 3247 }, { "epoch": 1.72, "learning_rate": 1.5263892848585043e-05, "loss": 0.1984, "step": 3248 }, { "epoch": 1.72, "learning_rate": 1.5260976788319506e-05, "loss": 0.2569, "step": 3249 }, { "epoch": 1.72, "learning_rate": 1.5258060109353622e-05, "loss": 0.2651, "step": 3250 }, { "epoch": 1.72, "learning_rate": 1.5255142812030394e-05, "loss": 0.1549, "step": 3251 }, { "epoch": 1.72, "learning_rate": 1.5252224896692907e-05, "loss": 0.1102, "step": 3252 }, { "epoch": 1.72, "learning_rate": 1.5249306363684309e-05, "loss": 0.1132, "step": 3253 }, { "epoch": 1.72, "learning_rate": 1.5246387213347826e-05, "loss": 0.19, "step": 3254 }, { "epoch": 1.72, "learning_rate": 1.5243467446026754e-05, "loss": 0.15, "step": 3255 }, { "epoch": 1.72, "learning_rate": 1.5240547062064463e-05, "loss": 0.2115, "step": 3256 }, { "epoch": 1.72, "learning_rate": 1.5237626061804398e-05, "loss": 0.1838, "step": 3257 }, { "epoch": 1.72, "learning_rate": 1.5234704445590072e-05, "loss": 0.2035, "step": 3258 }, { "epoch": 1.73, "learning_rate": 1.5231782213765073e-05, "loss": 0.1315, "step": 3259 }, { "epoch": 1.73, "learning_rate": 1.5228859366673059e-05, "loss": 0.1245, "step": 3260 }, { "epoch": 1.73, "learning_rate": 1.5225935904657761e-05, "loss": 0.208, "step": 3261 }, { "epoch": 1.73, "learning_rate": 1.522301182806299e-05, "loss": 0.2106, "step": 3262 }, { "epoch": 1.73, "learning_rate": 1.5220087137232617e-05, "loss": 0.1584, "step": 3263 }, { "epoch": 1.73, "learning_rate": 1.5217161832510592e-05, "loss": 0.1551, "step": 3264 }, { "epoch": 1.73, "learning_rate": 1.5214235914240937e-05, "loss": 0.1489, "step": 3265 }, { "epoch": 1.73, "learning_rate": 1.5211309382767745e-05, "loss": 0.1515, "step": 3266 }, { "epoch": 1.73, "learning_rate": 1.520838223843518e-05, "loss": 0.194, "step": 3267 }, { "epoch": 1.73, "learning_rate": 1.5205454481587478e-05, "loss": 0.1705, "step": 3268 }, { "epoch": 1.73, "learning_rate": 1.5202526112568956e-05, "loss": 0.2434, "step": 3269 }, { "epoch": 1.73, "learning_rate": 1.519959713172399e-05, "loss": 0.1577, "step": 3270 }, { "epoch": 1.73, "learning_rate": 1.519666753939703e-05, "loss": 0.2936, "step": 3271 }, { "epoch": 1.73, "learning_rate": 1.5193737335932607e-05, "loss": 0.0874, "step": 3272 }, { "epoch": 1.73, "learning_rate": 1.5190806521675317e-05, "loss": 0.2949, "step": 3273 }, { "epoch": 1.73, "learning_rate": 1.518787509696983e-05, "loss": 0.1728, "step": 3274 }, { "epoch": 1.73, "learning_rate": 1.5184943062160882e-05, "loss": 0.204, "step": 3275 }, { "epoch": 1.73, "learning_rate": 1.518201041759329e-05, "loss": 0.1731, "step": 3276 }, { "epoch": 1.73, "learning_rate": 1.5179077163611938e-05, "loss": 0.225, "step": 3277 }, { "epoch": 1.74, "learning_rate": 1.517614330056178e-05, "loss": 0.1731, "step": 3278 }, { "epoch": 1.74, "learning_rate": 1.5173208828787847e-05, "loss": 0.1181, "step": 3279 }, { "epoch": 1.74, "learning_rate": 1.5170273748635231e-05, "loss": 0.171, "step": 3280 }, { "epoch": 1.74, "learning_rate": 1.5167338060449113e-05, "loss": 0.1068, "step": 3281 }, { "epoch": 1.74, "learning_rate": 1.5164401764574725e-05, "loss": 0.1193, "step": 3282 }, { "epoch": 1.74, "learning_rate": 1.516146486135739e-05, "loss": 0.1719, "step": 3283 }, { "epoch": 1.74, "learning_rate": 1.5158527351142485e-05, "loss": 0.2107, "step": 3284 }, { "epoch": 1.74, "learning_rate": 1.5155589234275472e-05, "loss": 0.1514, "step": 3285 }, { "epoch": 1.74, "learning_rate": 1.5152650511101876e-05, "loss": 0.1565, "step": 3286 }, { "epoch": 1.74, "learning_rate": 1.5149711181967297e-05, "loss": 0.2125, "step": 3287 }, { "epoch": 1.74, "learning_rate": 1.5146771247217404e-05, "loss": 0.2498, "step": 3288 }, { "epoch": 1.74, "learning_rate": 1.5143830707197944e-05, "loss": 0.1434, "step": 3289 }, { "epoch": 1.74, "learning_rate": 1.5140889562254725e-05, "loss": 0.1682, "step": 3290 }, { "epoch": 1.74, "learning_rate": 1.513794781273363e-05, "loss": 0.1783, "step": 3291 }, { "epoch": 1.74, "learning_rate": 1.5135005458980617e-05, "loss": 0.1878, "step": 3292 }, { "epoch": 1.74, "learning_rate": 1.5132062501341708e-05, "loss": 0.1449, "step": 3293 }, { "epoch": 1.74, "learning_rate": 1.5129118940163008e-05, "loss": 0.19, "step": 3294 }, { "epoch": 1.74, "learning_rate": 1.5126174775790676e-05, "loss": 0.1585, "step": 3295 }, { "epoch": 1.74, "learning_rate": 1.5123230008570957e-05, "loss": 0.1972, "step": 3296 }, { "epoch": 1.75, "learning_rate": 1.512028463885016e-05, "loss": 0.2246, "step": 3297 }, { "epoch": 1.75, "learning_rate": 1.5117338666974662e-05, "loss": 0.1621, "step": 3298 }, { "epoch": 1.75, "learning_rate": 1.5114392093290922e-05, "loss": 0.1246, "step": 3299 }, { "epoch": 1.75, "learning_rate": 1.5111444918145455e-05, "loss": 0.1741, "step": 3300 }, { "epoch": 1.75, "learning_rate": 1.5108497141884857e-05, "loss": 0.1677, "step": 3301 }, { "epoch": 1.75, "learning_rate": 1.510554876485579e-05, "loss": 0.0647, "step": 3302 }, { "epoch": 1.75, "learning_rate": 1.5102599787404995e-05, "loss": 0.1366, "step": 3303 }, { "epoch": 1.75, "learning_rate": 1.5099650209879269e-05, "loss": 0.2991, "step": 3304 }, { "epoch": 1.75, "learning_rate": 1.509670003262549e-05, "loss": 0.2172, "step": 3305 }, { "epoch": 1.75, "learning_rate": 1.5093749255990609e-05, "loss": 0.1192, "step": 3306 }, { "epoch": 1.75, "learning_rate": 1.5090797880321635e-05, "loss": 0.1098, "step": 3307 }, { "epoch": 1.75, "learning_rate": 1.5087845905965659e-05, "loss": 0.2432, "step": 3308 }, { "epoch": 1.75, "learning_rate": 1.508489333326984e-05, "loss": 0.1454, "step": 3309 }, { "epoch": 1.75, "learning_rate": 1.5081940162581403e-05, "loss": 0.1729, "step": 3310 }, { "epoch": 1.75, "learning_rate": 1.5078986394247652e-05, "loss": 0.1645, "step": 3311 }, { "epoch": 1.75, "learning_rate": 1.5076032028615948e-05, "loss": 0.1982, "step": 3312 }, { "epoch": 1.75, "learning_rate": 1.5073077066033728e-05, "loss": 0.1837, "step": 3313 }, { "epoch": 1.75, "learning_rate": 1.5070121506848509e-05, "loss": 0.1676, "step": 3314 }, { "epoch": 1.75, "learning_rate": 1.5067165351407869e-05, "loss": 0.2655, "step": 3315 }, { "epoch": 1.76, "learning_rate": 1.5064208600059454e-05, "loss": 0.18, "step": 3316 }, { "epoch": 1.76, "learning_rate": 1.5061251253150982e-05, "loss": 0.1979, "step": 3317 }, { "epoch": 1.76, "learning_rate": 1.5058293311030246e-05, "loss": 0.2911, "step": 3318 }, { "epoch": 1.76, "learning_rate": 1.50553347740451e-05, "loss": 0.1824, "step": 3319 }, { "epoch": 1.76, "learning_rate": 1.5052375642543481e-05, "loss": 0.1074, "step": 3320 }, { "epoch": 1.76, "learning_rate": 1.5049415916873385e-05, "loss": 0.1291, "step": 3321 }, { "epoch": 1.76, "learning_rate": 1.5046455597382879e-05, "loss": 0.088, "step": 3322 }, { "epoch": 1.76, "learning_rate": 1.5043494684420102e-05, "loss": 0.1331, "step": 3323 }, { "epoch": 1.76, "learning_rate": 1.5040533178333263e-05, "loss": 0.1905, "step": 3324 }, { "epoch": 1.76, "learning_rate": 1.5037571079470646e-05, "loss": 0.2128, "step": 3325 }, { "epoch": 1.76, "learning_rate": 1.5034608388180592e-05, "loss": 0.19, "step": 3326 }, { "epoch": 1.76, "learning_rate": 1.5031645104811521e-05, "loss": 0.1991, "step": 3327 }, { "epoch": 1.76, "learning_rate": 1.5028681229711919e-05, "loss": 0.1422, "step": 3328 }, { "epoch": 1.76, "learning_rate": 1.5025716763230346e-05, "loss": 0.1581, "step": 3329 }, { "epoch": 1.76, "learning_rate": 1.5022751705715428e-05, "loss": 0.1381, "step": 3330 }, { "epoch": 1.76, "learning_rate": 1.5019786057515861e-05, "loss": 0.1066, "step": 3331 }, { "epoch": 1.76, "learning_rate": 1.5016819818980412e-05, "loss": 0.2444, "step": 3332 }, { "epoch": 1.76, "learning_rate": 1.5013852990457911e-05, "loss": 0.1496, "step": 3333 }, { "epoch": 1.76, "learning_rate": 1.5010885572297265e-05, "loss": 0.2993, "step": 3334 }, { "epoch": 1.77, "learning_rate": 1.5007917564847452e-05, "loss": 0.1552, "step": 3335 }, { "epoch": 1.77, "learning_rate": 1.500494896845751e-05, "loss": 0.1725, "step": 3336 }, { "epoch": 1.77, "learning_rate": 1.500197978347655e-05, "loss": 0.1237, "step": 3337 }, { "epoch": 1.77, "learning_rate": 1.4999010010253758e-05, "loss": 0.1701, "step": 3338 }, { "epoch": 1.77, "learning_rate": 1.4996039649138383e-05, "loss": 0.2149, "step": 3339 }, { "epoch": 1.77, "learning_rate": 1.4993068700479743e-05, "loss": 0.1992, "step": 3340 }, { "epoch": 1.77, "learning_rate": 1.4990097164627229e-05, "loss": 0.2251, "step": 3341 }, { "epoch": 1.77, "learning_rate": 1.4987125041930299e-05, "loss": 0.1808, "step": 3342 }, { "epoch": 1.77, "learning_rate": 1.498415233273848e-05, "loss": 0.0596, "step": 3343 }, { "epoch": 1.77, "learning_rate": 1.4981179037401366e-05, "loss": 0.1241, "step": 3344 }, { "epoch": 1.77, "learning_rate": 1.4978205156268624e-05, "loss": 0.1725, "step": 3345 }, { "epoch": 1.77, "learning_rate": 1.4975230689689986e-05, "loss": 0.1152, "step": 3346 }, { "epoch": 1.77, "learning_rate": 1.4972255638015258e-05, "loss": 0.1036, "step": 3347 }, { "epoch": 1.77, "learning_rate": 1.4969280001594309e-05, "loss": 0.1495, "step": 3348 }, { "epoch": 1.77, "learning_rate": 1.4966303780777076e-05, "loss": 0.2056, "step": 3349 }, { "epoch": 1.77, "learning_rate": 1.4963326975913573e-05, "loss": 0.1944, "step": 3350 }, { "epoch": 1.77, "learning_rate": 1.4960349587353878e-05, "loss": 0.2117, "step": 3351 }, { "epoch": 1.77, "learning_rate": 1.4957371615448133e-05, "loss": 0.2571, "step": 3352 }, { "epoch": 1.77, "learning_rate": 1.4954393060546553e-05, "loss": 0.147, "step": 3353 }, { "epoch": 1.78, "learning_rate": 1.4951413922999427e-05, "loss": 0.1924, "step": 3354 }, { "epoch": 1.78, "learning_rate": 1.49484342031571e-05, "loss": 0.1882, "step": 3355 }, { "epoch": 1.78, "learning_rate": 1.494545390137e-05, "loss": 0.1794, "step": 3356 }, { "epoch": 1.78, "learning_rate": 1.4942473017988607e-05, "loss": 0.1737, "step": 3357 }, { "epoch": 1.78, "learning_rate": 1.4939491553363486e-05, "loss": 0.2544, "step": 3358 }, { "epoch": 1.78, "learning_rate": 1.4936509507845257e-05, "loss": 0.1298, "step": 3359 }, { "epoch": 1.78, "learning_rate": 1.4933526881784618e-05, "loss": 0.185, "step": 3360 }, { "epoch": 1.78, "learning_rate": 1.4930543675532331e-05, "loss": 0.1596, "step": 3361 }, { "epoch": 1.78, "learning_rate": 1.4927559889439222e-05, "loss": 0.1673, "step": 3362 }, { "epoch": 1.78, "learning_rate": 1.4924575523856197e-05, "loss": 0.2909, "step": 3363 }, { "epoch": 1.78, "learning_rate": 1.4921590579134218e-05, "loss": 0.183, "step": 3364 }, { "epoch": 1.78, "learning_rate": 1.4918605055624315e-05, "loss": 0.1344, "step": 3365 }, { "epoch": 1.78, "learning_rate": 1.49156189536776e-05, "loss": 0.1351, "step": 3366 }, { "epoch": 1.78, "learning_rate": 1.491263227364524e-05, "loss": 0.1446, "step": 3367 }, { "epoch": 1.78, "learning_rate": 1.4909645015878473e-05, "loss": 0.1076, "step": 3368 }, { "epoch": 1.78, "learning_rate": 1.4906657180728608e-05, "loss": 0.1429, "step": 3369 }, { "epoch": 1.78, "learning_rate": 1.490366876854702e-05, "loss": 0.1786, "step": 3370 }, { "epoch": 1.78, "learning_rate": 1.490067977968515e-05, "loss": 0.209, "step": 3371 }, { "epoch": 1.78, "learning_rate": 1.4897690214494506e-05, "loss": 0.1417, "step": 3372 }, { "epoch": 1.79, "learning_rate": 1.4894700073326673e-05, "loss": 0.236, "step": 3373 }, { "epoch": 1.79, "learning_rate": 1.489170935653329e-05, "loss": 0.072, "step": 3374 }, { "epoch": 1.79, "learning_rate": 1.4888718064466076e-05, "loss": 0.1194, "step": 3375 }, { "epoch": 1.79, "learning_rate": 1.4885726197476809e-05, "loss": 0.2279, "step": 3376 }, { "epoch": 1.79, "learning_rate": 1.488273375591734e-05, "loss": 0.3112, "step": 3377 }, { "epoch": 1.79, "learning_rate": 1.4879740740139585e-05, "loss": 0.1959, "step": 3378 }, { "epoch": 1.79, "learning_rate": 1.487674715049553e-05, "loss": 0.2691, "step": 3379 }, { "epoch": 1.79, "learning_rate": 1.4873752987337223e-05, "loss": 0.1761, "step": 3380 }, { "epoch": 1.79, "learning_rate": 1.4870758251016779e-05, "loss": 0.1587, "step": 3381 }, { "epoch": 1.79, "learning_rate": 1.4867762941886395e-05, "loss": 0.1071, "step": 3382 }, { "epoch": 1.79, "learning_rate": 1.4864767060298319e-05, "loss": 0.2148, "step": 3383 }, { "epoch": 1.79, "learning_rate": 1.4861770606604873e-05, "loss": 0.2509, "step": 3384 }, { "epoch": 1.79, "learning_rate": 1.4858773581158443e-05, "loss": 0.1779, "step": 3385 }, { "epoch": 1.79, "learning_rate": 1.4855775984311485e-05, "loss": 0.2634, "step": 3386 }, { "epoch": 1.79, "learning_rate": 1.4852777816416526e-05, "loss": 0.1178, "step": 3387 }, { "epoch": 1.79, "learning_rate": 1.4849779077826152e-05, "loss": 0.1669, "step": 3388 }, { "epoch": 1.79, "learning_rate": 1.4846779768893022e-05, "loss": 0.1217, "step": 3389 }, { "epoch": 1.79, "learning_rate": 1.484377988996986e-05, "loss": 0.1244, "step": 3390 }, { "epoch": 1.79, "learning_rate": 1.4840779441409455e-05, "loss": 0.1973, "step": 3391 }, { "epoch": 1.8, "learning_rate": 1.4837778423564668e-05, "loss": 0.1895, "step": 3392 }, { "epoch": 1.8, "learning_rate": 1.4834776836788421e-05, "loss": 0.1164, "step": 3393 }, { "epoch": 1.8, "learning_rate": 1.4831774681433708e-05, "loss": 0.2915, "step": 3394 }, { "epoch": 1.8, "learning_rate": 1.4828771957853592e-05, "loss": 0.2024, "step": 3395 }, { "epoch": 1.8, "learning_rate": 1.4825768666401188e-05, "loss": 0.1894, "step": 3396 }, { "epoch": 1.8, "learning_rate": 1.4822764807429698e-05, "loss": 0.1315, "step": 3397 }, { "epoch": 1.8, "learning_rate": 1.4819760381292376e-05, "loss": 0.151, "step": 3398 }, { "epoch": 1.8, "learning_rate": 1.4816755388342557e-05, "loss": 0.2642, "step": 3399 }, { "epoch": 1.8, "learning_rate": 1.481374982893362e-05, "loss": 0.1467, "step": 3400 }, { "epoch": 1.8, "learning_rate": 1.481074370341903e-05, "loss": 0.1905, "step": 3401 }, { "epoch": 1.8, "learning_rate": 1.4807737012152314e-05, "loss": 0.1337, "step": 3402 }, { "epoch": 1.8, "learning_rate": 1.4804729755487071e-05, "loss": 0.2448, "step": 3403 }, { "epoch": 1.8, "learning_rate": 1.4801721933776944e-05, "loss": 0.2336, "step": 3404 }, { "epoch": 1.8, "learning_rate": 1.4798713547375672e-05, "loss": 0.1987, "step": 3405 }, { "epoch": 1.8, "learning_rate": 1.4795704596637039e-05, "loss": 0.1439, "step": 3406 }, { "epoch": 1.8, "learning_rate": 1.4792695081914907e-05, "loss": 0.1046, "step": 3407 }, { "epoch": 1.8, "learning_rate": 1.4789685003563199e-05, "loss": 0.2234, "step": 3408 }, { "epoch": 1.8, "learning_rate": 1.4786674361935904e-05, "loss": 0.1713, "step": 3409 }, { "epoch": 1.8, "learning_rate": 1.4783663157387083e-05, "loss": 0.0767, "step": 3410 }, { "epoch": 1.81, "learning_rate": 1.4780651390270856e-05, "loss": 0.1824, "step": 3411 }, { "epoch": 1.81, "learning_rate": 1.477763906094141e-05, "loss": 0.127, "step": 3412 }, { "epoch": 1.81, "learning_rate": 1.4774626169753008e-05, "loss": 0.1788, "step": 3413 }, { "epoch": 1.81, "learning_rate": 1.4771612717059962e-05, "loss": 0.2455, "step": 3414 }, { "epoch": 1.81, "learning_rate": 1.476859870321667e-05, "loss": 0.2204, "step": 3415 }, { "epoch": 1.81, "learning_rate": 1.4765584128577575e-05, "loss": 0.0896, "step": 3416 }, { "epoch": 1.81, "learning_rate": 1.47625689934972e-05, "loss": 0.1832, "step": 3417 }, { "epoch": 1.81, "learning_rate": 1.4759553298330132e-05, "loss": 0.207, "step": 3418 }, { "epoch": 1.81, "learning_rate": 1.4756537043431024e-05, "loss": 0.169, "step": 3419 }, { "epoch": 1.81, "learning_rate": 1.475352022915459e-05, "loss": 0.1793, "step": 3420 }, { "epoch": 1.81, "learning_rate": 1.4750502855855611e-05, "loss": 0.1747, "step": 3421 }, { "epoch": 1.81, "learning_rate": 1.474748492388894e-05, "loss": 0.1996, "step": 3422 }, { "epoch": 1.81, "learning_rate": 1.4744466433609486e-05, "loss": 0.2453, "step": 3423 }, { "epoch": 1.81, "learning_rate": 1.4741447385372233e-05, "loss": 0.2448, "step": 3424 }, { "epoch": 1.81, "learning_rate": 1.4738427779532228e-05, "loss": 0.1634, "step": 3425 }, { "epoch": 1.81, "learning_rate": 1.4735407616444577e-05, "loss": 0.219, "step": 3426 }, { "epoch": 1.81, "learning_rate": 1.4732386896464462e-05, "loss": 0.1175, "step": 3427 }, { "epoch": 1.81, "learning_rate": 1.4729365619947118e-05, "loss": 0.1299, "step": 3428 }, { "epoch": 1.82, "learning_rate": 1.4726343787247858e-05, "loss": 0.1209, "step": 3429 }, { "epoch": 1.82, "learning_rate": 1.4723321398722057e-05, "loss": 0.1119, "step": 3430 }, { "epoch": 1.82, "learning_rate": 1.4720298454725147e-05, "loss": 0.1302, "step": 3431 }, { "epoch": 1.82, "learning_rate": 1.4717274955612635e-05, "loss": 0.1639, "step": 3432 }, { "epoch": 1.82, "learning_rate": 1.471425090174009e-05, "loss": 0.161, "step": 3433 }, { "epoch": 1.82, "learning_rate": 1.4711226293463145e-05, "loss": 0.2101, "step": 3434 }, { "epoch": 1.82, "learning_rate": 1.4708201131137502e-05, "loss": 0.1458, "step": 3435 }, { "epoch": 1.82, "learning_rate": 1.4705175415118926e-05, "loss": 0.1867, "step": 3436 }, { "epoch": 1.82, "learning_rate": 1.4702149145763242e-05, "loss": 0.1462, "step": 3437 }, { "epoch": 1.82, "learning_rate": 1.4699122323426344e-05, "loss": 0.2031, "step": 3438 }, { "epoch": 1.82, "learning_rate": 1.4696094948464199e-05, "loss": 0.2547, "step": 3439 }, { "epoch": 1.82, "learning_rate": 1.469306702123283e-05, "loss": 0.2025, "step": 3440 }, { "epoch": 1.82, "learning_rate": 1.469003854208832e-05, "loss": 0.1371, "step": 3441 }, { "epoch": 1.82, "learning_rate": 1.468700951138683e-05, "loss": 0.2756, "step": 3442 }, { "epoch": 1.82, "learning_rate": 1.4683979929484576e-05, "loss": 0.1469, "step": 3443 }, { "epoch": 1.82, "learning_rate": 1.4680949796737845e-05, "loss": 0.1766, "step": 3444 }, { "epoch": 1.82, "learning_rate": 1.4677919113502983e-05, "loss": 0.2065, "step": 3445 }, { "epoch": 1.82, "learning_rate": 1.4674887880136408e-05, "loss": 0.1304, "step": 3446 }, { "epoch": 1.82, "learning_rate": 1.4671856096994594e-05, "loss": 0.2294, "step": 3447 }, { "epoch": 1.83, "learning_rate": 1.4668823764434088e-05, "loss": 0.1696, "step": 3448 }, { "epoch": 1.83, "learning_rate": 1.4665790882811494e-05, "loss": 0.2438, "step": 3449 }, { "epoch": 1.83, "learning_rate": 1.466275745248349e-05, "loss": 0.2191, "step": 3450 }, { "epoch": 1.83, "learning_rate": 1.4659723473806806e-05, "loss": 0.2225, "step": 3451 }, { "epoch": 1.83, "learning_rate": 1.465668894713825e-05, "loss": 0.2065, "step": 3452 }, { "epoch": 1.83, "learning_rate": 1.4653653872834684e-05, "loss": 0.2456, "step": 3453 }, { "epoch": 1.83, "learning_rate": 1.4650618251253033e-05, "loss": 0.1544, "step": 3454 }, { "epoch": 1.83, "learning_rate": 1.4647582082750302e-05, "loss": 0.2662, "step": 3455 }, { "epoch": 1.83, "learning_rate": 1.4644545367683546e-05, "loss": 0.1723, "step": 3456 }, { "epoch": 1.83, "learning_rate": 1.4641508106409887e-05, "loss": 0.1403, "step": 3457 }, { "epoch": 1.83, "learning_rate": 1.4638470299286512e-05, "loss": 0.2128, "step": 3458 }, { "epoch": 1.83, "learning_rate": 1.4635431946670672e-05, "loss": 0.0581, "step": 3459 }, { "epoch": 1.83, "learning_rate": 1.4632393048919685e-05, "loss": 0.153, "step": 3460 }, { "epoch": 1.83, "learning_rate": 1.4629353606390928e-05, "loss": 0.2406, "step": 3461 }, { "epoch": 1.83, "learning_rate": 1.462631361944185e-05, "loss": 0.1533, "step": 3462 }, { "epoch": 1.83, "learning_rate": 1.4623273088429952e-05, "loss": 0.1545, "step": 3463 }, { "epoch": 1.83, "learning_rate": 1.4620232013712813e-05, "loss": 0.3004, "step": 3464 }, { "epoch": 1.83, "learning_rate": 1.4617190395648064e-05, "loss": 0.2707, "step": 3465 }, { "epoch": 1.83, "learning_rate": 1.4614148234593406e-05, "loss": 0.2059, "step": 3466 }, { "epoch": 1.84, "learning_rate": 1.4611105530906604e-05, "loss": 0.1901, "step": 3467 }, { "epoch": 1.84, "learning_rate": 1.4608062284945486e-05, "loss": 0.1336, "step": 3468 }, { "epoch": 1.84, "learning_rate": 1.4605018497067938e-05, "loss": 0.2383, "step": 3469 }, { "epoch": 1.84, "learning_rate": 1.4601974167631919e-05, "loss": 0.2226, "step": 3470 }, { "epoch": 1.84, "learning_rate": 1.4598929296995448e-05, "loss": 0.3625, "step": 3471 }, { "epoch": 1.84, "learning_rate": 1.4595883885516609e-05, "loss": 0.222, "step": 3472 }, { "epoch": 1.84, "learning_rate": 1.459283793355354e-05, "loss": 0.1882, "step": 3473 }, { "epoch": 1.84, "learning_rate": 1.458979144146446e-05, "loss": 0.0806, "step": 3474 }, { "epoch": 1.84, "learning_rate": 1.4586744409607634e-05, "loss": 0.1439, "step": 3475 }, { "epoch": 1.84, "learning_rate": 1.4583696838341406e-05, "loss": 0.1804, "step": 3476 }, { "epoch": 1.84, "learning_rate": 1.458064872802417e-05, "loss": 0.2031, "step": 3477 }, { "epoch": 1.84, "learning_rate": 1.4577600079014392e-05, "loss": 0.177, "step": 3478 }, { "epoch": 1.84, "learning_rate": 1.4574550891670595e-05, "loss": 0.2646, "step": 3479 }, { "epoch": 1.84, "learning_rate": 1.4571501166351374e-05, "loss": 0.1895, "step": 3480 }, { "epoch": 1.84, "learning_rate": 1.4568450903415381e-05, "loss": 0.2051, "step": 3481 }, { "epoch": 1.84, "learning_rate": 1.456540010322133e-05, "loss": 0.1516, "step": 3482 }, { "epoch": 1.84, "learning_rate": 1.4562348766128e-05, "loss": 0.1673, "step": 3483 }, { "epoch": 1.84, "learning_rate": 1.455929689249424e-05, "loss": 0.12, "step": 3484 }, { "epoch": 1.84, "learning_rate": 1.4556244482678947e-05, "loss": 0.1597, "step": 3485 }, { "epoch": 1.85, "learning_rate": 1.4553191537041098e-05, "loss": 0.167, "step": 3486 }, { "epoch": 1.85, "learning_rate": 1.455013805593972e-05, "loss": 0.0798, "step": 3487 }, { "epoch": 1.85, "learning_rate": 1.4547084039733912e-05, "loss": 0.1907, "step": 3488 }, { "epoch": 1.85, "learning_rate": 1.4544029488782826e-05, "loss": 0.1532, "step": 3489 }, { "epoch": 1.85, "learning_rate": 1.4540974403445685e-05, "loss": 0.1696, "step": 3490 }, { "epoch": 1.85, "learning_rate": 1.4537918784081775e-05, "loss": 0.2392, "step": 3491 }, { "epoch": 1.85, "learning_rate": 1.4534862631050442e-05, "loss": 0.1224, "step": 3492 }, { "epoch": 1.85, "learning_rate": 1.453180594471109e-05, "loss": 0.2213, "step": 3493 }, { "epoch": 1.85, "learning_rate": 1.4528748725423199e-05, "loss": 0.1163, "step": 3494 }, { "epoch": 1.85, "learning_rate": 1.4525690973546293e-05, "loss": 0.1748, "step": 3495 }, { "epoch": 1.85, "learning_rate": 1.452263268943998e-05, "loss": 0.166, "step": 3496 }, { "epoch": 1.85, "learning_rate": 1.4519573873463913e-05, "loss": 0.265, "step": 3497 }, { "epoch": 1.85, "learning_rate": 1.4516514525977816e-05, "loss": 0.2009, "step": 3498 }, { "epoch": 1.85, "learning_rate": 1.4513454647341476e-05, "loss": 0.1991, "step": 3499 }, { "epoch": 1.85, "learning_rate": 1.4510394237914736e-05, "loss": 0.1655, "step": 3500 }, { "epoch": 1.85, "learning_rate": 1.4507333298057509e-05, "loss": 0.2057, "step": 3501 }, { "epoch": 1.85, "learning_rate": 1.4504271828129766e-05, "loss": 0.105, "step": 3502 }, { "epoch": 1.85, "learning_rate": 1.450120982849154e-05, "loss": 0.1415, "step": 3503 }, { "epoch": 1.85, "learning_rate": 1.4498147299502932e-05, "loss": 0.2088, "step": 3504 }, { "epoch": 1.86, "learning_rate": 1.4495084241524095e-05, "loss": 0.1922, "step": 3505 }, { "epoch": 1.86, "learning_rate": 1.449202065491525e-05, "loss": 0.1783, "step": 3506 }, { "epoch": 1.86, "learning_rate": 1.4488956540036687e-05, "loss": 0.2008, "step": 3507 }, { "epoch": 1.86, "learning_rate": 1.4485891897248752e-05, "loss": 0.1267, "step": 3508 }, { "epoch": 1.86, "learning_rate": 1.4482826726911843e-05, "loss": 0.1321, "step": 3509 }, { "epoch": 1.86, "learning_rate": 1.4479761029386434e-05, "loss": 0.1855, "step": 3510 }, { "epoch": 1.86, "learning_rate": 1.4476694805033058e-05, "loss": 0.0957, "step": 3511 }, { "epoch": 1.86, "learning_rate": 1.4473628054212309e-05, "loss": 0.1655, "step": 3512 }, { "epoch": 1.86, "learning_rate": 1.4470560777284841e-05, "loss": 0.1424, "step": 3513 }, { "epoch": 1.86, "learning_rate": 1.4467492974611372e-05, "loss": 0.1481, "step": 3514 }, { "epoch": 1.86, "learning_rate": 1.4464424646552679e-05, "loss": 0.1675, "step": 3515 }, { "epoch": 1.86, "learning_rate": 1.4461355793469606e-05, "loss": 0.0993, "step": 3516 }, { "epoch": 1.86, "learning_rate": 1.4458286415723056e-05, "loss": 0.1595, "step": 3517 }, { "epoch": 1.86, "learning_rate": 1.4455216513673989e-05, "loss": 0.1415, "step": 3518 }, { "epoch": 1.86, "learning_rate": 1.4452146087683434e-05, "loss": 0.1927, "step": 3519 }, { "epoch": 1.86, "learning_rate": 1.444907513811248e-05, "loss": 0.1963, "step": 3520 }, { "epoch": 1.86, "learning_rate": 1.4446003665322273e-05, "loss": 0.1671, "step": 3521 }, { "epoch": 1.86, "learning_rate": 1.4442931669674023e-05, "loss": 0.2063, "step": 3522 }, { "epoch": 1.86, "learning_rate": 1.4439859151529007e-05, "loss": 0.0913, "step": 3523 }, { "epoch": 1.87, "learning_rate": 1.4436786111248557e-05, "loss": 0.1561, "step": 3524 }, { "epoch": 1.87, "learning_rate": 1.4433712549194065e-05, "loss": 0.1264, "step": 3525 }, { "epoch": 1.87, "learning_rate": 1.4430638465726991e-05, "loss": 0.0814, "step": 3526 }, { "epoch": 1.87, "learning_rate": 1.442756386120885e-05, "loss": 0.1759, "step": 3527 }, { "epoch": 1.87, "learning_rate": 1.4424488736001223e-05, "loss": 0.1627, "step": 3528 }, { "epoch": 1.87, "learning_rate": 1.442141309046575e-05, "loss": 0.2009, "step": 3529 }, { "epoch": 1.87, "learning_rate": 1.4418336924964131e-05, "loss": 0.1783, "step": 3530 }, { "epoch": 1.87, "learning_rate": 1.4415260239858131e-05, "loss": 0.1894, "step": 3531 }, { "epoch": 1.87, "learning_rate": 1.4412183035509572e-05, "loss": 0.151, "step": 3532 }, { "epoch": 1.87, "learning_rate": 1.4409105312280339e-05, "loss": 0.1964, "step": 3533 }, { "epoch": 1.87, "learning_rate": 1.4406027070532377e-05, "loss": 0.1774, "step": 3534 }, { "epoch": 1.87, "learning_rate": 1.4402948310627695e-05, "loss": 0.2223, "step": 3535 }, { "epoch": 1.87, "learning_rate": 1.439986903292836e-05, "loss": 0.0889, "step": 3536 }, { "epoch": 1.87, "learning_rate": 1.43967892377965e-05, "loss": 0.3091, "step": 3537 }, { "epoch": 1.87, "learning_rate": 1.4393708925594304e-05, "loss": 0.0952, "step": 3538 }, { "epoch": 1.87, "learning_rate": 1.439062809668402e-05, "loss": 0.1523, "step": 3539 }, { "epoch": 1.87, "learning_rate": 1.4387546751427967e-05, "loss": 0.1554, "step": 3540 }, { "epoch": 1.87, "learning_rate": 1.4384464890188508e-05, "loss": 0.2969, "step": 3541 }, { "epoch": 1.87, "learning_rate": 1.438138251332808e-05, "loss": 0.21, "step": 3542 }, { "epoch": 1.88, "learning_rate": 1.4378299621209173e-05, "loss": 0.1687, "step": 3543 }, { "epoch": 1.88, "learning_rate": 1.4375216214194347e-05, "loss": 0.1889, "step": 3544 }, { "epoch": 1.88, "learning_rate": 1.437213229264621e-05, "loss": 0.2245, "step": 3545 }, { "epoch": 1.88, "learning_rate": 1.4369047856927438e-05, "loss": 0.1677, "step": 3546 }, { "epoch": 1.88, "learning_rate": 1.4365962907400767e-05, "loss": 0.2132, "step": 3547 }, { "epoch": 1.88, "learning_rate": 1.436287744442899e-05, "loss": 0.1783, "step": 3548 }, { "epoch": 1.88, "learning_rate": 1.435979146837497e-05, "loss": 0.1433, "step": 3549 }, { "epoch": 1.88, "learning_rate": 1.4356704979601617e-05, "loss": 0.1241, "step": 3550 }, { "epoch": 1.88, "learning_rate": 1.4353617978471909e-05, "loss": 0.2708, "step": 3551 }, { "epoch": 1.88, "learning_rate": 1.4350530465348884e-05, "loss": 0.2216, "step": 3552 }, { "epoch": 1.88, "learning_rate": 1.434744244059564e-05, "loss": 0.2693, "step": 3553 }, { "epoch": 1.88, "learning_rate": 1.434435390457533e-05, "loss": 0.2393, "step": 3554 }, { "epoch": 1.88, "learning_rate": 1.4341264857651175e-05, "loss": 0.2073, "step": 3555 }, { "epoch": 1.88, "learning_rate": 1.4338175300186451e-05, "loss": 0.1274, "step": 3556 }, { "epoch": 1.88, "learning_rate": 1.4335085232544498e-05, "loss": 0.1598, "step": 3557 }, { "epoch": 1.88, "learning_rate": 1.4331994655088713e-05, "loss": 0.1256, "step": 3558 }, { "epoch": 1.88, "learning_rate": 1.4328903568182548e-05, "loss": 0.2291, "step": 3559 }, { "epoch": 1.88, "learning_rate": 1.4325811972189527e-05, "loss": 0.1475, "step": 3560 }, { "epoch": 1.88, "learning_rate": 1.4322719867473227e-05, "loss": 0.2784, "step": 3561 }, { "epoch": 1.89, "learning_rate": 1.4319627254397279e-05, "loss": 0.1852, "step": 3562 }, { "epoch": 1.89, "learning_rate": 1.4316534133325384e-05, "loss": 0.2376, "step": 3563 }, { "epoch": 1.89, "learning_rate": 1.4313440504621298e-05, "loss": 0.1725, "step": 3564 }, { "epoch": 1.89, "learning_rate": 1.4310346368648842e-05, "loss": 0.1146, "step": 3565 }, { "epoch": 1.89, "learning_rate": 1.4307251725771885e-05, "loss": 0.2488, "step": 3566 }, { "epoch": 1.89, "learning_rate": 1.4304156576354364e-05, "loss": 0.1129, "step": 3567 }, { "epoch": 1.89, "learning_rate": 1.4301060920760277e-05, "loss": 0.1653, "step": 3568 }, { "epoch": 1.89, "learning_rate": 1.4297964759353678e-05, "loss": 0.1609, "step": 3569 }, { "epoch": 1.89, "learning_rate": 1.4294868092498678e-05, "loss": 0.2555, "step": 3570 }, { "epoch": 1.89, "learning_rate": 1.4291770920559453e-05, "loss": 0.1245, "step": 3571 }, { "epoch": 1.89, "learning_rate": 1.4288673243900237e-05, "loss": 0.2242, "step": 3572 }, { "epoch": 1.89, "learning_rate": 1.428557506288532e-05, "loss": 0.1273, "step": 3573 }, { "epoch": 1.89, "learning_rate": 1.4282476377879055e-05, "loss": 0.1605, "step": 3574 }, { "epoch": 1.89, "learning_rate": 1.4279377189245854e-05, "loss": 0.1227, "step": 3575 }, { "epoch": 1.89, "learning_rate": 1.4276277497350184e-05, "loss": 0.1221, "step": 3576 }, { "epoch": 1.89, "learning_rate": 1.427317730255658e-05, "loss": 0.2161, "step": 3577 }, { "epoch": 1.89, "learning_rate": 1.4270076605229624e-05, "loss": 0.1984, "step": 3578 }, { "epoch": 1.89, "learning_rate": 1.4266975405733965e-05, "loss": 0.2895, "step": 3579 }, { "epoch": 1.89, "learning_rate": 1.4263873704434312e-05, "loss": 0.1545, "step": 3580 }, { "epoch": 1.9, "learning_rate": 1.4260771501695432e-05, "loss": 0.1586, "step": 3581 }, { "epoch": 1.9, "learning_rate": 1.4257668797882144e-05, "loss": 0.1731, "step": 3582 }, { "epoch": 1.9, "learning_rate": 1.4254565593359338e-05, "loss": 0.1066, "step": 3583 }, { "epoch": 1.9, "learning_rate": 1.4251461888491948e-05, "loss": 0.2348, "step": 3584 }, { "epoch": 1.9, "learning_rate": 1.4248357683644989e-05, "loss": 0.2648, "step": 3585 }, { "epoch": 1.9, "learning_rate": 1.4245252979183508e-05, "loss": 0.1719, "step": 3586 }, { "epoch": 1.9, "learning_rate": 1.4242147775472628e-05, "loss": 0.1788, "step": 3587 }, { "epoch": 1.9, "learning_rate": 1.4239042072877528e-05, "loss": 0.1701, "step": 3588 }, { "epoch": 1.9, "learning_rate": 1.4235935871763442e-05, "loss": 0.1524, "step": 3589 }, { "epoch": 1.9, "learning_rate": 1.4232829172495667e-05, "loss": 0.1787, "step": 3590 }, { "epoch": 1.9, "learning_rate": 1.4229721975439557e-05, "loss": 0.2945, "step": 3591 }, { "epoch": 1.9, "learning_rate": 1.422661428096052e-05, "loss": 0.2198, "step": 3592 }, { "epoch": 1.9, "learning_rate": 1.4223506089424033e-05, "loss": 0.1774, "step": 3593 }, { "epoch": 1.9, "learning_rate": 1.4220397401195616e-05, "loss": 0.1902, "step": 3594 }, { "epoch": 1.9, "learning_rate": 1.4217288216640862e-05, "loss": 0.1086, "step": 3595 }, { "epoch": 1.9, "learning_rate": 1.421417853612542e-05, "loss": 0.1523, "step": 3596 }, { "epoch": 1.9, "learning_rate": 1.4211068360014987e-05, "loss": 0.3335, "step": 3597 }, { "epoch": 1.9, "learning_rate": 1.4207957688675331e-05, "loss": 0.2104, "step": 3598 }, { "epoch": 1.9, "learning_rate": 1.4204846522472266e-05, "loss": 0.1003, "step": 3599 }, { "epoch": 1.91, "learning_rate": 1.4201734861771676e-05, "loss": 0.1739, "step": 3600 }, { "epoch": 1.91, "learning_rate": 1.4198622706939498e-05, "loss": 0.2056, "step": 3601 }, { "epoch": 1.91, "learning_rate": 1.4195510058341726e-05, "loss": 0.1717, "step": 3602 }, { "epoch": 1.91, "learning_rate": 1.419239691634441e-05, "loss": 0.1547, "step": 3603 }, { "epoch": 1.91, "learning_rate": 1.4189283281313665e-05, "loss": 0.1278, "step": 3604 }, { "epoch": 1.91, "learning_rate": 1.418616915361566e-05, "loss": 0.1112, "step": 3605 }, { "epoch": 1.91, "learning_rate": 1.4183054533616619e-05, "loss": 0.1989, "step": 3606 }, { "epoch": 1.91, "learning_rate": 1.417993942168283e-05, "loss": 0.1244, "step": 3607 }, { "epoch": 1.91, "learning_rate": 1.4176823818180632e-05, "loss": 0.1147, "step": 3608 }, { "epoch": 1.91, "learning_rate": 1.4173707723476431e-05, "loss": 0.1404, "step": 3609 }, { "epoch": 1.91, "learning_rate": 1.4170591137936683e-05, "loss": 0.124, "step": 3610 }, { "epoch": 1.91, "learning_rate": 1.4167474061927902e-05, "loss": 0.1383, "step": 3611 }, { "epoch": 1.91, "learning_rate": 1.4164356495816662e-05, "loss": 0.2135, "step": 3612 }, { "epoch": 1.91, "learning_rate": 1.4161238439969597e-05, "loss": 0.1639, "step": 3613 }, { "epoch": 1.91, "learning_rate": 1.4158119894753397e-05, "loss": 0.1123, "step": 3614 }, { "epoch": 1.91, "learning_rate": 1.4155000860534802e-05, "loss": 0.2306, "step": 3615 }, { "epoch": 1.91, "learning_rate": 1.4151881337680622e-05, "loss": 0.1018, "step": 3616 }, { "epoch": 1.91, "learning_rate": 1.414876132655772e-05, "loss": 0.1631, "step": 3617 }, { "epoch": 1.92, "learning_rate": 1.4145640827533009e-05, "loss": 0.0861, "step": 3618 }, { "epoch": 1.92, "learning_rate": 1.4142519840973468e-05, "loss": 0.1423, "step": 3619 }, { "epoch": 1.92, "learning_rate": 1.4139398367246129e-05, "loss": 0.155, "step": 3620 }, { "epoch": 1.92, "learning_rate": 1.4136276406718086e-05, "loss": 0.2381, "step": 3621 }, { "epoch": 1.92, "learning_rate": 1.4133153959756487e-05, "loss": 0.1356, "step": 3622 }, { "epoch": 1.92, "learning_rate": 1.4130031026728535e-05, "loss": 0.2158, "step": 3623 }, { "epoch": 1.92, "learning_rate": 1.4126907608001494e-05, "loss": 0.1649, "step": 3624 }, { "epoch": 1.92, "learning_rate": 1.4123783703942683e-05, "loss": 0.1816, "step": 3625 }, { "epoch": 1.92, "learning_rate": 1.4120659314919479e-05, "loss": 0.2265, "step": 3626 }, { "epoch": 1.92, "learning_rate": 1.4117534441299317e-05, "loss": 0.2838, "step": 3627 }, { "epoch": 1.92, "learning_rate": 1.4114409083449685e-05, "loss": 0.1415, "step": 3628 }, { "epoch": 1.92, "learning_rate": 1.4111283241738136e-05, "loss": 0.0605, "step": 3629 }, { "epoch": 1.92, "learning_rate": 1.4108156916532268e-05, "loss": 0.1559, "step": 3630 }, { "epoch": 1.92, "learning_rate": 1.4105030108199745e-05, "loss": 0.2748, "step": 3631 }, { "epoch": 1.92, "learning_rate": 1.4101902817108286e-05, "loss": 0.1248, "step": 3632 }, { "epoch": 1.92, "learning_rate": 1.4098775043625667e-05, "loss": 0.1254, "step": 3633 }, { "epoch": 1.92, "learning_rate": 1.4095646788119719e-05, "loss": 0.1861, "step": 3634 }, { "epoch": 1.92, "learning_rate": 1.409251805095833e-05, "loss": 0.1112, "step": 3635 }, { "epoch": 1.92, "learning_rate": 1.408938883250944e-05, "loss": 0.2263, "step": 3636 }, { "epoch": 1.93, "learning_rate": 1.4086259133141064e-05, "loss": 0.1319, "step": 3637 }, { "epoch": 1.93, "learning_rate": 1.408312895322125e-05, "loss": 0.0666, "step": 3638 }, { "epoch": 1.93, "learning_rate": 1.4079998293118113e-05, "loss": 0.1569, "step": 3639 }, { "epoch": 1.93, "learning_rate": 1.4076867153199827e-05, "loss": 0.1727, "step": 3640 }, { "epoch": 1.93, "learning_rate": 1.4073735533834618e-05, "loss": 0.1455, "step": 3641 }, { "epoch": 1.93, "learning_rate": 1.4070603435390773e-05, "loss": 0.2156, "step": 3642 }, { "epoch": 1.93, "learning_rate": 1.406747085823663e-05, "loss": 0.1335, "step": 3643 }, { "epoch": 1.93, "learning_rate": 1.4064337802740586e-05, "loss": 0.2606, "step": 3644 }, { "epoch": 1.93, "learning_rate": 1.4061204269271096e-05, "loss": 0.2732, "step": 3645 }, { "epoch": 1.93, "learning_rate": 1.4058070258196664e-05, "loss": 0.1222, "step": 3646 }, { "epoch": 1.93, "learning_rate": 1.405493576988586e-05, "loss": 0.2141, "step": 3647 }, { "epoch": 1.93, "learning_rate": 1.4051800804707305e-05, "loss": 0.2281, "step": 3648 }, { "epoch": 1.93, "learning_rate": 1.4048665363029677e-05, "loss": 0.1912, "step": 3649 }, { "epoch": 1.93, "learning_rate": 1.4045529445221705e-05, "loss": 0.1434, "step": 3650 }, { "epoch": 1.93, "learning_rate": 1.4042393051652185e-05, "loss": 0.1415, "step": 3651 }, { "epoch": 1.93, "learning_rate": 1.4039256182689958e-05, "loss": 0.1573, "step": 3652 }, { "epoch": 1.93, "learning_rate": 1.4036118838703927e-05, "loss": 0.1272, "step": 3653 }, { "epoch": 1.93, "learning_rate": 1.403298102006305e-05, "loss": 0.1285, "step": 3654 }, { "epoch": 1.93, "learning_rate": 1.4029842727136337e-05, "loss": 0.2314, "step": 3655 }, { "epoch": 1.94, "learning_rate": 1.4026703960292862e-05, "loss": 0.2028, "step": 3656 }, { "epoch": 1.94, "learning_rate": 1.4023564719901746e-05, "loss": 0.1888, "step": 3657 }, { "epoch": 1.94, "learning_rate": 1.402042500633217e-05, "loss": 0.2195, "step": 3658 }, { "epoch": 1.94, "learning_rate": 1.401728481995337e-05, "loss": 0.2198, "step": 3659 }, { "epoch": 1.94, "learning_rate": 1.4014144161134637e-05, "loss": 0.2014, "step": 3660 }, { "epoch": 1.94, "learning_rate": 1.401100303024532e-05, "loss": 0.1829, "step": 3661 }, { "epoch": 1.94, "learning_rate": 1.4007861427654821e-05, "loss": 0.1686, "step": 3662 }, { "epoch": 1.94, "learning_rate": 1.4004719353732597e-05, "loss": 0.2041, "step": 3663 }, { "epoch": 1.94, "learning_rate": 1.400157680884816e-05, "loss": 0.142, "step": 3664 }, { "epoch": 1.94, "learning_rate": 1.3998433793371087e-05, "loss": 0.1715, "step": 3665 }, { "epoch": 1.94, "learning_rate": 1.3995290307670992e-05, "loss": 0.1274, "step": 3666 }, { "epoch": 1.94, "learning_rate": 1.399214635211756e-05, "loss": 0.2088, "step": 3667 }, { "epoch": 1.94, "learning_rate": 1.3989001927080522e-05, "loss": 0.1826, "step": 3668 }, { "epoch": 1.94, "learning_rate": 1.3985857032929676e-05, "loss": 0.2171, "step": 3669 }, { "epoch": 1.94, "learning_rate": 1.3982711670034862e-05, "loss": 0.1545, "step": 3670 }, { "epoch": 1.94, "learning_rate": 1.3979565838765977e-05, "loss": 0.1663, "step": 3671 }, { "epoch": 1.94, "learning_rate": 1.397641953949298e-05, "loss": 0.1498, "step": 3672 }, { "epoch": 1.94, "learning_rate": 1.397327277258588e-05, "loss": 0.1346, "step": 3673 }, { "epoch": 1.94, "learning_rate": 1.397012553841475e-05, "loss": 0.183, "step": 3674 }, { "epoch": 1.95, "learning_rate": 1.39669778373497e-05, "loss": 0.1902, "step": 3675 }, { "epoch": 1.95, "learning_rate": 1.396382966976091e-05, "loss": 0.1541, "step": 3676 }, { "epoch": 1.95, "learning_rate": 1.396068103601861e-05, "loss": 0.1941, "step": 3677 }, { "epoch": 1.95, "learning_rate": 1.3957531936493082e-05, "loss": 0.2046, "step": 3678 }, { "epoch": 1.95, "learning_rate": 1.3954382371554674e-05, "loss": 0.3764, "step": 3679 }, { "epoch": 1.95, "learning_rate": 1.395123234157377e-05, "loss": 0.1344, "step": 3680 }, { "epoch": 1.95, "learning_rate": 1.3948081846920826e-05, "loss": 0.1351, "step": 3681 }, { "epoch": 1.95, "learning_rate": 1.3944930887966346e-05, "loss": 0.1978, "step": 3682 }, { "epoch": 1.95, "learning_rate": 1.3941779465080885e-05, "loss": 0.1285, "step": 3683 }, { "epoch": 1.95, "learning_rate": 1.3938627578635055e-05, "loss": 0.1537, "step": 3684 }, { "epoch": 1.95, "learning_rate": 1.3935475228999529e-05, "loss": 0.1224, "step": 3685 }, { "epoch": 1.95, "learning_rate": 1.3932322416545027e-05, "loss": 0.1922, "step": 3686 }, { "epoch": 1.95, "learning_rate": 1.3929169141642324e-05, "loss": 0.1394, "step": 3687 }, { "epoch": 1.95, "learning_rate": 1.3926015404662248e-05, "loss": 0.1701, "step": 3688 }, { "epoch": 1.95, "learning_rate": 1.3922861205975689e-05, "loss": 0.148, "step": 3689 }, { "epoch": 1.95, "learning_rate": 1.3919706545953587e-05, "loss": 0.1911, "step": 3690 }, { "epoch": 1.95, "learning_rate": 1.3916551424966934e-05, "loss": 0.1924, "step": 3691 }, { "epoch": 1.95, "learning_rate": 1.3913395843386775e-05, "loss": 0.0765, "step": 3692 }, { "epoch": 1.95, "learning_rate": 1.3910239801584213e-05, "loss": 0.158, "step": 3693 }, { "epoch": 1.96, "learning_rate": 1.3907083299930409e-05, "loss": 0.0999, "step": 3694 }, { "epoch": 1.96, "learning_rate": 1.3903926338796567e-05, "loss": 0.1845, "step": 3695 }, { "epoch": 1.96, "learning_rate": 1.3900768918553958e-05, "loss": 0.2301, "step": 3696 }, { "epoch": 1.96, "learning_rate": 1.3897611039573895e-05, "loss": 0.0928, "step": 3697 }, { "epoch": 1.96, "learning_rate": 1.3894452702227751e-05, "loss": 0.1568, "step": 3698 }, { "epoch": 1.96, "learning_rate": 1.3891293906886953e-05, "loss": 0.1486, "step": 3699 }, { "epoch": 1.96, "learning_rate": 1.3888134653922984e-05, "loss": 0.1613, "step": 3700 }, { "epoch": 1.96, "learning_rate": 1.3884974943707373e-05, "loss": 0.2389, "step": 3701 }, { "epoch": 1.96, "learning_rate": 1.388181477661171e-05, "loss": 0.0823, "step": 3702 }, { "epoch": 1.96, "learning_rate": 1.3878654153007636e-05, "loss": 0.2342, "step": 3703 }, { "epoch": 1.96, "learning_rate": 1.3875493073266844e-05, "loss": 0.2129, "step": 3704 }, { "epoch": 1.96, "learning_rate": 1.3872331537761085e-05, "loss": 0.1718, "step": 3705 }, { "epoch": 1.96, "learning_rate": 1.3869169546862165e-05, "loss": 0.1337, "step": 3706 }, { "epoch": 1.96, "learning_rate": 1.3866007100941931e-05, "loss": 0.1757, "step": 3707 }, { "epoch": 1.96, "learning_rate": 1.3862844200372299e-05, "loss": 0.1459, "step": 3708 }, { "epoch": 1.96, "learning_rate": 1.3859680845525228e-05, "loss": 0.1213, "step": 3709 }, { "epoch": 1.96, "learning_rate": 1.385651703677274e-05, "loss": 0.1771, "step": 3710 }, { "epoch": 1.96, "learning_rate": 1.3853352774486897e-05, "loss": 0.2908, "step": 3711 }, { "epoch": 1.96, "learning_rate": 1.3850188059039828e-05, "loss": 0.2322, "step": 3712 }, { "epoch": 1.97, "learning_rate": 1.3847022890803705e-05, "loss": 0.2232, "step": 3713 }, { "epoch": 1.97, "learning_rate": 1.384385727015076e-05, "loss": 0.1438, "step": 3714 }, { "epoch": 1.97, "learning_rate": 1.3840691197453275e-05, "loss": 0.1852, "step": 3715 }, { "epoch": 1.97, "learning_rate": 1.3837524673083586e-05, "loss": 0.1926, "step": 3716 }, { "epoch": 1.97, "learning_rate": 1.3834357697414082e-05, "loss": 0.1275, "step": 3717 }, { "epoch": 1.97, "learning_rate": 1.3831190270817208e-05, "loss": 0.1313, "step": 3718 }, { "epoch": 1.97, "learning_rate": 1.3828022393665451e-05, "loss": 0.2674, "step": 3719 }, { "epoch": 1.97, "learning_rate": 1.3824854066331366e-05, "loss": 0.2919, "step": 3720 }, { "epoch": 1.97, "learning_rate": 1.3821685289187552e-05, "loss": 0.114, "step": 3721 }, { "epoch": 1.97, "learning_rate": 1.3818516062606664e-05, "loss": 0.1257, "step": 3722 }, { "epoch": 1.97, "learning_rate": 1.3815346386961409e-05, "loss": 0.2637, "step": 3723 }, { "epoch": 1.97, "learning_rate": 1.3812176262624541e-05, "loss": 0.0788, "step": 3724 }, { "epoch": 1.97, "learning_rate": 1.3809005689968874e-05, "loss": 0.1478, "step": 3725 }, { "epoch": 1.97, "learning_rate": 1.3805834669367284e-05, "loss": 0.1421, "step": 3726 }, { "epoch": 1.97, "learning_rate": 1.3802663201192676e-05, "loss": 0.1795, "step": 3727 }, { "epoch": 1.97, "learning_rate": 1.3799491285818028e-05, "loss": 0.1809, "step": 3728 }, { "epoch": 1.97, "learning_rate": 1.3796318923616357e-05, "loss": 0.1417, "step": 3729 }, { "epoch": 1.97, "learning_rate": 1.379314611496074e-05, "loss": 0.2185, "step": 3730 }, { "epoch": 1.97, "learning_rate": 1.3789972860224308e-05, "loss": 0.113, "step": 3731 }, { "epoch": 1.98, "learning_rate": 1.378679915978024e-05, "loss": 0.1657, "step": 3732 }, { "epoch": 1.98, "learning_rate": 1.3783625014001768e-05, "loss": 0.1915, "step": 3733 }, { "epoch": 1.98, "learning_rate": 1.3780450423262177e-05, "loss": 0.0884, "step": 3734 }, { "epoch": 1.98, "learning_rate": 1.3777275387934808e-05, "loss": 0.1944, "step": 3735 }, { "epoch": 1.98, "learning_rate": 1.3774099908393045e-05, "loss": 0.1697, "step": 3736 }, { "epoch": 1.98, "learning_rate": 1.3770923985010335e-05, "loss": 0.1612, "step": 3737 }, { "epoch": 1.98, "learning_rate": 1.3767747618160175e-05, "loss": 0.1264, "step": 3738 }, { "epoch": 1.98, "learning_rate": 1.3764570808216102e-05, "loss": 0.1524, "step": 3739 }, { "epoch": 1.98, "learning_rate": 1.3761393555551718e-05, "loss": 0.2743, "step": 3740 }, { "epoch": 1.98, "learning_rate": 1.375821586054068e-05, "loss": 0.1965, "step": 3741 }, { "epoch": 1.98, "learning_rate": 1.3755037723556686e-05, "loss": 0.0865, "step": 3742 }, { "epoch": 1.98, "learning_rate": 1.375185914497349e-05, "loss": 0.1815, "step": 3743 }, { "epoch": 1.98, "learning_rate": 1.37486801251649e-05, "loss": 0.1758, "step": 3744 }, { "epoch": 1.98, "learning_rate": 1.3745500664504771e-05, "loss": 0.2089, "step": 3745 }, { "epoch": 1.98, "learning_rate": 1.3742320763367016e-05, "loss": 0.2349, "step": 3746 }, { "epoch": 1.98, "learning_rate": 1.3739140422125601e-05, "loss": 0.2813, "step": 3747 }, { "epoch": 1.98, "learning_rate": 1.3735959641154533e-05, "loss": 0.1271, "step": 3748 }, { "epoch": 1.98, "learning_rate": 1.373277842082788e-05, "loss": 0.272, "step": 3749 }, { "epoch": 1.98, "learning_rate": 1.3729596761519764e-05, "loss": 0.1819, "step": 3750 }, { "epoch": 1.99, "learning_rate": 1.3726414663604345e-05, "loss": 0.2469, "step": 3751 }, { "epoch": 1.99, "learning_rate": 1.372323212745585e-05, "loss": 0.2521, "step": 3752 }, { "epoch": 1.99, "learning_rate": 1.3720049153448551e-05, "loss": 0.1597, "step": 3753 }, { "epoch": 1.99, "learning_rate": 1.3716865741956769e-05, "loss": 0.2418, "step": 3754 }, { "epoch": 1.99, "learning_rate": 1.3713681893354877e-05, "loss": 0.1171, "step": 3755 }, { "epoch": 1.99, "learning_rate": 1.3710497608017304e-05, "loss": 0.2232, "step": 3756 }, { "epoch": 1.99, "learning_rate": 1.370731288631853e-05, "loss": 0.1234, "step": 3757 }, { "epoch": 1.99, "learning_rate": 1.3704127728633081e-05, "loss": 0.1871, "step": 3758 }, { "epoch": 1.99, "learning_rate": 1.3700942135335539e-05, "loss": 0.1985, "step": 3759 }, { "epoch": 1.99, "learning_rate": 1.3697756106800532e-05, "loss": 0.1435, "step": 3760 }, { "epoch": 1.99, "learning_rate": 1.3694569643402745e-05, "loss": 0.2063, "step": 3761 }, { "epoch": 1.99, "learning_rate": 1.3691382745516914e-05, "loss": 0.2224, "step": 3762 }, { "epoch": 1.99, "learning_rate": 1.3688195413517819e-05, "loss": 0.1976, "step": 3763 }, { "epoch": 1.99, "learning_rate": 1.3685007647780303e-05, "loss": 0.1382, "step": 3764 }, { "epoch": 1.99, "learning_rate": 1.3681819448679246e-05, "loss": 0.2375, "step": 3765 }, { "epoch": 1.99, "learning_rate": 1.3678630816589589e-05, "loss": 0.173, "step": 3766 }, { "epoch": 1.99, "learning_rate": 1.367544175188632e-05, "loss": 0.1753, "step": 3767 }, { "epoch": 1.99, "learning_rate": 1.3672252254944481e-05, "loss": 0.2525, "step": 3768 }, { "epoch": 1.99, "learning_rate": 1.3669062326139159e-05, "loss": 0.2096, "step": 3769 }, { "epoch": 2.0, "learning_rate": 1.3665871965845501e-05, "loss": 0.1877, "step": 3770 }, { "epoch": 2.0, "learning_rate": 1.366268117443869e-05, "loss": 0.1949, "step": 3771 }, { "epoch": 2.0, "learning_rate": 1.3659489952293979e-05, "loss": 0.1573, "step": 3772 }, { "epoch": 2.0, "learning_rate": 1.3656298299786654e-05, "loss": 0.2457, "step": 3773 }, { "epoch": 2.0, "learning_rate": 1.3653106217292065e-05, "loss": 0.1874, "step": 3774 }, { "epoch": 2.0, "learning_rate": 1.3649913705185603e-05, "loss": 0.1405, "step": 3775 }, { "epoch": 2.0, "learning_rate": 1.3646720763842712e-05, "loss": 0.1668, "step": 3776 }, { "epoch": 2.0, "learning_rate": 1.3643527393638887e-05, "loss": 0.1477, "step": 3777 }, { "epoch": 2.0, "learning_rate": 1.364033359494968e-05, "loss": 0.196, "step": 3778 }, { "epoch": 2.0, "learning_rate": 1.3637139368150682e-05, "loss": 0.1639, "step": 3779 }, { "epoch": 2.0, "learning_rate": 1.3633944713617544e-05, "loss": 0.1103, "step": 3780 }, { "epoch": 2.0, "learning_rate": 1.363074963172596e-05, "loss": 0.171, "step": 3781 }, { "epoch": 2.0, "learning_rate": 1.3627554122851679e-05, "loss": 0.0938, "step": 3782 }, { "epoch": 2.0, "learning_rate": 1.3624358187370494e-05, "loss": 0.147, "step": 3783 }, { "epoch": 2.0, "learning_rate": 1.362116182565826e-05, "loss": 0.1347, "step": 3784 }, { "epoch": 2.0, "learning_rate": 1.361796503809087e-05, "loss": 0.1113, "step": 3785 }, { "epoch": 2.0, "learning_rate": 1.3614767825044275e-05, "loss": 0.1391, "step": 3786 }, { "epoch": 2.0, "learning_rate": 1.3611570186894469e-05, "loss": 0.1454, "step": 3787 }, { "epoch": 2.01, "learning_rate": 1.3608372124017506e-05, "loss": 0.0898, "step": 3788 }, { "epoch": 2.01, "learning_rate": 1.3605173636789479e-05, "loss": 0.1514, "step": 3789 }, { "epoch": 2.01, "learning_rate": 1.3601974725586539e-05, "loss": 0.1242, "step": 3790 }, { "epoch": 2.01, "learning_rate": 1.359877539078488e-05, "loss": 0.1625, "step": 3791 }, { "epoch": 2.01, "learning_rate": 1.359557563276075e-05, "loss": 0.1587, "step": 3792 }, { "epoch": 2.01, "learning_rate": 1.359237545189045e-05, "loss": 0.1195, "step": 3793 }, { "epoch": 2.01, "learning_rate": 1.3589174848550326e-05, "loss": 0.1758, "step": 3794 }, { "epoch": 2.01, "learning_rate": 1.3585973823116773e-05, "loss": 0.1327, "step": 3795 }, { "epoch": 2.01, "learning_rate": 1.3582772375966238e-05, "loss": 0.0897, "step": 3796 }, { "epoch": 2.01, "learning_rate": 1.3579570507475215e-05, "loss": 0.1498, "step": 3797 }, { "epoch": 2.01, "learning_rate": 1.357636821802025e-05, "loss": 0.115, "step": 3798 }, { "epoch": 2.01, "learning_rate": 1.3573165507977946e-05, "loss": 0.1443, "step": 3799 }, { "epoch": 2.01, "learning_rate": 1.3569962377724937e-05, "loss": 0.1761, "step": 3800 }, { "epoch": 2.01, "learning_rate": 1.3566758827637921e-05, "loss": 0.1433, "step": 3801 }, { "epoch": 2.01, "learning_rate": 1.3563554858093642e-05, "loss": 0.2216, "step": 3802 }, { "epoch": 2.01, "learning_rate": 1.356035046946889e-05, "loss": 0.1192, "step": 3803 }, { "epoch": 2.01, "learning_rate": 1.355714566214051e-05, "loss": 0.1715, "step": 3804 }, { "epoch": 2.01, "learning_rate": 1.355394043648539e-05, "loss": 0.1353, "step": 3805 }, { "epoch": 2.01, "learning_rate": 1.3550734792880473e-05, "loss": 0.1194, "step": 3806 }, { "epoch": 2.02, "learning_rate": 1.3547528731702749e-05, "loss": 0.1865, "step": 3807 }, { "epoch": 2.02, "learning_rate": 1.3544322253329254e-05, "loss": 0.1034, "step": 3808 }, { "epoch": 2.02, "learning_rate": 1.3541115358137076e-05, "loss": 0.0746, "step": 3809 }, { "epoch": 2.02, "learning_rate": 1.3537908046503352e-05, "loss": 0.2261, "step": 3810 }, { "epoch": 2.02, "learning_rate": 1.3534700318805272e-05, "loss": 0.1232, "step": 3811 }, { "epoch": 2.02, "learning_rate": 1.3531492175420063e-05, "loss": 0.1687, "step": 3812 }, { "epoch": 2.02, "learning_rate": 1.3528283616725008e-05, "loss": 0.199, "step": 3813 }, { "epoch": 2.02, "learning_rate": 1.3525074643097449e-05, "loss": 0.1204, "step": 3814 }, { "epoch": 2.02, "learning_rate": 1.3521865254914761e-05, "loss": 0.0974, "step": 3815 }, { "epoch": 2.02, "learning_rate": 1.3518655452554373e-05, "loss": 0.121, "step": 3816 }, { "epoch": 2.02, "learning_rate": 1.3515445236393766e-05, "loss": 0.1383, "step": 3817 }, { "epoch": 2.02, "learning_rate": 1.3512234606810462e-05, "loss": 0.1699, "step": 3818 }, { "epoch": 2.02, "learning_rate": 1.3509023564182044e-05, "loss": 0.1172, "step": 3819 }, { "epoch": 2.02, "learning_rate": 1.3505812108886131e-05, "loss": 0.1781, "step": 3820 }, { "epoch": 2.02, "learning_rate": 1.3502600241300398e-05, "loss": 0.1167, "step": 3821 }, { "epoch": 2.02, "learning_rate": 1.3499387961802569e-05, "loss": 0.1351, "step": 3822 }, { "epoch": 2.02, "learning_rate": 1.349617527077041e-05, "loss": 0.1543, "step": 3823 }, { "epoch": 2.02, "learning_rate": 1.349296216858174e-05, "loss": 0.1564, "step": 3824 }, { "epoch": 2.02, "learning_rate": 1.3489748655614428e-05, "loss": 0.0799, "step": 3825 }, { "epoch": 2.03, "learning_rate": 1.3486534732246383e-05, "loss": 0.1442, "step": 3826 }, { "epoch": 2.03, "learning_rate": 1.3483320398855577e-05, "loss": 0.1529, "step": 3827 }, { "epoch": 2.03, "learning_rate": 1.3480105655820017e-05, "loss": 0.1415, "step": 3828 }, { "epoch": 2.03, "learning_rate": 1.3476890503517757e-05, "loss": 0.1456, "step": 3829 }, { "epoch": 2.03, "learning_rate": 1.347367494232691e-05, "loss": 0.1509, "step": 3830 }, { "epoch": 2.03, "learning_rate": 1.347045897262564e-05, "loss": 0.1029, "step": 3831 }, { "epoch": 2.03, "learning_rate": 1.3467242594792136e-05, "loss": 0.2245, "step": 3832 }, { "epoch": 2.03, "learning_rate": 1.3464025809204658e-05, "loss": 0.1227, "step": 3833 }, { "epoch": 2.03, "learning_rate": 1.34608086162415e-05, "loss": 0.1425, "step": 3834 }, { "epoch": 2.03, "learning_rate": 1.345759101628102e-05, "loss": 0.156, "step": 3835 }, { "epoch": 2.03, "learning_rate": 1.3454373009701604e-05, "loss": 0.1267, "step": 3836 }, { "epoch": 2.03, "learning_rate": 1.3451154596881699e-05, "loss": 0.1291, "step": 3837 }, { "epoch": 2.03, "learning_rate": 1.3447935778199797e-05, "loss": 0.097, "step": 3838 }, { "epoch": 2.03, "learning_rate": 1.3444716554034436e-05, "loss": 0.1074, "step": 3839 }, { "epoch": 2.03, "learning_rate": 1.3441496924764201e-05, "loss": 0.1743, "step": 3840 }, { "epoch": 2.03, "learning_rate": 1.3438276890767727e-05, "loss": 0.1396, "step": 3841 }, { "epoch": 2.03, "learning_rate": 1.3435056452423698e-05, "loss": 0.1365, "step": 3842 }, { "epoch": 2.03, "learning_rate": 1.3431835610110844e-05, "loss": 0.1137, "step": 3843 }, { "epoch": 2.03, "learning_rate": 1.3428614364207935e-05, "loss": 0.286, "step": 3844 }, { "epoch": 2.04, "learning_rate": 1.3425392715093801e-05, "loss": 0.1294, "step": 3845 }, { "epoch": 2.04, "learning_rate": 1.3422170663147313e-05, "loss": 0.0842, "step": 3846 }, { "epoch": 2.04, "learning_rate": 1.3418948208747393e-05, "loss": 0.1434, "step": 3847 }, { "epoch": 2.04, "learning_rate": 1.3415725352273001e-05, "loss": 0.1847, "step": 3848 }, { "epoch": 2.04, "learning_rate": 1.3412502094103154e-05, "loss": 0.1404, "step": 3849 }, { "epoch": 2.04, "learning_rate": 1.340927843461691e-05, "loss": 0.0567, "step": 3850 }, { "epoch": 2.04, "learning_rate": 1.3406054374193385e-05, "loss": 0.1425, "step": 3851 }, { "epoch": 2.04, "learning_rate": 1.3402829913211727e-05, "loss": 0.152, "step": 3852 }, { "epoch": 2.04, "learning_rate": 1.3399605052051142e-05, "loss": 0.0976, "step": 3853 }, { "epoch": 2.04, "learning_rate": 1.3396379791090877e-05, "loss": 0.1428, "step": 3854 }, { "epoch": 2.04, "learning_rate": 1.3393154130710231e-05, "loss": 0.1608, "step": 3855 }, { "epoch": 2.04, "learning_rate": 1.3389928071288545e-05, "loss": 0.1957, "step": 3856 }, { "epoch": 2.04, "learning_rate": 1.338670161320521e-05, "loss": 0.0994, "step": 3857 }, { "epoch": 2.04, "learning_rate": 1.3383474756839665e-05, "loss": 0.1611, "step": 3858 }, { "epoch": 2.04, "learning_rate": 1.3380247502571393e-05, "loss": 0.148, "step": 3859 }, { "epoch": 2.04, "learning_rate": 1.3377019850779924e-05, "loss": 0.125, "step": 3860 }, { "epoch": 2.04, "learning_rate": 1.3373791801844838e-05, "loss": 0.171, "step": 3861 }, { "epoch": 2.04, "learning_rate": 1.3370563356145757e-05, "loss": 0.0764, "step": 3862 }, { "epoch": 2.04, "learning_rate": 1.3367334514062356e-05, "loss": 0.107, "step": 3863 }, { "epoch": 2.05, "learning_rate": 1.3364105275974346e-05, "loss": 0.1016, "step": 3864 }, { "epoch": 2.05, "learning_rate": 1.3360875642261495e-05, "loss": 0.0826, "step": 3865 }, { "epoch": 2.05, "learning_rate": 1.3357645613303616e-05, "loss": 0.1597, "step": 3866 }, { "epoch": 2.05, "learning_rate": 1.3354415189480564e-05, "loss": 0.1245, "step": 3867 }, { "epoch": 2.05, "learning_rate": 1.335118437117224e-05, "loss": 0.0992, "step": 3868 }, { "epoch": 2.05, "learning_rate": 1.3347953158758597e-05, "loss": 0.1682, "step": 3869 }, { "epoch": 2.05, "learning_rate": 1.3344721552619633e-05, "loss": 0.1273, "step": 3870 }, { "epoch": 2.05, "learning_rate": 1.3341489553135385e-05, "loss": 0.1187, "step": 3871 }, { "epoch": 2.05, "learning_rate": 1.3338257160685948e-05, "loss": 0.1209, "step": 3872 }, { "epoch": 2.05, "learning_rate": 1.3335024375651456e-05, "loss": 0.0786, "step": 3873 }, { "epoch": 2.05, "learning_rate": 1.3331791198412088e-05, "loss": 0.1512, "step": 3874 }, { "epoch": 2.05, "learning_rate": 1.332855762934807e-05, "loss": 0.1593, "step": 3875 }, { "epoch": 2.05, "learning_rate": 1.3325323668839682e-05, "loss": 0.1714, "step": 3876 }, { "epoch": 2.05, "learning_rate": 1.332208931726724e-05, "loss": 0.1329, "step": 3877 }, { "epoch": 2.05, "learning_rate": 1.3318854575011106e-05, "loss": 0.1158, "step": 3878 }, { "epoch": 2.05, "learning_rate": 1.3315619442451699e-05, "loss": 0.1863, "step": 3879 }, { "epoch": 2.05, "learning_rate": 1.3312383919969467e-05, "loss": 0.1449, "step": 3880 }, { "epoch": 2.05, "learning_rate": 1.330914800794492e-05, "loss": 0.1051, "step": 3881 }, { "epoch": 2.05, "learning_rate": 1.3305911706758607e-05, "loss": 0.103, "step": 3882 }, { "epoch": 2.06, "learning_rate": 1.3302675016791123e-05, "loss": 0.1129, "step": 3883 }, { "epoch": 2.06, "learning_rate": 1.3299437938423106e-05, "loss": 0.1412, "step": 3884 }, { "epoch": 2.06, "learning_rate": 1.329620047203524e-05, "loss": 0.1791, "step": 3885 }, { "epoch": 2.06, "learning_rate": 1.3292962618008263e-05, "loss": 0.135, "step": 3886 }, { "epoch": 2.06, "learning_rate": 1.3289724376722951e-05, "loss": 0.171, "step": 3887 }, { "epoch": 2.06, "learning_rate": 1.3286485748560125e-05, "loss": 0.1035, "step": 3888 }, { "epoch": 2.06, "learning_rate": 1.3283246733900653e-05, "loss": 0.1221, "step": 3889 }, { "epoch": 2.06, "learning_rate": 1.3280007333125448e-05, "loss": 0.1171, "step": 3890 }, { "epoch": 2.06, "learning_rate": 1.3276767546615475e-05, "loss": 0.1381, "step": 3891 }, { "epoch": 2.06, "learning_rate": 1.3273527374751733e-05, "loss": 0.1359, "step": 3892 }, { "epoch": 2.06, "learning_rate": 1.3270286817915277e-05, "loss": 0.14, "step": 3893 }, { "epoch": 2.06, "learning_rate": 1.3267045876487197e-05, "loss": 0.0875, "step": 3894 }, { "epoch": 2.06, "learning_rate": 1.3263804550848639e-05, "loss": 0.1539, "step": 3895 }, { "epoch": 2.06, "learning_rate": 1.326056284138078e-05, "loss": 0.125, "step": 3896 }, { "epoch": 2.06, "learning_rate": 1.325732074846486e-05, "loss": 0.118, "step": 3897 }, { "epoch": 2.06, "learning_rate": 1.3254078272482151e-05, "loss": 0.1052, "step": 3898 }, { "epoch": 2.06, "learning_rate": 1.3250835413813978e-05, "loss": 0.1275, "step": 3899 }, { "epoch": 2.06, "learning_rate": 1.3247592172841703e-05, "loss": 0.1206, "step": 3900 }, { "epoch": 2.06, "learning_rate": 1.3244348549946735e-05, "loss": 0.1693, "step": 3901 }, { "epoch": 2.07, "learning_rate": 1.3241104545510533e-05, "loss": 0.1565, "step": 3902 }, { "epoch": 2.07, "learning_rate": 1.3237860159914599e-05, "loss": 0.1206, "step": 3903 }, { "epoch": 2.07, "learning_rate": 1.3234615393540477e-05, "loss": 0.1748, "step": 3904 }, { "epoch": 2.07, "learning_rate": 1.3231370246769757e-05, "loss": 0.134, "step": 3905 }, { "epoch": 2.07, "learning_rate": 1.3228124719984075e-05, "loss": 0.1482, "step": 3906 }, { "epoch": 2.07, "learning_rate": 1.3224878813565108e-05, "loss": 0.178, "step": 3907 }, { "epoch": 2.07, "learning_rate": 1.3221632527894583e-05, "loss": 0.1625, "step": 3908 }, { "epoch": 2.07, "learning_rate": 1.321838586335427e-05, "loss": 0.1466, "step": 3909 }, { "epoch": 2.07, "learning_rate": 1.321513882032598e-05, "loss": 0.1143, "step": 3910 }, { "epoch": 2.07, "learning_rate": 1.3211891399191575e-05, "loss": 0.1826, "step": 3911 }, { "epoch": 2.07, "learning_rate": 1.3208643600332952e-05, "loss": 0.1356, "step": 3912 }, { "epoch": 2.07, "learning_rate": 1.3205395424132062e-05, "loss": 0.1126, "step": 3913 }, { "epoch": 2.07, "learning_rate": 1.3202146870970896e-05, "loss": 0.0832, "step": 3914 }, { "epoch": 2.07, "learning_rate": 1.319889794123149e-05, "loss": 0.1218, "step": 3915 }, { "epoch": 2.07, "learning_rate": 1.3195648635295921e-05, "loss": 0.2385, "step": 3916 }, { "epoch": 2.07, "learning_rate": 1.3192398953546314e-05, "loss": 0.1461, "step": 3917 }, { "epoch": 2.07, "learning_rate": 1.3189148896364842e-05, "loss": 0.1635, "step": 3918 }, { "epoch": 2.07, "learning_rate": 1.3185898464133716e-05, "loss": 0.059, "step": 3919 }, { "epoch": 2.07, "learning_rate": 1.3182647657235188e-05, "loss": 0.1027, "step": 3920 }, { "epoch": 2.08, "learning_rate": 1.3179396476051563e-05, "loss": 0.1986, "step": 3921 }, { "epoch": 2.08, "learning_rate": 1.317614492096518e-05, "loss": 0.1574, "step": 3922 }, { "epoch": 2.08, "learning_rate": 1.3172892992358438e-05, "loss": 0.1659, "step": 3923 }, { "epoch": 2.08, "learning_rate": 1.3169640690613761e-05, "loss": 0.1227, "step": 3924 }, { "epoch": 2.08, "learning_rate": 1.316638801611363e-05, "loss": 0.1175, "step": 3925 }, { "epoch": 2.08, "learning_rate": 1.3163134969240563e-05, "loss": 0.1504, "step": 3926 }, { "epoch": 2.08, "learning_rate": 1.3159881550377126e-05, "loss": 0.1445, "step": 3927 }, { "epoch": 2.08, "learning_rate": 1.3156627759905923e-05, "loss": 0.1641, "step": 3928 }, { "epoch": 2.08, "learning_rate": 1.315337359820961e-05, "loss": 0.0842, "step": 3929 }, { "epoch": 2.08, "learning_rate": 1.3150119065670881e-05, "loss": 0.14, "step": 3930 }, { "epoch": 2.08, "learning_rate": 1.3146864162672475e-05, "loss": 0.0901, "step": 3931 }, { "epoch": 2.08, "learning_rate": 1.3143608889597175e-05, "loss": 0.0968, "step": 3932 }, { "epoch": 2.08, "learning_rate": 1.3140353246827803e-05, "loss": 0.1483, "step": 3933 }, { "epoch": 2.08, "learning_rate": 1.3137097234747236e-05, "loss": 0.0691, "step": 3934 }, { "epoch": 2.08, "learning_rate": 1.313384085373838e-05, "loss": 0.169, "step": 3935 }, { "epoch": 2.08, "learning_rate": 1.3130584104184196e-05, "loss": 0.1559, "step": 3936 }, { "epoch": 2.08, "learning_rate": 1.3127326986467679e-05, "loss": 0.1077, "step": 3937 }, { "epoch": 2.08, "learning_rate": 1.3124069500971874e-05, "loss": 0.1161, "step": 3938 }, { "epoch": 2.08, "learning_rate": 1.3120811648079868e-05, "loss": 0.1864, "step": 3939 }, { "epoch": 2.09, "learning_rate": 1.3117553428174792e-05, "loss": 0.117, "step": 3940 }, { "epoch": 2.09, "learning_rate": 1.3114294841639813e-05, "loss": 0.1033, "step": 3941 }, { "epoch": 2.09, "learning_rate": 1.3111035888858153e-05, "loss": 0.0949, "step": 3942 }, { "epoch": 2.09, "learning_rate": 1.3107776570213067e-05, "loss": 0.0895, "step": 3943 }, { "epoch": 2.09, "learning_rate": 1.3104516886087854e-05, "loss": 0.1938, "step": 3944 }, { "epoch": 2.09, "learning_rate": 1.3101256836865866e-05, "loss": 0.1357, "step": 3945 }, { "epoch": 2.09, "learning_rate": 1.3097996422930486e-05, "loss": 0.101, "step": 3946 }, { "epoch": 2.09, "learning_rate": 1.3094735644665144e-05, "loss": 0.1409, "step": 3947 }, { "epoch": 2.09, "learning_rate": 1.3091474502453316e-05, "loss": 0.0888, "step": 3948 }, { "epoch": 2.09, "learning_rate": 1.3088212996678518e-05, "loss": 0.1219, "step": 3949 }, { "epoch": 2.09, "learning_rate": 1.3084951127724305e-05, "loss": 0.1134, "step": 3950 }, { "epoch": 2.09, "learning_rate": 1.3081688895974285e-05, "loss": 0.1359, "step": 3951 }, { "epoch": 2.09, "learning_rate": 1.30784263018121e-05, "loss": 0.1432, "step": 3952 }, { "epoch": 2.09, "learning_rate": 1.3075163345621433e-05, "loss": 0.2018, "step": 3953 }, { "epoch": 2.09, "learning_rate": 1.3071900027786016e-05, "loss": 0.2426, "step": 3954 }, { "epoch": 2.09, "learning_rate": 1.3068636348689621e-05, "loss": 0.1062, "step": 3955 }, { "epoch": 2.09, "learning_rate": 1.3065372308716067e-05, "loss": 0.1243, "step": 3956 }, { "epoch": 2.09, "learning_rate": 1.3062107908249205e-05, "loss": 0.1586, "step": 3957 }, { "epoch": 2.1, "learning_rate": 1.3058843147672936e-05, "loss": 0.1722, "step": 3958 }, { "epoch": 2.1, "learning_rate": 1.3055578027371198e-05, "loss": 0.1054, "step": 3959 }, { "epoch": 2.1, "learning_rate": 1.3052312547727985e-05, "loss": 0.1527, "step": 3960 }, { "epoch": 2.1, "learning_rate": 1.3049046709127316e-05, "loss": 0.1144, "step": 3961 }, { "epoch": 2.1, "learning_rate": 1.3045780511953261e-05, "loss": 0.1733, "step": 3962 }, { "epoch": 2.1, "learning_rate": 1.3042513956589928e-05, "loss": 0.1267, "step": 3963 }, { "epoch": 2.1, "learning_rate": 1.3039247043421474e-05, "loss": 0.1175, "step": 3964 }, { "epoch": 2.1, "learning_rate": 1.3035979772832095e-05, "loss": 0.0841, "step": 3965 }, { "epoch": 2.1, "learning_rate": 1.303271214520602e-05, "loss": 0.1659, "step": 3966 }, { "epoch": 2.1, "learning_rate": 1.3029444160927539e-05, "loss": 0.0743, "step": 3967 }, { "epoch": 2.1, "learning_rate": 1.3026175820380965e-05, "loss": 0.13, "step": 3968 }, { "epoch": 2.1, "learning_rate": 1.3022907123950659e-05, "loss": 0.1276, "step": 3969 }, { "epoch": 2.1, "learning_rate": 1.3019638072021032e-05, "loss": 0.1383, "step": 3970 }, { "epoch": 2.1, "learning_rate": 1.3016368664976526e-05, "loss": 0.1253, "step": 3971 }, { "epoch": 2.1, "learning_rate": 1.3013098903201635e-05, "loss": 0.1844, "step": 3972 }, { "epoch": 2.1, "learning_rate": 1.3009828787080882e-05, "loss": 0.0851, "step": 3973 }, { "epoch": 2.1, "learning_rate": 1.300655831699884e-05, "loss": 0.0937, "step": 3974 }, { "epoch": 2.1, "learning_rate": 1.3003287493340124e-05, "loss": 0.1392, "step": 3975 }, { "epoch": 2.1, "learning_rate": 1.300001631648939e-05, "loss": 0.128, "step": 3976 }, { "epoch": 2.11, "learning_rate": 1.2996744786831332e-05, "loss": 0.2165, "step": 3977 }, { "epoch": 2.11, "learning_rate": 1.2993472904750686e-05, "loss": 0.1519, "step": 3978 }, { "epoch": 2.11, "learning_rate": 1.2990200670632235e-05, "loss": 0.1384, "step": 3979 }, { "epoch": 2.11, "learning_rate": 1.29869280848608e-05, "loss": 0.1585, "step": 3980 }, { "epoch": 2.11, "learning_rate": 1.2983655147821236e-05, "loss": 0.1419, "step": 3981 }, { "epoch": 2.11, "learning_rate": 1.2980381859898454e-05, "loss": 0.2438, "step": 3982 }, { "epoch": 2.11, "learning_rate": 1.2977108221477396e-05, "loss": 0.1186, "step": 3983 }, { "epoch": 2.11, "learning_rate": 1.297383423294305e-05, "loss": 0.1281, "step": 3984 }, { "epoch": 2.11, "learning_rate": 1.297055989468044e-05, "loss": 0.0666, "step": 3985 }, { "epoch": 2.11, "learning_rate": 1.2967285207074632e-05, "loss": 0.1046, "step": 3986 }, { "epoch": 2.11, "learning_rate": 1.296401017051074e-05, "loss": 0.1748, "step": 3987 }, { "epoch": 2.11, "learning_rate": 1.2960734785373911e-05, "loss": 0.0951, "step": 3988 }, { "epoch": 2.11, "learning_rate": 1.295745905204934e-05, "loss": 0.0863, "step": 3989 }, { "epoch": 2.11, "learning_rate": 1.2954182970922249e-05, "loss": 0.0907, "step": 3990 }, { "epoch": 2.11, "learning_rate": 1.2950906542377923e-05, "loss": 0.0993, "step": 3991 }, { "epoch": 2.11, "learning_rate": 1.2947629766801675e-05, "loss": 0.1411, "step": 3992 }, { "epoch": 2.11, "learning_rate": 1.2944352644578852e-05, "loss": 0.0789, "step": 3993 }, { "epoch": 2.11, "learning_rate": 1.2941075176094855e-05, "loss": 0.1078, "step": 3994 }, { "epoch": 2.11, "learning_rate": 1.2937797361735119e-05, "loss": 0.1765, "step": 3995 }, { "epoch": 2.12, "learning_rate": 1.2934519201885119e-05, "loss": 0.1436, "step": 3996 }, { "epoch": 2.12, "learning_rate": 1.2931240696930374e-05, "loss": 0.0919, "step": 3997 }, { "epoch": 2.12, "learning_rate": 1.2927961847256443e-05, "loss": 0.111, "step": 3998 }, { "epoch": 2.12, "learning_rate": 1.2924682653248923e-05, "loss": 0.1005, "step": 3999 }, { "epoch": 2.12, "learning_rate": 1.2921403115293455e-05, "loss": 0.1348, "step": 4000 }, { "epoch": 2.12, "learning_rate": 1.2918123233775717e-05, "loss": 0.1231, "step": 4001 }, { "epoch": 2.12, "learning_rate": 1.291484300908143e-05, "loss": 0.0736, "step": 4002 }, { "epoch": 2.12, "learning_rate": 1.2911562441596354e-05, "loss": 0.1209, "step": 4003 }, { "epoch": 2.12, "learning_rate": 1.290828153170629e-05, "loss": 0.1222, "step": 4004 }, { "epoch": 2.12, "learning_rate": 1.2905000279797077e-05, "loss": 0.1776, "step": 4005 }, { "epoch": 2.12, "learning_rate": 1.2901718686254598e-05, "loss": 0.1709, "step": 4006 }, { "epoch": 2.12, "learning_rate": 1.2898436751464773e-05, "loss": 0.0978, "step": 4007 }, { "epoch": 2.12, "learning_rate": 1.2895154475813567e-05, "loss": 0.0986, "step": 4008 }, { "epoch": 2.12, "learning_rate": 1.2891871859686977e-05, "loss": 0.0561, "step": 4009 }, { "epoch": 2.12, "learning_rate": 1.2888588903471047e-05, "loss": 0.1337, "step": 4010 }, { "epoch": 2.12, "learning_rate": 1.2885305607551856e-05, "loss": 0.104, "step": 4011 }, { "epoch": 2.12, "learning_rate": 1.2882021972315531e-05, "loss": 0.124, "step": 4012 }, { "epoch": 2.12, "learning_rate": 1.2878737998148228e-05, "loss": 0.1473, "step": 4013 }, { "epoch": 2.12, "learning_rate": 1.287545368543615e-05, "loss": 0.1158, "step": 4014 }, { "epoch": 2.13, "learning_rate": 1.2872169034565536e-05, "loss": 0.1816, "step": 4015 }, { "epoch": 2.13, "learning_rate": 1.2868884045922671e-05, "loss": 0.2043, "step": 4016 }, { "epoch": 2.13, "learning_rate": 1.2865598719893875e-05, "loss": 0.1321, "step": 4017 }, { "epoch": 2.13, "learning_rate": 1.2862313056865507e-05, "loss": 0.1138, "step": 4018 }, { "epoch": 2.13, "learning_rate": 1.2859027057223964e-05, "loss": 0.1093, "step": 4019 }, { "epoch": 2.13, "learning_rate": 1.2855740721355695e-05, "loss": 0.0679, "step": 4020 }, { "epoch": 2.13, "learning_rate": 1.2852454049647163e-05, "loss": 0.0644, "step": 4021 }, { "epoch": 2.13, "learning_rate": 1.2849167042484903e-05, "loss": 0.1928, "step": 4022 }, { "epoch": 2.13, "learning_rate": 1.284587970025546e-05, "loss": 0.1888, "step": 4023 }, { "epoch": 2.13, "learning_rate": 1.2842592023345444e-05, "loss": 0.1108, "step": 4024 }, { "epoch": 2.13, "learning_rate": 1.283930401214148e-05, "loss": 0.1076, "step": 4025 }, { "epoch": 2.13, "learning_rate": 1.283601566703025e-05, "loss": 0.1474, "step": 4026 }, { "epoch": 2.13, "learning_rate": 1.2832726988398468e-05, "loss": 0.0903, "step": 4027 }, { "epoch": 2.13, "learning_rate": 1.2829437976632892e-05, "loss": 0.1063, "step": 4028 }, { "epoch": 2.13, "learning_rate": 1.282614863212031e-05, "loss": 0.1413, "step": 4029 }, { "epoch": 2.13, "learning_rate": 1.2822858955247556e-05, "loss": 0.1597, "step": 4030 }, { "epoch": 2.13, "learning_rate": 1.2819568946401505e-05, "loss": 0.0964, "step": 4031 }, { "epoch": 2.13, "learning_rate": 1.2816278605969066e-05, "loss": 0.2378, "step": 4032 }, { "epoch": 2.13, "learning_rate": 1.2812987934337189e-05, "loss": 0.1072, "step": 4033 }, { "epoch": 2.14, "learning_rate": 1.2809696931892863e-05, "loss": 0.1303, "step": 4034 }, { "epoch": 2.14, "learning_rate": 1.2806405599023117e-05, "loss": 0.1151, "step": 4035 }, { "epoch": 2.14, "learning_rate": 1.2803113936115014e-05, "loss": 0.0978, "step": 4036 }, { "epoch": 2.14, "learning_rate": 1.2799821943555663e-05, "loss": 0.0576, "step": 4037 }, { "epoch": 2.14, "learning_rate": 1.279652962173221e-05, "loss": 0.175, "step": 4038 }, { "epoch": 2.14, "learning_rate": 1.2793236971031833e-05, "loss": 0.0919, "step": 4039 }, { "epoch": 2.14, "learning_rate": 1.278994399184176e-05, "loss": 0.1341, "step": 4040 }, { "epoch": 2.14, "learning_rate": 1.2786650684549242e-05, "loss": 0.2184, "step": 4041 }, { "epoch": 2.14, "learning_rate": 1.2783357049541584e-05, "loss": 0.1409, "step": 4042 }, { "epoch": 2.14, "learning_rate": 1.2780063087206123e-05, "loss": 0.1241, "step": 4043 }, { "epoch": 2.14, "learning_rate": 1.2776768797930237e-05, "loss": 0.1117, "step": 4044 }, { "epoch": 2.14, "learning_rate": 1.2773474182101336e-05, "loss": 0.1239, "step": 4045 }, { "epoch": 2.14, "learning_rate": 1.2770179240106875e-05, "loss": 0.1337, "step": 4046 }, { "epoch": 2.14, "learning_rate": 1.2766883972334343e-05, "loss": 0.1395, "step": 4047 }, { "epoch": 2.14, "learning_rate": 1.2763588379171273e-05, "loss": 0.2369, "step": 4048 }, { "epoch": 2.14, "learning_rate": 1.276029246100523e-05, "loss": 0.135, "step": 4049 }, { "epoch": 2.14, "learning_rate": 1.2756996218223821e-05, "loss": 0.1533, "step": 4050 }, { "epoch": 2.14, "learning_rate": 1.2753699651214688e-05, "loss": 0.1567, "step": 4051 }, { "epoch": 2.14, "learning_rate": 1.2750402760365517e-05, "loss": 0.0846, "step": 4052 }, { "epoch": 2.15, "learning_rate": 1.2747105546064027e-05, "loss": 0.1325, "step": 4053 }, { "epoch": 2.15, "learning_rate": 1.2743808008697975e-05, "loss": 0.0961, "step": 4054 }, { "epoch": 2.15, "learning_rate": 1.2740510148655156e-05, "loss": 0.0794, "step": 4055 }, { "epoch": 2.15, "learning_rate": 1.2737211966323408e-05, "loss": 0.1345, "step": 4056 }, { "epoch": 2.15, "learning_rate": 1.2733913462090602e-05, "loss": 0.1111, "step": 4057 }, { "epoch": 2.15, "learning_rate": 1.2730614636344647e-05, "loss": 0.1307, "step": 4058 }, { "epoch": 2.15, "learning_rate": 1.2727315489473491e-05, "loss": 0.1794, "step": 4059 }, { "epoch": 2.15, "learning_rate": 1.272401602186512e-05, "loss": 0.0919, "step": 4060 }, { "epoch": 2.15, "learning_rate": 1.272071623390756e-05, "loss": 0.1805, "step": 4061 }, { "epoch": 2.15, "learning_rate": 1.2717416125988868e-05, "loss": 0.1405, "step": 4062 }, { "epoch": 2.15, "learning_rate": 1.271411569849714e-05, "loss": 0.091, "step": 4063 }, { "epoch": 2.15, "learning_rate": 1.271081495182052e-05, "loss": 0.2481, "step": 4064 }, { "epoch": 2.15, "learning_rate": 1.2707513886347177e-05, "loss": 0.1097, "step": 4065 }, { "epoch": 2.15, "learning_rate": 1.2704212502465325e-05, "loss": 0.1661, "step": 4066 }, { "epoch": 2.15, "learning_rate": 1.2700910800563208e-05, "loss": 0.1389, "step": 4067 }, { "epoch": 2.15, "learning_rate": 1.2697608781029117e-05, "loss": 0.1143, "step": 4068 }, { "epoch": 2.15, "learning_rate": 1.2694306444251372e-05, "loss": 0.207, "step": 4069 }, { "epoch": 2.15, "learning_rate": 1.2691003790618335e-05, "loss": 0.2017, "step": 4070 }, { "epoch": 2.15, "learning_rate": 1.2687700820518404e-05, "loss": 0.1153, "step": 4071 }, { "epoch": 2.16, "learning_rate": 1.2684397534340018e-05, "loss": 0.1149, "step": 4072 }, { "epoch": 2.16, "learning_rate": 1.2681093932471642e-05, "loss": 0.2238, "step": 4073 }, { "epoch": 2.16, "learning_rate": 1.2677790015301793e-05, "loss": 0.1234, "step": 4074 }, { "epoch": 2.16, "learning_rate": 1.267448578321901e-05, "loss": 0.1476, "step": 4075 }, { "epoch": 2.16, "learning_rate": 1.267118123661188e-05, "loss": 0.1102, "step": 4076 }, { "epoch": 2.16, "learning_rate": 1.2667876375869031e-05, "loss": 0.108, "step": 4077 }, { "epoch": 2.16, "learning_rate": 1.2664571201379108e-05, "loss": 0.0652, "step": 4078 }, { "epoch": 2.16, "learning_rate": 1.266126571353081e-05, "loss": 0.1121, "step": 4079 }, { "epoch": 2.16, "learning_rate": 1.2657959912712872e-05, "loss": 0.1269, "step": 4080 }, { "epoch": 2.16, "learning_rate": 1.2654653799314063e-05, "loss": 0.1202, "step": 4081 }, { "epoch": 2.16, "learning_rate": 1.2651347373723182e-05, "loss": 0.1999, "step": 4082 }, { "epoch": 2.16, "learning_rate": 1.2648040636329073e-05, "loss": 0.143, "step": 4083 }, { "epoch": 2.16, "learning_rate": 1.2644733587520612e-05, "loss": 0.2167, "step": 4084 }, { "epoch": 2.16, "learning_rate": 1.264142622768672e-05, "loss": 0.1193, "step": 4085 }, { "epoch": 2.16, "learning_rate": 1.2638118557216346e-05, "loss": 0.1144, "step": 4086 }, { "epoch": 2.16, "learning_rate": 1.2634810576498473e-05, "loss": 0.1024, "step": 4087 }, { "epoch": 2.16, "learning_rate": 1.263150228592213e-05, "loss": 0.1496, "step": 4088 }, { "epoch": 2.16, "learning_rate": 1.262819368587638e-05, "loss": 0.1046, "step": 4089 }, { "epoch": 2.16, "learning_rate": 1.2624884776750315e-05, "loss": 0.169, "step": 4090 }, { "epoch": 2.17, "learning_rate": 1.2621575558933072e-05, "loss": 0.1449, "step": 4091 }, { "epoch": 2.17, "learning_rate": 1.2618266032813822e-05, "loss": 0.1836, "step": 4092 }, { "epoch": 2.17, "learning_rate": 1.261495619878177e-05, "loss": 0.1174, "step": 4093 }, { "epoch": 2.17, "learning_rate": 1.2611646057226155e-05, "loss": 0.0918, "step": 4094 }, { "epoch": 2.17, "learning_rate": 1.260833560853626e-05, "loss": 0.1439, "step": 4095 }, { "epoch": 2.17, "learning_rate": 1.2605024853101397e-05, "loss": 0.0911, "step": 4096 }, { "epoch": 2.17, "learning_rate": 1.2601713791310921e-05, "loss": 0.1471, "step": 4097 }, { "epoch": 2.17, "learning_rate": 1.2598402423554214e-05, "loss": 0.1201, "step": 4098 }, { "epoch": 2.17, "learning_rate": 1.2595090750220702e-05, "loss": 0.1494, "step": 4099 }, { "epoch": 2.17, "learning_rate": 1.2591778771699841e-05, "loss": 0.1511, "step": 4100 }, { "epoch": 2.17, "learning_rate": 1.2588466488381129e-05, "loss": 0.1295, "step": 4101 }, { "epoch": 2.17, "learning_rate": 1.2585153900654094e-05, "loss": 0.1375, "step": 4102 }, { "epoch": 2.17, "learning_rate": 1.2581841008908302e-05, "loss": 0.1639, "step": 4103 }, { "epoch": 2.17, "learning_rate": 1.2578527813533357e-05, "loss": 0.1706, "step": 4104 }, { "epoch": 2.17, "learning_rate": 1.2575214314918896e-05, "loss": 0.1467, "step": 4105 }, { "epoch": 2.17, "learning_rate": 1.2571900513454589e-05, "loss": 0.1209, "step": 4106 }, { "epoch": 2.17, "learning_rate": 1.256858640953015e-05, "loss": 0.0979, "step": 4107 }, { "epoch": 2.17, "learning_rate": 1.2565272003535322e-05, "loss": 0.1147, "step": 4108 }, { "epoch": 2.17, "learning_rate": 1.2561957295859886e-05, "loss": 0.1377, "step": 4109 }, { "epoch": 2.18, "learning_rate": 1.2558642286893656e-05, "loss": 0.1257, "step": 4110 }, { "epoch": 2.18, "learning_rate": 1.2555326977026483e-05, "loss": 0.1203, "step": 4111 }, { "epoch": 2.18, "learning_rate": 1.2552011366648253e-05, "loss": 0.0951, "step": 4112 }, { "epoch": 2.18, "learning_rate": 1.2548695456148891e-05, "loss": 0.1419, "step": 4113 }, { "epoch": 2.18, "learning_rate": 1.2545379245918352e-05, "loss": 0.1185, "step": 4114 }, { "epoch": 2.18, "learning_rate": 1.2542062736346625e-05, "loss": 0.1466, "step": 4115 }, { "epoch": 2.18, "learning_rate": 1.2538745927823741e-05, "loss": 0.1323, "step": 4116 }, { "epoch": 2.18, "learning_rate": 1.2535428820739766e-05, "loss": 0.1706, "step": 4117 }, { "epoch": 2.18, "learning_rate": 1.2532111415484792e-05, "loss": 0.1304, "step": 4118 }, { "epoch": 2.18, "learning_rate": 1.252879371244895e-05, "loss": 0.1739, "step": 4119 }, { "epoch": 2.18, "learning_rate": 1.2525475712022415e-05, "loss": 0.1596, "step": 4120 }, { "epoch": 2.18, "learning_rate": 1.2522157414595386e-05, "loss": 0.1245, "step": 4121 }, { "epoch": 2.18, "learning_rate": 1.25188388205581e-05, "loss": 0.1421, "step": 4122 }, { "epoch": 2.18, "learning_rate": 1.2515519930300832e-05, "loss": 0.1609, "step": 4123 }, { "epoch": 2.18, "learning_rate": 1.251220074421389e-05, "loss": 0.1053, "step": 4124 }, { "epoch": 2.18, "learning_rate": 1.2508881262687612e-05, "loss": 0.1543, "step": 4125 }, { "epoch": 2.18, "learning_rate": 1.2505561486112377e-05, "loss": 0.2353, "step": 4126 }, { "epoch": 2.18, "learning_rate": 1.25022414148786e-05, "loss": 0.2486, "step": 4127 }, { "epoch": 2.18, "learning_rate": 1.2498921049376722e-05, "loss": 0.2263, "step": 4128 }, { "epoch": 2.19, "learning_rate": 1.2495600389997231e-05, "loss": 0.0638, "step": 4129 }, { "epoch": 2.19, "learning_rate": 1.2492279437130632e-05, "loss": 0.1634, "step": 4130 }, { "epoch": 2.19, "learning_rate": 1.2488958191167485e-05, "loss": 0.1239, "step": 4131 }, { "epoch": 2.19, "learning_rate": 1.248563665249837e-05, "loss": 0.1626, "step": 4132 }, { "epoch": 2.19, "learning_rate": 1.2482314821513909e-05, "loss": 0.1921, "step": 4133 }, { "epoch": 2.19, "learning_rate": 1.2478992698604748e-05, "loss": 0.098, "step": 4134 }, { "epoch": 2.19, "learning_rate": 1.2475670284161583e-05, "loss": 0.1128, "step": 4135 }, { "epoch": 2.19, "learning_rate": 1.247234757857513e-05, "loss": 0.0938, "step": 4136 }, { "epoch": 2.19, "learning_rate": 1.2469024582236151e-05, "loss": 0.0753, "step": 4137 }, { "epoch": 2.19, "learning_rate": 1.246570129553543e-05, "loss": 0.1785, "step": 4138 }, { "epoch": 2.19, "learning_rate": 1.2462377718863796e-05, "loss": 0.1998, "step": 4139 }, { "epoch": 2.19, "learning_rate": 1.2459053852612106e-05, "loss": 0.0999, "step": 4140 }, { "epoch": 2.19, "learning_rate": 1.2455729697171253e-05, "loss": 0.0928, "step": 4141 }, { "epoch": 2.19, "learning_rate": 1.2452405252932163e-05, "loss": 0.122, "step": 4142 }, { "epoch": 2.19, "learning_rate": 1.2449080520285796e-05, "loss": 0.1681, "step": 4143 }, { "epoch": 2.19, "learning_rate": 1.2445755499623149e-05, "loss": 0.1012, "step": 4144 }, { "epoch": 2.19, "learning_rate": 1.2442430191335251e-05, "loss": 0.1301, "step": 4145 }, { "epoch": 2.19, "learning_rate": 1.2439104595813158e-05, "loss": 0.1012, "step": 4146 }, { "epoch": 2.2, "learning_rate": 1.2435778713447973e-05, "loss": 0.186, "step": 4147 }, { "epoch": 2.2, "learning_rate": 1.2432452544630822e-05, "loss": 0.1356, "step": 4148 }, { "epoch": 2.2, "learning_rate": 1.2429126089752872e-05, "loss": 0.1513, "step": 4149 }, { "epoch": 2.2, "learning_rate": 1.2425799349205314e-05, "loss": 0.1691, "step": 4150 }, { "epoch": 2.2, "learning_rate": 1.2422472323379385e-05, "loss": 0.1435, "step": 4151 }, { "epoch": 2.2, "learning_rate": 1.2419145012666342e-05, "loss": 0.1296, "step": 4152 }, { "epoch": 2.2, "learning_rate": 1.2415817417457495e-05, "loss": 0.2134, "step": 4153 }, { "epoch": 2.2, "learning_rate": 1.2412489538144162e-05, "loss": 0.1996, "step": 4154 }, { "epoch": 2.2, "learning_rate": 1.2409161375117714e-05, "loss": 0.1036, "step": 4155 }, { "epoch": 2.2, "learning_rate": 1.2405832928769548e-05, "loss": 0.1418, "step": 4156 }, { "epoch": 2.2, "learning_rate": 1.2402504199491095e-05, "loss": 0.089, "step": 4157 }, { "epoch": 2.2, "learning_rate": 1.239917518767382e-05, "loss": 0.1377, "step": 4158 }, { "epoch": 2.2, "learning_rate": 1.2395845893709222e-05, "loss": 0.1089, "step": 4159 }, { "epoch": 2.2, "learning_rate": 1.2392516317988827e-05, "loss": 0.0701, "step": 4160 }, { "epoch": 2.2, "learning_rate": 1.2389186460904206e-05, "loss": 0.1168, "step": 4161 }, { "epoch": 2.2, "learning_rate": 1.2385856322846953e-05, "loss": 0.0555, "step": 4162 }, { "epoch": 2.2, "learning_rate": 1.2382525904208698e-05, "loss": 0.1762, "step": 4163 }, { "epoch": 2.2, "learning_rate": 1.2379195205381103e-05, "loss": 0.0625, "step": 4164 }, { "epoch": 2.2, "learning_rate": 1.2375864226755868e-05, "loss": 0.2069, "step": 4165 }, { "epoch": 2.21, "learning_rate": 1.2372532968724717e-05, "loss": 0.1869, "step": 4166 }, { "epoch": 2.21, "learning_rate": 1.2369201431679414e-05, "loss": 0.1559, "step": 4167 }, { "epoch": 2.21, "learning_rate": 1.2365869616011756e-05, "loss": 0.1444, "step": 4168 }, { "epoch": 2.21, "learning_rate": 1.2362537522113572e-05, "loss": 0.108, "step": 4169 }, { "epoch": 2.21, "learning_rate": 1.2359205150376714e-05, "loss": 0.1444, "step": 4170 }, { "epoch": 2.21, "learning_rate": 1.2355872501193083e-05, "loss": 0.0938, "step": 4171 }, { "epoch": 2.21, "learning_rate": 1.23525395749546e-05, "loss": 0.0708, "step": 4172 }, { "epoch": 2.21, "learning_rate": 1.2349206372053227e-05, "loss": 0.1701, "step": 4173 }, { "epoch": 2.21, "learning_rate": 1.2345872892880948e-05, "loss": 0.1358, "step": 4174 }, { "epoch": 2.21, "learning_rate": 1.2342539137829795e-05, "loss": 0.1593, "step": 4175 }, { "epoch": 2.21, "learning_rate": 1.2339205107291816e-05, "loss": 0.2218, "step": 4176 }, { "epoch": 2.21, "learning_rate": 1.2335870801659104e-05, "loss": 0.1581, "step": 4177 }, { "epoch": 2.21, "learning_rate": 1.2332536221323776e-05, "loss": 0.1073, "step": 4178 }, { "epoch": 2.21, "learning_rate": 1.2329201366677988e-05, "loss": 0.1841, "step": 4179 }, { "epoch": 2.21, "learning_rate": 1.2325866238113921e-05, "loss": 0.1664, "step": 4180 }, { "epoch": 2.21, "learning_rate": 1.2322530836023796e-05, "loss": 0.1224, "step": 4181 }, { "epoch": 2.21, "learning_rate": 1.2319195160799859e-05, "loss": 0.1585, "step": 4182 }, { "epoch": 2.21, "learning_rate": 1.2315859212834393e-05, "loss": 0.1346, "step": 4183 }, { "epoch": 2.21, "learning_rate": 1.2312522992519714e-05, "loss": 0.2391, "step": 4184 }, { "epoch": 2.22, "learning_rate": 1.2309186500248162e-05, "loss": 0.0889, "step": 4185 }, { "epoch": 2.22, "learning_rate": 1.230584973641212e-05, "loss": 0.1867, "step": 4186 }, { "epoch": 2.22, "learning_rate": 1.2302512701403995e-05, "loss": 0.1479, "step": 4187 }, { "epoch": 2.22, "learning_rate": 1.2299175395616224e-05, "loss": 0.1183, "step": 4188 }, { "epoch": 2.22, "learning_rate": 1.2295837819441286e-05, "loss": 0.1488, "step": 4189 }, { "epoch": 2.22, "learning_rate": 1.2292499973271687e-05, "loss": 0.1496, "step": 4190 }, { "epoch": 2.22, "learning_rate": 1.2289161857499958e-05, "loss": 0.1333, "step": 4191 }, { "epoch": 2.22, "learning_rate": 1.2285823472518673e-05, "loss": 0.1274, "step": 4192 }, { "epoch": 2.22, "learning_rate": 1.2282484818720428e-05, "loss": 0.0921, "step": 4193 }, { "epoch": 2.22, "learning_rate": 1.2279145896497858e-05, "loss": 0.0636, "step": 4194 }, { "epoch": 2.22, "learning_rate": 1.227580670624362e-05, "loss": 0.1242, "step": 4195 }, { "epoch": 2.22, "learning_rate": 1.2272467248350417e-05, "loss": 0.0862, "step": 4196 }, { "epoch": 2.22, "learning_rate": 1.226912752321097e-05, "loss": 0.1351, "step": 4197 }, { "epoch": 2.22, "learning_rate": 1.226578753121804e-05, "loss": 0.1679, "step": 4198 }, { "epoch": 2.22, "learning_rate": 1.2262447272764413e-05, "loss": 0.0871, "step": 4199 }, { "epoch": 2.22, "learning_rate": 1.2259106748242913e-05, "loss": 0.1316, "step": 4200 }, { "epoch": 2.22, "learning_rate": 1.2255765958046386e-05, "loss": 0.1525, "step": 4201 }, { "epoch": 2.22, "learning_rate": 1.2252424902567723e-05, "loss": 0.1357, "step": 4202 }, { "epoch": 2.22, "learning_rate": 1.224908358219983e-05, "loss": 0.1502, "step": 4203 }, { "epoch": 2.23, "learning_rate": 1.2245741997335656e-05, "loss": 0.1725, "step": 4204 }, { "epoch": 2.23, "learning_rate": 1.2242400148368176e-05, "loss": 0.1696, "step": 4205 }, { "epoch": 2.23, "learning_rate": 1.2239058035690404e-05, "loss": 0.1431, "step": 4206 }, { "epoch": 2.23, "learning_rate": 1.2235715659695368e-05, "loss": 0.1239, "step": 4207 }, { "epoch": 2.23, "learning_rate": 1.2232373020776147e-05, "loss": 0.1346, "step": 4208 }, { "epoch": 2.23, "learning_rate": 1.222903011932583e-05, "loss": 0.1019, "step": 4209 }, { "epoch": 2.23, "learning_rate": 1.2225686955737563e-05, "loss": 0.1964, "step": 4210 }, { "epoch": 2.23, "learning_rate": 1.2222343530404497e-05, "loss": 0.1915, "step": 4211 }, { "epoch": 2.23, "learning_rate": 1.2218999843719827e-05, "loss": 0.0911, "step": 4212 }, { "epoch": 2.23, "learning_rate": 1.221565589607678e-05, "loss": 0.1359, "step": 4213 }, { "epoch": 2.23, "learning_rate": 1.2212311687868606e-05, "loss": 0.1711, "step": 4214 }, { "epoch": 2.23, "learning_rate": 1.2208967219488592e-05, "loss": 0.1896, "step": 4215 }, { "epoch": 2.23, "learning_rate": 1.2205622491330054e-05, "loss": 0.1418, "step": 4216 }, { "epoch": 2.23, "learning_rate": 1.2202277503786337e-05, "loss": 0.133, "step": 4217 }, { "epoch": 2.23, "learning_rate": 1.2198932257250822e-05, "loss": 0.0724, "step": 4218 }, { "epoch": 2.23, "learning_rate": 1.2195586752116904e-05, "loss": 0.125, "step": 4219 }, { "epoch": 2.23, "learning_rate": 1.2192240988778035e-05, "loss": 0.1641, "step": 4220 }, { "epoch": 2.23, "learning_rate": 1.2188894967627674e-05, "loss": 0.1043, "step": 4221 }, { "epoch": 2.23, "learning_rate": 1.218554868905932e-05, "loss": 0.1843, "step": 4222 }, { "epoch": 2.24, "learning_rate": 1.2182202153466507e-05, "loss": 0.2349, "step": 4223 }, { "epoch": 2.24, "learning_rate": 1.2178855361242786e-05, "loss": 0.165, "step": 4224 }, { "epoch": 2.24, "learning_rate": 1.2175508312781748e-05, "loss": 0.0906, "step": 4225 }, { "epoch": 2.24, "learning_rate": 1.2172161008477014e-05, "loss": 0.1221, "step": 4226 }, { "epoch": 2.24, "learning_rate": 1.2168813448722234e-05, "loss": 0.1533, "step": 4227 }, { "epoch": 2.24, "learning_rate": 1.2165465633911084e-05, "loss": 0.1388, "step": 4228 }, { "epoch": 2.24, "learning_rate": 1.2162117564437275e-05, "loss": 0.1389, "step": 4229 }, { "epoch": 2.24, "learning_rate": 1.2158769240694544e-05, "loss": 0.1782, "step": 4230 }, { "epoch": 2.24, "learning_rate": 1.2155420663076663e-05, "loss": 0.0618, "step": 4231 }, { "epoch": 2.24, "learning_rate": 1.2152071831977425e-05, "loss": 0.1762, "step": 4232 }, { "epoch": 2.24, "learning_rate": 1.2148722747790667e-05, "loss": 0.1163, "step": 4233 }, { "epoch": 2.24, "learning_rate": 1.214537341091024e-05, "loss": 0.1715, "step": 4234 }, { "epoch": 2.24, "learning_rate": 1.2142023821730036e-05, "loss": 0.1197, "step": 4235 }, { "epoch": 2.24, "learning_rate": 1.213867398064397e-05, "loss": 0.0855, "step": 4236 }, { "epoch": 2.24, "learning_rate": 1.2135323888045991e-05, "loss": 0.105, "step": 4237 }, { "epoch": 2.24, "learning_rate": 1.213197354433008e-05, "loss": 0.104, "step": 4238 }, { "epoch": 2.24, "learning_rate": 1.2128622949890236e-05, "loss": 0.1492, "step": 4239 }, { "epoch": 2.24, "learning_rate": 1.2125272105120496e-05, "loss": 0.1221, "step": 4240 }, { "epoch": 2.24, "learning_rate": 1.212192101041493e-05, "loss": 0.1501, "step": 4241 }, { "epoch": 2.25, "learning_rate": 1.2118569666167631e-05, "loss": 0.1738, "step": 4242 }, { "epoch": 2.25, "learning_rate": 1.2115218072772724e-05, "loss": 0.1152, "step": 4243 }, { "epoch": 2.25, "learning_rate": 1.2111866230624358e-05, "loss": 0.2047, "step": 4244 }, { "epoch": 2.25, "learning_rate": 1.2108514140116721e-05, "loss": 0.116, "step": 4245 }, { "epoch": 2.25, "learning_rate": 1.2105161801644018e-05, "loss": 0.1149, "step": 4246 }, { "epoch": 2.25, "learning_rate": 1.21018092156005e-05, "loss": 0.2212, "step": 4247 }, { "epoch": 2.25, "learning_rate": 1.2098456382380428e-05, "loss": 0.1373, "step": 4248 }, { "epoch": 2.25, "learning_rate": 1.2095103302378106e-05, "loss": 0.1391, "step": 4249 }, { "epoch": 2.25, "learning_rate": 1.2091749975987861e-05, "loss": 0.159, "step": 4250 }, { "epoch": 2.25, "learning_rate": 1.208839640360405e-05, "loss": 0.121, "step": 4251 }, { "epoch": 2.25, "learning_rate": 1.208504258562106e-05, "loss": 0.0606, "step": 4252 }, { "epoch": 2.25, "learning_rate": 1.2081688522433305e-05, "loss": 0.1507, "step": 4253 }, { "epoch": 2.25, "learning_rate": 1.2078334214435232e-05, "loss": 0.1314, "step": 4254 }, { "epoch": 2.25, "learning_rate": 1.2074979662021309e-05, "loss": 0.0689, "step": 4255 }, { "epoch": 2.25, "learning_rate": 1.207162486558604e-05, "loss": 0.1477, "step": 4256 }, { "epoch": 2.25, "learning_rate": 1.2068269825523953e-05, "loss": 0.1696, "step": 4257 }, { "epoch": 2.25, "learning_rate": 1.2064914542229614e-05, "loss": 0.079, "step": 4258 }, { "epoch": 2.25, "learning_rate": 1.20615590160976e-05, "loss": 0.1383, "step": 4259 }, { "epoch": 2.25, "learning_rate": 1.205820324752253e-05, "loss": 0.1195, "step": 4260 }, { "epoch": 2.26, "learning_rate": 1.2054847236899053e-05, "loss": 0.1098, "step": 4261 }, { "epoch": 2.26, "learning_rate": 1.2051490984621839e-05, "loss": 0.1217, "step": 4262 }, { "epoch": 2.26, "learning_rate": 1.2048134491085588e-05, "loss": 0.1594, "step": 4263 }, { "epoch": 2.26, "learning_rate": 1.204477775668503e-05, "loss": 0.2188, "step": 4264 }, { "epoch": 2.26, "learning_rate": 1.2041420781814928e-05, "loss": 0.181, "step": 4265 }, { "epoch": 2.26, "learning_rate": 1.2038063566870062e-05, "loss": 0.0572, "step": 4266 }, { "epoch": 2.26, "learning_rate": 1.2034706112245247e-05, "loss": 0.0723, "step": 4267 }, { "epoch": 2.26, "learning_rate": 1.2031348418335331e-05, "loss": 0.1494, "step": 4268 }, { "epoch": 2.26, "learning_rate": 1.2027990485535179e-05, "loss": 0.1621, "step": 4269 }, { "epoch": 2.26, "learning_rate": 1.2024632314239695e-05, "loss": 0.1452, "step": 4270 }, { "epoch": 2.26, "learning_rate": 1.20212739048438e-05, "loss": 0.1427, "step": 4271 }, { "epoch": 2.26, "learning_rate": 1.2017915257742451e-05, "loss": 0.1404, "step": 4272 }, { "epoch": 2.26, "learning_rate": 1.2014556373330637e-05, "loss": 0.161, "step": 4273 }, { "epoch": 2.26, "learning_rate": 1.2011197252003364e-05, "loss": 0.1505, "step": 4274 }, { "epoch": 2.26, "learning_rate": 1.2007837894155669e-05, "loss": 0.1895, "step": 4275 }, { "epoch": 2.26, "learning_rate": 1.200447830018262e-05, "loss": 0.1795, "step": 4276 }, { "epoch": 2.26, "learning_rate": 1.200111847047931e-05, "loss": 0.1132, "step": 4277 }, { "epoch": 2.26, "learning_rate": 1.1997758405440867e-05, "loss": 0.1041, "step": 4278 }, { "epoch": 2.26, "learning_rate": 1.1994398105462435e-05, "loss": 0.1507, "step": 4279 }, { "epoch": 2.27, "learning_rate": 1.199103757093919e-05, "loss": 0.1946, "step": 4280 }, { "epoch": 2.27, "learning_rate": 1.1987676802266341e-05, "loss": 0.1172, "step": 4281 }, { "epoch": 2.27, "learning_rate": 1.1984315799839122e-05, "loss": 0.1392, "step": 4282 }, { "epoch": 2.27, "learning_rate": 1.1980954564052787e-05, "loss": 0.2338, "step": 4283 }, { "epoch": 2.27, "learning_rate": 1.1977593095302626e-05, "loss": 0.1026, "step": 4284 }, { "epoch": 2.27, "learning_rate": 1.1974231393983954e-05, "loss": 0.1379, "step": 4285 }, { "epoch": 2.27, "learning_rate": 1.1970869460492116e-05, "loss": 0.1195, "step": 4286 }, { "epoch": 2.27, "learning_rate": 1.1967507295222478e-05, "loss": 0.1209, "step": 4287 }, { "epoch": 2.27, "learning_rate": 1.1964144898570438e-05, "loss": 0.153, "step": 4288 }, { "epoch": 2.27, "learning_rate": 1.1960782270931416e-05, "loss": 0.0868, "step": 4289 }, { "epoch": 2.27, "learning_rate": 1.1957419412700873e-05, "loss": 0.1787, "step": 4290 }, { "epoch": 2.27, "learning_rate": 1.1954056324274274e-05, "loss": 0.0953, "step": 4291 }, { "epoch": 2.27, "learning_rate": 1.1950693006047134e-05, "loss": 0.1333, "step": 4292 }, { "epoch": 2.27, "learning_rate": 1.194732945841498e-05, "loss": 0.1368, "step": 4293 }, { "epoch": 2.27, "learning_rate": 1.1943965681773377e-05, "loss": 0.0977, "step": 4294 }, { "epoch": 2.27, "learning_rate": 1.1940601676517907e-05, "loss": 0.1114, "step": 4295 }, { "epoch": 2.27, "learning_rate": 1.193723744304418e-05, "loss": 0.1863, "step": 4296 }, { "epoch": 2.27, "learning_rate": 1.1933872981747843e-05, "loss": 0.0974, "step": 4297 }, { "epoch": 2.27, "learning_rate": 1.1930508293024559e-05, "loss": 0.1723, "step": 4298 }, { "epoch": 2.28, "learning_rate": 1.192714337727002e-05, "loss": 0.0855, "step": 4299 }, { "epoch": 2.28, "learning_rate": 1.192377823487995e-05, "loss": 0.1072, "step": 4300 }, { "epoch": 2.28, "learning_rate": 1.1920412866250093e-05, "loss": 0.1401, "step": 4301 }, { "epoch": 2.28, "learning_rate": 1.1917047271776223e-05, "loss": 0.191, "step": 4302 }, { "epoch": 2.28, "learning_rate": 1.191368145185414e-05, "loss": 0.0951, "step": 4303 }, { "epoch": 2.28, "learning_rate": 1.191031540687967e-05, "loss": 0.1609, "step": 4304 }, { "epoch": 2.28, "learning_rate": 1.1906949137248667e-05, "loss": 0.1348, "step": 4305 }, { "epoch": 2.28, "learning_rate": 1.190358264335701e-05, "loss": 0.171, "step": 4306 }, { "epoch": 2.28, "learning_rate": 1.1900215925600607e-05, "loss": 0.1413, "step": 4307 }, { "epoch": 2.28, "learning_rate": 1.1896848984375385e-05, "loss": 0.1616, "step": 4308 }, { "epoch": 2.28, "learning_rate": 1.1893481820077306e-05, "loss": 0.1617, "step": 4309 }, { "epoch": 2.28, "learning_rate": 1.1890114433102354e-05, "loss": 0.1255, "step": 4310 }, { "epoch": 2.28, "learning_rate": 1.188674682384654e-05, "loss": 0.1506, "step": 4311 }, { "epoch": 2.28, "learning_rate": 1.18833789927059e-05, "loss": 0.1873, "step": 4312 }, { "epoch": 2.28, "learning_rate": 1.1880010940076495e-05, "loss": 0.1119, "step": 4313 }, { "epoch": 2.28, "learning_rate": 1.1876642666354418e-05, "loss": 0.1811, "step": 4314 }, { "epoch": 2.28, "learning_rate": 1.1873274171935785e-05, "loss": 0.1062, "step": 4315 }, { "epoch": 2.28, "learning_rate": 1.1869905457216733e-05, "loss": 0.2399, "step": 4316 }, { "epoch": 2.29, "learning_rate": 1.186653652259343e-05, "loss": 0.1602, "step": 4317 }, { "epoch": 2.29, "learning_rate": 1.186316736846207e-05, "loss": 0.1566, "step": 4318 }, { "epoch": 2.29, "learning_rate": 1.1859797995218869e-05, "loss": 0.272, "step": 4319 }, { "epoch": 2.29, "learning_rate": 1.1856428403260074e-05, "loss": 0.2018, "step": 4320 }, { "epoch": 2.29, "learning_rate": 1.1853058592981954e-05, "loss": 0.2015, "step": 4321 }, { "epoch": 2.29, "learning_rate": 1.1849688564780805e-05, "loss": 0.1365, "step": 4322 }, { "epoch": 2.29, "learning_rate": 1.184631831905295e-05, "loss": 0.116, "step": 4323 }, { "epoch": 2.29, "learning_rate": 1.1842947856194733e-05, "loss": 0.1329, "step": 4324 }, { "epoch": 2.29, "learning_rate": 1.183957717660253e-05, "loss": 0.1325, "step": 4325 }, { "epoch": 2.29, "learning_rate": 1.1836206280672736e-05, "loss": 0.1057, "step": 4326 }, { "epoch": 2.29, "learning_rate": 1.1832835168801774e-05, "loss": 0.0941, "step": 4327 }, { "epoch": 2.29, "learning_rate": 1.1829463841386095e-05, "loss": 0.0964, "step": 4328 }, { "epoch": 2.29, "learning_rate": 1.1826092298822172e-05, "loss": 0.127, "step": 4329 }, { "epoch": 2.29, "learning_rate": 1.1822720541506506e-05, "loss": 0.112, "step": 4330 }, { "epoch": 2.29, "learning_rate": 1.181934856983562e-05, "loss": 0.1563, "step": 4331 }, { "epoch": 2.29, "learning_rate": 1.1815976384206062e-05, "loss": 0.18, "step": 4332 }, { "epoch": 2.29, "learning_rate": 1.1812603985014412e-05, "loss": 0.1571, "step": 4333 }, { "epoch": 2.29, "learning_rate": 1.1809231372657266e-05, "loss": 0.1236, "step": 4334 }, { "epoch": 2.29, "learning_rate": 1.1805858547531255e-05, "loss": 0.1879, "step": 4335 }, { "epoch": 2.3, "learning_rate": 1.1802485510033024e-05, "loss": 0.103, "step": 4336 }, { "epoch": 2.3, "learning_rate": 1.1799112260559249e-05, "loss": 0.1782, "step": 4337 }, { "epoch": 2.3, "learning_rate": 1.179573879950663e-05, "loss": 0.0639, "step": 4338 }, { "epoch": 2.3, "learning_rate": 1.1792365127271894e-05, "loss": 0.1173, "step": 4339 }, { "epoch": 2.3, "learning_rate": 1.1788991244251788e-05, "loss": 0.1602, "step": 4340 }, { "epoch": 2.3, "learning_rate": 1.1785617150843091e-05, "loss": 0.131, "step": 4341 }, { "epoch": 2.3, "learning_rate": 1.17822428474426e-05, "loss": 0.0939, "step": 4342 }, { "epoch": 2.3, "learning_rate": 1.177886833444714e-05, "loss": 0.1174, "step": 4343 }, { "epoch": 2.3, "learning_rate": 1.1775493612253556e-05, "loss": 0.1604, "step": 4344 }, { "epoch": 2.3, "learning_rate": 1.1772118681258726e-05, "loss": 0.1697, "step": 4345 }, { "epoch": 2.3, "learning_rate": 1.1768743541859549e-05, "loss": 0.0711, "step": 4346 }, { "epoch": 2.3, "learning_rate": 1.1765368194452944e-05, "loss": 0.1063, "step": 4347 }, { "epoch": 2.3, "learning_rate": 1.1761992639435858e-05, "loss": 0.1258, "step": 4348 }, { "epoch": 2.3, "learning_rate": 1.1758616877205262e-05, "loss": 0.1097, "step": 4349 }, { "epoch": 2.3, "learning_rate": 1.1755240908158152e-05, "loss": 0.1217, "step": 4350 }, { "epoch": 2.3, "learning_rate": 1.1751864732691553e-05, "loss": 0.1564, "step": 4351 }, { "epoch": 2.3, "learning_rate": 1.1748488351202503e-05, "loss": 0.1764, "step": 4352 }, { "epoch": 2.3, "learning_rate": 1.1745111764088075e-05, "loss": 0.0701, "step": 4353 }, { "epoch": 2.3, "learning_rate": 1.1741734971745358e-05, "loss": 0.1422, "step": 4354 }, { "epoch": 2.31, "learning_rate": 1.173835797457147e-05, "loss": 0.0887, "step": 4355 }, { "epoch": 2.31, "learning_rate": 1.173498077296355e-05, "loss": 0.17, "step": 4356 }, { "epoch": 2.31, "learning_rate": 1.173160336731877e-05, "loss": 0.0945, "step": 4357 }, { "epoch": 2.31, "learning_rate": 1.1728225758034314e-05, "loss": 0.2148, "step": 4358 }, { "epoch": 2.31, "learning_rate": 1.1724847945507392e-05, "loss": 0.1323, "step": 4359 }, { "epoch": 2.31, "learning_rate": 1.1721469930135248e-05, "loss": 0.1162, "step": 4360 }, { "epoch": 2.31, "learning_rate": 1.1718091712315136e-05, "loss": 0.1072, "step": 4361 }, { "epoch": 2.31, "learning_rate": 1.1714713292444344e-05, "loss": 0.1606, "step": 4362 }, { "epoch": 2.31, "learning_rate": 1.1711334670920183e-05, "loss": 0.091, "step": 4363 }, { "epoch": 2.31, "learning_rate": 1.1707955848139978e-05, "loss": 0.1311, "step": 4364 }, { "epoch": 2.31, "learning_rate": 1.1704576824501086e-05, "loss": 0.1767, "step": 4365 }, { "epoch": 2.31, "learning_rate": 1.1701197600400891e-05, "loss": 0.1285, "step": 4366 }, { "epoch": 2.31, "learning_rate": 1.1697818176236795e-05, "loss": 0.1435, "step": 4367 }, { "epoch": 2.31, "learning_rate": 1.1694438552406221e-05, "loss": 0.1368, "step": 4368 }, { "epoch": 2.31, "learning_rate": 1.1691058729306622e-05, "loss": 0.1316, "step": 4369 }, { "epoch": 2.31, "learning_rate": 1.168767870733547e-05, "loss": 0.1927, "step": 4370 }, { "epoch": 2.31, "learning_rate": 1.168429848689026e-05, "loss": 0.1085, "step": 4371 }, { "epoch": 2.31, "learning_rate": 1.1680918068368515e-05, "loss": 0.0741, "step": 4372 }, { "epoch": 2.31, "learning_rate": 1.1677537452167776e-05, "loss": 0.0509, "step": 4373 }, { "epoch": 2.32, "learning_rate": 1.1674156638685608e-05, "loss": 0.1, "step": 4374 }, { "epoch": 2.32, "learning_rate": 1.1670775628319606e-05, "loss": 0.1214, "step": 4375 }, { "epoch": 2.32, "learning_rate": 1.166739442146738e-05, "loss": 0.0595, "step": 4376 }, { "epoch": 2.32, "learning_rate": 1.1664013018526565e-05, "loss": 0.1392, "step": 4377 }, { "epoch": 2.32, "learning_rate": 1.1660631419894823e-05, "loss": 0.08, "step": 4378 }, { "epoch": 2.32, "learning_rate": 1.1657249625969835e-05, "loss": 0.123, "step": 4379 }, { "epoch": 2.32, "learning_rate": 1.1653867637149301e-05, "loss": 0.172, "step": 4380 }, { "epoch": 2.32, "learning_rate": 1.1650485453830955e-05, "loss": 0.1585, "step": 4381 }, { "epoch": 2.32, "learning_rate": 1.1647103076412547e-05, "loss": 0.1603, "step": 4382 }, { "epoch": 2.32, "learning_rate": 1.1643720505291851e-05, "loss": 0.1594, "step": 4383 }, { "epoch": 2.32, "learning_rate": 1.1640337740866663e-05, "loss": 0.0969, "step": 4384 }, { "epoch": 2.32, "learning_rate": 1.16369547835348e-05, "loss": 0.0825, "step": 4385 }, { "epoch": 2.32, "learning_rate": 1.16335716336941e-05, "loss": 0.1334, "step": 4386 }, { "epoch": 2.32, "learning_rate": 1.1630188291742441e-05, "loss": 0.1207, "step": 4387 }, { "epoch": 2.32, "learning_rate": 1.1626804758077699e-05, "loss": 0.1445, "step": 4388 }, { "epoch": 2.32, "learning_rate": 1.1623421033097787e-05, "loss": 0.1874, "step": 4389 }, { "epoch": 2.32, "learning_rate": 1.1620037117200635e-05, "loss": 0.0708, "step": 4390 }, { "epoch": 2.32, "learning_rate": 1.1616653010784203e-05, "loss": 0.118, "step": 4391 }, { "epoch": 2.32, "learning_rate": 1.1613268714246462e-05, "loss": 0.1412, "step": 4392 }, { "epoch": 2.33, "learning_rate": 1.1609884227985413e-05, "loss": 0.0843, "step": 4393 }, { "epoch": 2.33, "learning_rate": 1.160649955239908e-05, "loss": 0.2176, "step": 4394 }, { "epoch": 2.33, "learning_rate": 1.1603114687885506e-05, "loss": 0.1305, "step": 4395 }, { "epoch": 2.33, "learning_rate": 1.1599729634842754e-05, "loss": 0.126, "step": 4396 }, { "epoch": 2.33, "learning_rate": 1.1596344393668914e-05, "loss": 0.1109, "step": 4397 }, { "epoch": 2.33, "learning_rate": 1.1592958964762101e-05, "loss": 0.1275, "step": 4398 }, { "epoch": 2.33, "learning_rate": 1.1589573348520444e-05, "loss": 0.0852, "step": 4399 }, { "epoch": 2.33, "learning_rate": 1.1586187545342097e-05, "loss": 0.1466, "step": 4400 }, { "epoch": 2.33, "learning_rate": 1.1582801555625236e-05, "loss": 0.1252, "step": 4401 }, { "epoch": 2.33, "learning_rate": 1.157941537976806e-05, "loss": 0.1743, "step": 4402 }, { "epoch": 2.33, "learning_rate": 1.157602901816879e-05, "loss": 0.2346, "step": 4403 }, { "epoch": 2.33, "learning_rate": 1.157264247122567e-05, "loss": 0.1201, "step": 4404 }, { "epoch": 2.33, "learning_rate": 1.156925573933696e-05, "loss": 0.1132, "step": 4405 }, { "epoch": 2.33, "learning_rate": 1.156586882290095e-05, "loss": 0.1107, "step": 4406 }, { "epoch": 2.33, "learning_rate": 1.1562481722315944e-05, "loss": 0.1843, "step": 4407 }, { "epoch": 2.33, "learning_rate": 1.1559094437980272e-05, "loss": 0.08, "step": 4408 }, { "epoch": 2.33, "learning_rate": 1.1555706970292287e-05, "loss": 0.1755, "step": 4409 }, { "epoch": 2.33, "learning_rate": 1.1552319319650358e-05, "loss": 0.0889, "step": 4410 }, { "epoch": 2.33, "learning_rate": 1.1548931486452882e-05, "loss": 0.1926, "step": 4411 }, { "epoch": 2.34, "learning_rate": 1.154554347109827e-05, "loss": 0.0751, "step": 4412 }, { "epoch": 2.34, "learning_rate": 1.1542155273984963e-05, "loss": 0.1795, "step": 4413 }, { "epoch": 2.34, "learning_rate": 1.1538766895511417e-05, "loss": 0.1403, "step": 4414 }, { "epoch": 2.34, "learning_rate": 1.1535378336076115e-05, "loss": 0.1331, "step": 4415 }, { "epoch": 2.34, "learning_rate": 1.1531989596077553e-05, "loss": 0.1169, "step": 4416 }, { "epoch": 2.34, "learning_rate": 1.152860067591425e-05, "loss": 0.1623, "step": 4417 }, { "epoch": 2.34, "learning_rate": 1.1525211575984758e-05, "loss": 0.0869, "step": 4418 }, { "epoch": 2.34, "learning_rate": 1.1521822296687639e-05, "loss": 0.0663, "step": 4419 }, { "epoch": 2.34, "learning_rate": 1.1518432838421472e-05, "loss": 0.2135, "step": 4420 }, { "epoch": 2.34, "learning_rate": 1.151504320158487e-05, "loss": 0.1538, "step": 4421 }, { "epoch": 2.34, "learning_rate": 1.1511653386576455e-05, "loss": 0.1289, "step": 4422 }, { "epoch": 2.34, "learning_rate": 1.1508263393794882e-05, "loss": 0.1567, "step": 4423 }, { "epoch": 2.34, "learning_rate": 1.1504873223638814e-05, "loss": 0.1331, "step": 4424 }, { "epoch": 2.34, "learning_rate": 1.1501482876506947e-05, "loss": 0.1239, "step": 4425 }, { "epoch": 2.34, "learning_rate": 1.1498092352797985e-05, "loss": 0.1261, "step": 4426 }, { "epoch": 2.34, "learning_rate": 1.1494701652910664e-05, "loss": 0.1746, "step": 4427 }, { "epoch": 2.34, "learning_rate": 1.1491310777243738e-05, "loss": 0.2418, "step": 4428 }, { "epoch": 2.34, "learning_rate": 1.1487919726195975e-05, "loss": 0.1406, "step": 4429 }, { "epoch": 2.34, "learning_rate": 1.1484528500166174e-05, "loss": 0.1237, "step": 4430 }, { "epoch": 2.35, "learning_rate": 1.1481137099553146e-05, "loss": 0.2274, "step": 4431 }, { "epoch": 2.35, "learning_rate": 1.1477745524755729e-05, "loss": 0.0965, "step": 4432 }, { "epoch": 2.35, "learning_rate": 1.1474353776172773e-05, "loss": 0.153, "step": 4433 }, { "epoch": 2.35, "learning_rate": 1.1470961854203158e-05, "loss": 0.1634, "step": 4434 }, { "epoch": 2.35, "learning_rate": 1.1467569759245779e-05, "loss": 0.0956, "step": 4435 }, { "epoch": 2.35, "learning_rate": 1.1464177491699553e-05, "loss": 0.1048, "step": 4436 }, { "epoch": 2.35, "learning_rate": 1.1460785051963414e-05, "loss": 0.1311, "step": 4437 }, { "epoch": 2.35, "learning_rate": 1.1457392440436319e-05, "loss": 0.116, "step": 4438 }, { "epoch": 2.35, "learning_rate": 1.1453999657517252e-05, "loss": 0.1012, "step": 4439 }, { "epoch": 2.35, "learning_rate": 1.1450606703605207e-05, "loss": 0.1392, "step": 4440 }, { "epoch": 2.35, "learning_rate": 1.1447213579099197e-05, "loss": 0.1214, "step": 4441 }, { "epoch": 2.35, "learning_rate": 1.1443820284398263e-05, "loss": 0.1673, "step": 4442 }, { "epoch": 2.35, "learning_rate": 1.1440426819901465e-05, "loss": 0.2235, "step": 4443 }, { "epoch": 2.35, "learning_rate": 1.1437033186007876e-05, "loss": 0.0794, "step": 4444 }, { "epoch": 2.35, "learning_rate": 1.1433639383116595e-05, "loss": 0.1455, "step": 4445 }, { "epoch": 2.35, "learning_rate": 1.1430245411626739e-05, "loss": 0.1054, "step": 4446 }, { "epoch": 2.35, "learning_rate": 1.1426851271937448e-05, "loss": 0.1803, "step": 4447 }, { "epoch": 2.35, "learning_rate": 1.1423456964447875e-05, "loss": 0.1504, "step": 4448 }, { "epoch": 2.35, "learning_rate": 1.14200624895572e-05, "loss": 0.1468, "step": 4449 }, { "epoch": 2.36, "learning_rate": 1.1416667847664615e-05, "loss": 0.1432, "step": 4450 }, { "epoch": 2.36, "learning_rate": 1.1413273039169344e-05, "loss": 0.1043, "step": 4451 }, { "epoch": 2.36, "learning_rate": 1.1409878064470616e-05, "loss": 0.096, "step": 4452 }, { "epoch": 2.36, "learning_rate": 1.1406482923967685e-05, "loss": 0.1989, "step": 4453 }, { "epoch": 2.36, "learning_rate": 1.1403087618059827e-05, "loss": 0.0949, "step": 4454 }, { "epoch": 2.36, "learning_rate": 1.1399692147146342e-05, "loss": 0.1118, "step": 4455 }, { "epoch": 2.36, "learning_rate": 1.1396296511626539e-05, "loss": 0.1287, "step": 4456 }, { "epoch": 2.36, "learning_rate": 1.1392900711899748e-05, "loss": 0.1267, "step": 4457 }, { "epoch": 2.36, "learning_rate": 1.1389504748365323e-05, "loss": 0.139, "step": 4458 }, { "epoch": 2.36, "learning_rate": 1.1386108621422636e-05, "loss": 0.0903, "step": 4459 }, { "epoch": 2.36, "learning_rate": 1.1382712331471082e-05, "loss": 0.144, "step": 4460 }, { "epoch": 2.36, "learning_rate": 1.1379315878910064e-05, "loss": 0.1222, "step": 4461 }, { "epoch": 2.36, "learning_rate": 1.1375919264139014e-05, "loss": 0.1258, "step": 4462 }, { "epoch": 2.36, "learning_rate": 1.137252248755738e-05, "loss": 0.1518, "step": 4463 }, { "epoch": 2.36, "learning_rate": 1.1369125549564628e-05, "loss": 0.1195, "step": 4464 }, { "epoch": 2.36, "learning_rate": 1.1365728450560245e-05, "loss": 0.1651, "step": 4465 }, { "epoch": 2.36, "learning_rate": 1.1362331190943736e-05, "loss": 0.138, "step": 4466 }, { "epoch": 2.36, "learning_rate": 1.1358933771114623e-05, "loss": 0.1065, "step": 4467 }, { "epoch": 2.36, "learning_rate": 1.1355536191472457e-05, "loss": 0.1372, "step": 4468 }, { "epoch": 2.37, "learning_rate": 1.1352138452416785e-05, "loss": 0.1031, "step": 4469 }, { "epoch": 2.37, "learning_rate": 1.1348740554347197e-05, "loss": 0.1295, "step": 4470 }, { "epoch": 2.37, "learning_rate": 1.1345342497663293e-05, "loss": 0.3377, "step": 4471 }, { "epoch": 2.37, "learning_rate": 1.1341944282764685e-05, "loss": 0.0789, "step": 4472 }, { "epoch": 2.37, "learning_rate": 1.1338545910051012e-05, "loss": 0.1031, "step": 4473 }, { "epoch": 2.37, "learning_rate": 1.1335147379921929e-05, "loss": 0.1503, "step": 4474 }, { "epoch": 2.37, "learning_rate": 1.1331748692777107e-05, "loss": 0.1318, "step": 4475 }, { "epoch": 2.37, "learning_rate": 1.1328349849016243e-05, "loss": 0.1084, "step": 4476 }, { "epoch": 2.37, "learning_rate": 1.132495084903904e-05, "loss": 0.1442, "step": 4477 }, { "epoch": 2.37, "learning_rate": 1.1321551693245231e-05, "loss": 0.1504, "step": 4478 }, { "epoch": 2.37, "learning_rate": 1.1318152382034565e-05, "loss": 0.1111, "step": 4479 }, { "epoch": 2.37, "learning_rate": 1.13147529158068e-05, "loss": 0.1528, "step": 4480 }, { "epoch": 2.37, "learning_rate": 1.1311353294961727e-05, "loss": 0.1406, "step": 4481 }, { "epoch": 2.37, "learning_rate": 1.1307953519899142e-05, "loss": 0.1251, "step": 4482 }, { "epoch": 2.37, "learning_rate": 1.1304553591018865e-05, "loss": 0.1351, "step": 4483 }, { "epoch": 2.37, "learning_rate": 1.1301153508720738e-05, "loss": 0.1321, "step": 4484 }, { "epoch": 2.37, "learning_rate": 1.1297753273404614e-05, "loss": 0.1464, "step": 4485 }, { "epoch": 2.37, "learning_rate": 1.1294352885470367e-05, "loss": 0.1524, "step": 4486 }, { "epoch": 2.38, "learning_rate": 1.129095234531789e-05, "loss": 0.1669, "step": 4487 }, { "epoch": 2.38, "learning_rate": 1.1287551653347091e-05, "loss": 0.1879, "step": 4488 }, { "epoch": 2.38, "learning_rate": 1.1284150809957896e-05, "loss": 0.1618, "step": 4489 }, { "epoch": 2.38, "learning_rate": 1.1280749815550251e-05, "loss": 0.1414, "step": 4490 }, { "epoch": 2.38, "learning_rate": 1.1277348670524125e-05, "loss": 0.2028, "step": 4491 }, { "epoch": 2.38, "learning_rate": 1.1273947375279493e-05, "loss": 0.1295, "step": 4492 }, { "epoch": 2.38, "learning_rate": 1.1270545930216353e-05, "loss": 0.1458, "step": 4493 }, { "epoch": 2.38, "learning_rate": 1.126714433573472e-05, "loss": 0.1402, "step": 4494 }, { "epoch": 2.38, "learning_rate": 1.1263742592234634e-05, "loss": 0.1324, "step": 4495 }, { "epoch": 2.38, "learning_rate": 1.1260340700116142e-05, "loss": 0.102, "step": 4496 }, { "epoch": 2.38, "learning_rate": 1.1256938659779312e-05, "loss": 0.1446, "step": 4497 }, { "epoch": 2.38, "learning_rate": 1.1253536471624233e-05, "loss": 0.1205, "step": 4498 }, { "epoch": 2.38, "learning_rate": 1.1250134136051005e-05, "loss": 0.1235, "step": 4499 }, { "epoch": 2.38, "learning_rate": 1.124673165345975e-05, "loss": 0.1213, "step": 4500 }, { "epoch": 2.38, "learning_rate": 1.1243329024250608e-05, "loss": 0.1748, "step": 4501 }, { "epoch": 2.38, "learning_rate": 1.1239926248823731e-05, "loss": 0.0766, "step": 4502 }, { "epoch": 2.38, "learning_rate": 1.1236523327579296e-05, "loss": 0.1825, "step": 4503 }, { "epoch": 2.38, "learning_rate": 1.1233120260917489e-05, "loss": 0.1177, "step": 4504 }, { "epoch": 2.38, "learning_rate": 1.1229717049238515e-05, "loss": 0.1616, "step": 4505 }, { "epoch": 2.39, "learning_rate": 1.1226313692942603e-05, "loss": 0.081, "step": 4506 }, { "epoch": 2.39, "learning_rate": 1.1222910192429991e-05, "loss": 0.1173, "step": 4507 }, { "epoch": 2.39, "learning_rate": 1.121950654810094e-05, "loss": 0.1358, "step": 4508 }, { "epoch": 2.39, "learning_rate": 1.1216102760355718e-05, "loss": 0.218, "step": 4509 }, { "epoch": 2.39, "learning_rate": 1.1212698829594624e-05, "loss": 0.1379, "step": 4510 }, { "epoch": 2.39, "learning_rate": 1.1209294756217957e-05, "loss": 0.1233, "step": 4511 }, { "epoch": 2.39, "learning_rate": 1.1205890540626054e-05, "loss": 0.166, "step": 4512 }, { "epoch": 2.39, "learning_rate": 1.1202486183219247e-05, "loss": 0.1812, "step": 4513 }, { "epoch": 2.39, "learning_rate": 1.11990816843979e-05, "loss": 0.1614, "step": 4514 }, { "epoch": 2.39, "learning_rate": 1.1195677044562387e-05, "loss": 0.049, "step": 4515 }, { "epoch": 2.39, "learning_rate": 1.11922722641131e-05, "loss": 0.1062, "step": 4516 }, { "epoch": 2.39, "learning_rate": 1.1188867343450447e-05, "loss": 0.1664, "step": 4517 }, { "epoch": 2.39, "learning_rate": 1.118546228297485e-05, "loss": 0.2086, "step": 4518 }, { "epoch": 2.39, "learning_rate": 1.1182057083086754e-05, "loss": 0.0892, "step": 4519 }, { "epoch": 2.39, "learning_rate": 1.117865174418662e-05, "loss": 0.083, "step": 4520 }, { "epoch": 2.39, "learning_rate": 1.117524626667491e-05, "loss": 0.1242, "step": 4521 }, { "epoch": 2.39, "learning_rate": 1.1171840650952126e-05, "loss": 0.1146, "step": 4522 }, { "epoch": 2.39, "learning_rate": 1.116843489741877e-05, "loss": 0.1717, "step": 4523 }, { "epoch": 2.39, "learning_rate": 1.1165029006475365e-05, "loss": 0.1111, "step": 4524 }, { "epoch": 2.4, "learning_rate": 1.1161622978522451e-05, "loss": 0.1326, "step": 4525 }, { "epoch": 2.4, "learning_rate": 1.1158216813960578e-05, "loss": 0.093, "step": 4526 }, { "epoch": 2.4, "learning_rate": 1.1154810513190321e-05, "loss": 0.1581, "step": 4527 }, { "epoch": 2.4, "learning_rate": 1.1151404076612274e-05, "loss": 0.1575, "step": 4528 }, { "epoch": 2.4, "learning_rate": 1.1147997504627028e-05, "loss": 0.0995, "step": 4529 }, { "epoch": 2.4, "learning_rate": 1.1144590797635208e-05, "loss": 0.0713, "step": 4530 }, { "epoch": 2.4, "learning_rate": 1.1141183956037449e-05, "loss": 0.1396, "step": 4531 }, { "epoch": 2.4, "learning_rate": 1.11377769802344e-05, "loss": 0.1523, "step": 4532 }, { "epoch": 2.4, "learning_rate": 1.113436987062673e-05, "loss": 0.0724, "step": 4533 }, { "epoch": 2.4, "learning_rate": 1.1130962627615117e-05, "loss": 0.116, "step": 4534 }, { "epoch": 2.4, "learning_rate": 1.1127555251600261e-05, "loss": 0.1461, "step": 4535 }, { "epoch": 2.4, "learning_rate": 1.1124147742982879e-05, "loss": 0.1389, "step": 4536 }, { "epoch": 2.4, "learning_rate": 1.1120740102163694e-05, "loss": 0.1467, "step": 4537 }, { "epoch": 2.4, "learning_rate": 1.1117332329543457e-05, "loss": 0.1603, "step": 4538 }, { "epoch": 2.4, "learning_rate": 1.1113924425522925e-05, "loss": 0.1346, "step": 4539 }, { "epoch": 2.4, "learning_rate": 1.1110516390502872e-05, "loss": 0.1659, "step": 4540 }, { "epoch": 2.4, "learning_rate": 1.1107108224884094e-05, "loss": 0.1286, "step": 4541 }, { "epoch": 2.4, "learning_rate": 1.110369992906739e-05, "loss": 0.1827, "step": 4542 }, { "epoch": 2.4, "learning_rate": 1.1100291503453585e-05, "loss": 0.0825, "step": 4543 }, { "epoch": 2.41, "learning_rate": 1.109688294844352e-05, "loss": 0.1337, "step": 4544 }, { "epoch": 2.41, "learning_rate": 1.1093474264438045e-05, "loss": 0.0663, "step": 4545 }, { "epoch": 2.41, "learning_rate": 1.1090065451838027e-05, "loss": 0.1941, "step": 4546 }, { "epoch": 2.41, "learning_rate": 1.1086656511044345e-05, "loss": 0.1393, "step": 4547 }, { "epoch": 2.41, "learning_rate": 1.10832474424579e-05, "loss": 0.094, "step": 4548 }, { "epoch": 2.41, "learning_rate": 1.1079838246479606e-05, "loss": 0.0959, "step": 4549 }, { "epoch": 2.41, "learning_rate": 1.1076428923510388e-05, "loss": 0.0878, "step": 4550 }, { "epoch": 2.41, "learning_rate": 1.1073019473951188e-05, "loss": 0.1185, "step": 4551 }, { "epoch": 2.41, "learning_rate": 1.1069609898202966e-05, "loss": 0.1441, "step": 4552 }, { "epoch": 2.41, "learning_rate": 1.1066200196666692e-05, "loss": 0.2119, "step": 4553 }, { "epoch": 2.41, "learning_rate": 1.1062790369743356e-05, "loss": 0.1408, "step": 4554 }, { "epoch": 2.41, "learning_rate": 1.1059380417833956e-05, "loss": 0.1897, "step": 4555 }, { "epoch": 2.41, "learning_rate": 1.105597034133951e-05, "loss": 0.1287, "step": 4556 }, { "epoch": 2.41, "learning_rate": 1.1052560140661053e-05, "loss": 0.1692, "step": 4557 }, { "epoch": 2.41, "learning_rate": 1.1049149816199623e-05, "loss": 0.1407, "step": 4558 }, { "epoch": 2.41, "learning_rate": 1.1045739368356287e-05, "loss": 0.1005, "step": 4559 }, { "epoch": 2.41, "learning_rate": 1.1042328797532116e-05, "loss": 0.0784, "step": 4560 }, { "epoch": 2.41, "learning_rate": 1.1038918104128203e-05, "loss": 0.1647, "step": 4561 }, { "epoch": 2.41, "learning_rate": 1.1035507288545648e-05, "loss": 0.104, "step": 4562 }, { "epoch": 2.42, "learning_rate": 1.1032096351185567e-05, "loss": 0.1548, "step": 4563 }, { "epoch": 2.42, "learning_rate": 1.10286852924491e-05, "loss": 0.1609, "step": 4564 }, { "epoch": 2.42, "learning_rate": 1.1025274112737388e-05, "loss": 0.1063, "step": 4565 }, { "epoch": 2.42, "learning_rate": 1.1021862812451594e-05, "loss": 0.0898, "step": 4566 }, { "epoch": 2.42, "learning_rate": 1.1018451391992891e-05, "loss": 0.1485, "step": 4567 }, { "epoch": 2.42, "learning_rate": 1.101503985176247e-05, "loss": 0.1719, "step": 4568 }, { "epoch": 2.42, "learning_rate": 1.1011628192161531e-05, "loss": 0.1611, "step": 4569 }, { "epoch": 2.42, "learning_rate": 1.10082164135913e-05, "loss": 0.0872, "step": 4570 }, { "epoch": 2.42, "learning_rate": 1.1004804516452997e-05, "loss": 0.1143, "step": 4571 }, { "epoch": 2.42, "learning_rate": 1.1001392501147876e-05, "loss": 0.108, "step": 4572 }, { "epoch": 2.42, "learning_rate": 1.099798036807719e-05, "loss": 0.0923, "step": 4573 }, { "epoch": 2.42, "learning_rate": 1.0994568117642216e-05, "loss": 0.1366, "step": 4574 }, { "epoch": 2.42, "learning_rate": 1.099115575024424e-05, "loss": 0.1913, "step": 4575 }, { "epoch": 2.42, "learning_rate": 1.0987743266284562e-05, "loss": 0.1419, "step": 4576 }, { "epoch": 2.42, "learning_rate": 1.09843306661645e-05, "loss": 0.0893, "step": 4577 }, { "epoch": 2.42, "learning_rate": 1.0980917950285372e-05, "loss": 0.182, "step": 4578 }, { "epoch": 2.42, "learning_rate": 1.0977505119048529e-05, "loss": 0.1212, "step": 4579 }, { "epoch": 2.42, "learning_rate": 1.0974092172855322e-05, "loss": 0.1324, "step": 4580 }, { "epoch": 2.42, "learning_rate": 1.0970679112107124e-05, "loss": 0.169, "step": 4581 }, { "epoch": 2.43, "learning_rate": 1.0967265937205309e-05, "loss": 0.0731, "step": 4582 }, { "epoch": 2.43, "learning_rate": 1.0963852648551278e-05, "loss": 0.1519, "step": 4583 }, { "epoch": 2.43, "learning_rate": 1.096043924654644e-05, "loss": 0.0863, "step": 4584 }, { "epoch": 2.43, "learning_rate": 1.0957025731592216e-05, "loss": 0.1504, "step": 4585 }, { "epoch": 2.43, "learning_rate": 1.0953612104090043e-05, "loss": 0.1158, "step": 4586 }, { "epoch": 2.43, "learning_rate": 1.0950198364441369e-05, "loss": 0.1031, "step": 4587 }, { "epoch": 2.43, "learning_rate": 1.0946784513047652e-05, "loss": 0.1287, "step": 4588 }, { "epoch": 2.43, "learning_rate": 1.0943370550310374e-05, "loss": 0.0673, "step": 4589 }, { "epoch": 2.43, "learning_rate": 1.093995647663102e-05, "loss": 0.0567, "step": 4590 }, { "epoch": 2.43, "learning_rate": 1.0936542292411087e-05, "loss": 0.1561, "step": 4591 }, { "epoch": 2.43, "learning_rate": 1.0933127998052096e-05, "loss": 0.1414, "step": 4592 }, { "epoch": 2.43, "learning_rate": 1.0929713593955573e-05, "loss": 0.0879, "step": 4593 }, { "epoch": 2.43, "learning_rate": 1.0926299080523053e-05, "loss": 0.1857, "step": 4594 }, { "epoch": 2.43, "learning_rate": 1.0922884458156094e-05, "loss": 0.1246, "step": 4595 }, { "epoch": 2.43, "learning_rate": 1.091946972725626e-05, "loss": 0.1319, "step": 4596 }, { "epoch": 2.43, "learning_rate": 1.0916054888225131e-05, "loss": 0.1167, "step": 4597 }, { "epoch": 2.43, "learning_rate": 1.0912639941464293e-05, "loss": 0.1703, "step": 4598 }, { "epoch": 2.43, "learning_rate": 1.0909224887375357e-05, "loss": 0.1852, "step": 4599 }, { "epoch": 2.43, "learning_rate": 1.0905809726359932e-05, "loss": 0.1279, "step": 4600 }, { "epoch": 2.44, "learning_rate": 1.0902394458819658e-05, "loss": 0.0868, "step": 4601 }, { "epoch": 2.44, "learning_rate": 1.0898979085156167e-05, "loss": 0.1036, "step": 4602 }, { "epoch": 2.44, "learning_rate": 1.0895563605771113e-05, "loss": 0.1691, "step": 4603 }, { "epoch": 2.44, "learning_rate": 1.0892148021066168e-05, "loss": 0.1238, "step": 4604 }, { "epoch": 2.44, "learning_rate": 1.088873233144301e-05, "loss": 0.1252, "step": 4605 }, { "epoch": 2.44, "learning_rate": 1.0885316537303328e-05, "loss": 0.1684, "step": 4606 }, { "epoch": 2.44, "learning_rate": 1.0881900639048823e-05, "loss": 0.1452, "step": 4607 }, { "epoch": 2.44, "learning_rate": 1.0878484637081217e-05, "loss": 0.1731, "step": 4608 }, { "epoch": 2.44, "learning_rate": 1.0875068531802232e-05, "loss": 0.179, "step": 4609 }, { "epoch": 2.44, "learning_rate": 1.0871652323613615e-05, "loss": 0.179, "step": 4610 }, { "epoch": 2.44, "learning_rate": 1.0868236012917112e-05, "loss": 0.086, "step": 4611 }, { "epoch": 2.44, "learning_rate": 1.0864819600114491e-05, "loss": 0.1154, "step": 4612 }, { "epoch": 2.44, "learning_rate": 1.0861403085607528e-05, "loss": 0.0893, "step": 4613 }, { "epoch": 2.44, "learning_rate": 1.085798646979801e-05, "loss": 0.0742, "step": 4614 }, { "epoch": 2.44, "learning_rate": 1.0854569753087733e-05, "loss": 0.1785, "step": 4615 }, { "epoch": 2.44, "learning_rate": 1.085115293587852e-05, "loss": 0.0581, "step": 4616 }, { "epoch": 2.44, "learning_rate": 1.0847736018572186e-05, "loss": 0.0666, "step": 4617 }, { "epoch": 2.44, "learning_rate": 1.0844319001570568e-05, "loss": 0.1134, "step": 4618 }, { "epoch": 2.44, "learning_rate": 1.0840901885275515e-05, "loss": 0.1687, "step": 4619 }, { "epoch": 2.45, "learning_rate": 1.0837484670088885e-05, "loss": 0.0828, "step": 4620 }, { "epoch": 2.45, "learning_rate": 1.0834067356412548e-05, "loss": 0.1603, "step": 4621 }, { "epoch": 2.45, "learning_rate": 1.0830649944648388e-05, "loss": 0.0839, "step": 4622 }, { "epoch": 2.45, "learning_rate": 1.0827232435198297e-05, "loss": 0.1241, "step": 4623 }, { "epoch": 2.45, "learning_rate": 1.0823814828464184e-05, "loss": 0.1469, "step": 4624 }, { "epoch": 2.45, "learning_rate": 1.082039712484796e-05, "loss": 0.0793, "step": 4625 }, { "epoch": 2.45, "learning_rate": 1.0816979324751557e-05, "loss": 0.1736, "step": 4626 }, { "epoch": 2.45, "learning_rate": 1.0813561428576914e-05, "loss": 0.1219, "step": 4627 }, { "epoch": 2.45, "learning_rate": 1.0810143436725982e-05, "loss": 0.1028, "step": 4628 }, { "epoch": 2.45, "learning_rate": 1.0806725349600724e-05, "loss": 0.1348, "step": 4629 }, { "epoch": 2.45, "learning_rate": 1.0803307167603105e-05, "loss": 0.1263, "step": 4630 }, { "epoch": 2.45, "learning_rate": 1.0799888891135122e-05, "loss": 0.1533, "step": 4631 }, { "epoch": 2.45, "learning_rate": 1.079647052059876e-05, "loss": 0.1721, "step": 4632 }, { "epoch": 2.45, "learning_rate": 1.0793052056396037e-05, "loss": 0.1391, "step": 4633 }, { "epoch": 2.45, "learning_rate": 1.0789633498928958e-05, "loss": 0.088, "step": 4634 }, { "epoch": 2.45, "learning_rate": 1.0786214848599557e-05, "loss": 0.0601, "step": 4635 }, { "epoch": 2.45, "learning_rate": 1.0782796105809872e-05, "loss": 0.1468, "step": 4636 }, { "epoch": 2.45, "learning_rate": 1.077937727096196e-05, "loss": 0.1493, "step": 4637 }, { "epoch": 2.45, "learning_rate": 1.0775958344457872e-05, "loss": 0.1497, "step": 4638 }, { "epoch": 2.46, "learning_rate": 1.0772539326699687e-05, "loss": 0.1452, "step": 4639 }, { "epoch": 2.46, "learning_rate": 1.0769120218089484e-05, "loss": 0.0673, "step": 4640 }, { "epoch": 2.46, "learning_rate": 1.0765701019029358e-05, "loss": 0.0626, "step": 4641 }, { "epoch": 2.46, "learning_rate": 1.0762281729921413e-05, "loss": 0.0912, "step": 4642 }, { "epoch": 2.46, "learning_rate": 1.0758862351167765e-05, "loss": 0.1017, "step": 4643 }, { "epoch": 2.46, "learning_rate": 1.0755442883170534e-05, "loss": 0.1144, "step": 4644 }, { "epoch": 2.46, "learning_rate": 1.0752023326331865e-05, "loss": 0.1086, "step": 4645 }, { "epoch": 2.46, "learning_rate": 1.074860368105389e-05, "loss": 0.1879, "step": 4646 }, { "epoch": 2.46, "learning_rate": 1.0745183947738778e-05, "loss": 0.1623, "step": 4647 }, { "epoch": 2.46, "learning_rate": 1.074176412678869e-05, "loss": 0.1071, "step": 4648 }, { "epoch": 2.46, "learning_rate": 1.0738344218605806e-05, "loss": 0.0847, "step": 4649 }, { "epoch": 2.46, "learning_rate": 1.073492422359231e-05, "loss": 0.1376, "step": 4650 }, { "epoch": 2.46, "learning_rate": 1.0731504142150402e-05, "loss": 0.0878, "step": 4651 }, { "epoch": 2.46, "learning_rate": 1.0728083974682286e-05, "loss": 0.1506, "step": 4652 }, { "epoch": 2.46, "learning_rate": 1.0724663721590188e-05, "loss": 0.1172, "step": 4653 }, { "epoch": 2.46, "learning_rate": 1.072124338327633e-05, "loss": 0.118, "step": 4654 }, { "epoch": 2.46, "learning_rate": 1.0717822960142947e-05, "loss": 0.1112, "step": 4655 }, { "epoch": 2.46, "learning_rate": 1.0714402452592295e-05, "loss": 0.1548, "step": 4656 }, { "epoch": 2.46, "learning_rate": 1.0710981861026624e-05, "loss": 0.188, "step": 4657 }, { "epoch": 2.47, "learning_rate": 1.0707561185848208e-05, "loss": 0.1096, "step": 4658 }, { "epoch": 2.47, "learning_rate": 1.070414042745932e-05, "loss": 0.1769, "step": 4659 }, { "epoch": 2.47, "learning_rate": 1.070071958626225e-05, "loss": 0.1086, "step": 4660 }, { "epoch": 2.47, "learning_rate": 1.0697298662659292e-05, "loss": 0.1509, "step": 4661 }, { "epoch": 2.47, "learning_rate": 1.0693877657052757e-05, "loss": 0.1173, "step": 4662 }, { "epoch": 2.47, "learning_rate": 1.0690456569844961e-05, "loss": 0.2208, "step": 4663 }, { "epoch": 2.47, "learning_rate": 1.0687035401438226e-05, "loss": 0.1917, "step": 4664 }, { "epoch": 2.47, "learning_rate": 1.0683614152234894e-05, "loss": 0.1731, "step": 4665 }, { "epoch": 2.47, "learning_rate": 1.0680192822637304e-05, "loss": 0.0885, "step": 4666 }, { "epoch": 2.47, "learning_rate": 1.0676771413047809e-05, "loss": 0.218, "step": 4667 }, { "epoch": 2.47, "learning_rate": 1.0673349923868779e-05, "loss": 0.1512, "step": 4668 }, { "epoch": 2.47, "learning_rate": 1.0669928355502585e-05, "loss": 0.1409, "step": 4669 }, { "epoch": 2.47, "learning_rate": 1.0666506708351612e-05, "loss": 0.1249, "step": 4670 }, { "epoch": 2.47, "learning_rate": 1.0663084982818245e-05, "loss": 0.1647, "step": 4671 }, { "epoch": 2.47, "learning_rate": 1.065966317930489e-05, "loss": 0.1319, "step": 4672 }, { "epoch": 2.47, "learning_rate": 1.065624129821396e-05, "loss": 0.1119, "step": 4673 }, { "epoch": 2.47, "learning_rate": 1.0652819339947869e-05, "loss": 0.1595, "step": 4674 }, { "epoch": 2.47, "learning_rate": 1.0649397304909049e-05, "loss": 0.1265, "step": 4675 }, { "epoch": 2.48, "learning_rate": 1.0645975193499933e-05, "loss": 0.0911, "step": 4676 }, { "epoch": 2.48, "learning_rate": 1.0642553006122973e-05, "loss": 0.1316, "step": 4677 }, { "epoch": 2.48, "learning_rate": 1.0639130743180621e-05, "loss": 0.1511, "step": 4678 }, { "epoch": 2.48, "learning_rate": 1.0635708405075342e-05, "loss": 0.1373, "step": 4679 }, { "epoch": 2.48, "learning_rate": 1.0632285992209609e-05, "loss": 0.0802, "step": 4680 }, { "epoch": 2.48, "learning_rate": 1.0628863504985905e-05, "loss": 0.1009, "step": 4681 }, { "epoch": 2.48, "learning_rate": 1.062544094380672e-05, "loss": 0.0932, "step": 4682 }, { "epoch": 2.48, "learning_rate": 1.062201830907455e-05, "loss": 0.1347, "step": 4683 }, { "epoch": 2.48, "learning_rate": 1.061859560119191e-05, "loss": 0.1803, "step": 4684 }, { "epoch": 2.48, "learning_rate": 1.0615172820561312e-05, "loss": 0.0709, "step": 4685 }, { "epoch": 2.48, "learning_rate": 1.0611749967585285e-05, "loss": 0.1585, "step": 4686 }, { "epoch": 2.48, "learning_rate": 1.0608327042666358e-05, "loss": 0.2369, "step": 4687 }, { "epoch": 2.48, "learning_rate": 1.060490404620707e-05, "loss": 0.1193, "step": 4688 }, { "epoch": 2.48, "learning_rate": 1.060148097860998e-05, "loss": 0.1568, "step": 4689 }, { "epoch": 2.48, "learning_rate": 1.0598057840277644e-05, "loss": 0.1372, "step": 4690 }, { "epoch": 2.48, "learning_rate": 1.0594634631612628e-05, "loss": 0.071, "step": 4691 }, { "epoch": 2.48, "learning_rate": 1.0591211353017508e-05, "loss": 0.1427, "step": 4692 }, { "epoch": 2.48, "learning_rate": 1.0587788004894867e-05, "loss": 0.1733, "step": 4693 }, { "epoch": 2.48, "learning_rate": 1.0584364587647298e-05, "loss": 0.1049, "step": 4694 }, { "epoch": 2.49, "learning_rate": 1.0580941101677402e-05, "loss": 0.088, "step": 4695 }, { "epoch": 2.49, "learning_rate": 1.0577517547387783e-05, "loss": 0.1191, "step": 4696 }, { "epoch": 2.49, "learning_rate": 1.0574093925181064e-05, "loss": 0.2173, "step": 4697 }, { "epoch": 2.49, "learning_rate": 1.0570670235459862e-05, "loss": 0.1097, "step": 4698 }, { "epoch": 2.49, "learning_rate": 1.0567246478626814e-05, "loss": 0.1269, "step": 4699 }, { "epoch": 2.49, "learning_rate": 1.0563822655084557e-05, "loss": 0.1035, "step": 4700 }, { "epoch": 2.49, "learning_rate": 1.056039876523574e-05, "loss": 0.1402, "step": 4701 }, { "epoch": 2.49, "learning_rate": 1.055697480948302e-05, "loss": 0.1676, "step": 4702 }, { "epoch": 2.49, "learning_rate": 1.0553550788229057e-05, "loss": 0.1082, "step": 4703 }, { "epoch": 2.49, "learning_rate": 1.0550126701876523e-05, "loss": 0.1302, "step": 4704 }, { "epoch": 2.49, "learning_rate": 1.0546702550828101e-05, "loss": 0.1723, "step": 4705 }, { "epoch": 2.49, "learning_rate": 1.0543278335486475e-05, "loss": 0.1269, "step": 4706 }, { "epoch": 2.49, "learning_rate": 1.0539854056254334e-05, "loss": 0.1541, "step": 4707 }, { "epoch": 2.49, "learning_rate": 1.0536429713534386e-05, "loss": 0.1163, "step": 4708 }, { "epoch": 2.49, "learning_rate": 1.0533005307729334e-05, "loss": 0.1089, "step": 4709 }, { "epoch": 2.49, "learning_rate": 1.0529580839241901e-05, "loss": 0.0628, "step": 4710 }, { "epoch": 2.49, "learning_rate": 1.0526156308474809e-05, "loss": 0.1084, "step": 4711 }, { "epoch": 2.49, "learning_rate": 1.0522731715830783e-05, "loss": 0.1576, "step": 4712 }, { "epoch": 2.49, "learning_rate": 1.0519307061712568e-05, "loss": 0.0568, "step": 4713 }, { "epoch": 2.5, "learning_rate": 1.0515882346522906e-05, "loss": 0.1741, "step": 4714 }, { "epoch": 2.5, "learning_rate": 1.0512457570664549e-05, "loss": 0.1348, "step": 4715 }, { "epoch": 2.5, "learning_rate": 1.0509032734540261e-05, "loss": 0.2729, "step": 4716 }, { "epoch": 2.5, "learning_rate": 1.0505607838552805e-05, "loss": 0.1626, "step": 4717 }, { "epoch": 2.5, "learning_rate": 1.0502182883104958e-05, "loss": 0.1566, "step": 4718 }, { "epoch": 2.5, "learning_rate": 1.0498757868599496e-05, "loss": 0.1415, "step": 4719 }, { "epoch": 2.5, "learning_rate": 1.0495332795439212e-05, "loss": 0.1534, "step": 4720 }, { "epoch": 2.5, "learning_rate": 1.04919076640269e-05, "loss": 0.1326, "step": 4721 }, { "epoch": 2.5, "learning_rate": 1.048848247476536e-05, "loss": 0.1509, "step": 4722 }, { "epoch": 2.5, "learning_rate": 1.04850572280574e-05, "loss": 0.1965, "step": 4723 }, { "epoch": 2.5, "learning_rate": 1.0481631924305838e-05, "loss": 0.1652, "step": 4724 }, { "epoch": 2.5, "learning_rate": 1.047820656391349e-05, "loss": 0.0717, "step": 4725 }, { "epoch": 2.5, "learning_rate": 1.0474781147283193e-05, "loss": 0.1576, "step": 4726 }, { "epoch": 2.5, "learning_rate": 1.0471355674817776e-05, "loss": 0.132, "step": 4727 }, { "epoch": 2.5, "learning_rate": 1.0467930146920083e-05, "loss": 0.1507, "step": 4728 }, { "epoch": 2.5, "learning_rate": 1.0464504563992958e-05, "loss": 0.1514, "step": 4729 }, { "epoch": 2.5, "learning_rate": 1.0461078926439262e-05, "loss": 0.0929, "step": 4730 }, { "epoch": 2.5, "learning_rate": 1.0457653234661855e-05, "loss": 0.1618, "step": 4731 }, { "epoch": 2.5, "learning_rate": 1.0454227489063603e-05, "loss": 0.1125, "step": 4732 }, { "epoch": 2.51, "learning_rate": 1.0450801690047379e-05, "loss": 0.0855, "step": 4733 }, { "epoch": 2.51, "learning_rate": 1.0447375838016062e-05, "loss": 0.113, "step": 4734 }, { "epoch": 2.51, "learning_rate": 1.0443949933372544e-05, "loss": 0.1274, "step": 4735 }, { "epoch": 2.51, "learning_rate": 1.0440523976519711e-05, "loss": 0.1697, "step": 4736 }, { "epoch": 2.51, "learning_rate": 1.0437097967860468e-05, "loss": 0.1743, "step": 4737 }, { "epoch": 2.51, "learning_rate": 1.0433671907797717e-05, "loss": 0.0924, "step": 4738 }, { "epoch": 2.51, "learning_rate": 1.0430245796734366e-05, "loss": 0.1415, "step": 4739 }, { "epoch": 2.51, "learning_rate": 1.0426819635073334e-05, "loss": 0.1337, "step": 4740 }, { "epoch": 2.51, "learning_rate": 1.0423393423217542e-05, "loss": 0.1391, "step": 4741 }, { "epoch": 2.51, "learning_rate": 1.0419967161569925e-05, "loss": 0.0957, "step": 4742 }, { "epoch": 2.51, "learning_rate": 1.0416540850533411e-05, "loss": 0.1541, "step": 4743 }, { "epoch": 2.51, "learning_rate": 1.0413114490510946e-05, "loss": 0.139, "step": 4744 }, { "epoch": 2.51, "learning_rate": 1.0409688081905467e-05, "loss": 0.1345, "step": 4745 }, { "epoch": 2.51, "learning_rate": 1.0406261625119933e-05, "loss": 0.1134, "step": 4746 }, { "epoch": 2.51, "learning_rate": 1.0402835120557302e-05, "loss": 0.0982, "step": 4747 }, { "epoch": 2.51, "learning_rate": 1.0399408568620535e-05, "loss": 0.1753, "step": 4748 }, { "epoch": 2.51, "learning_rate": 1.0395981969712598e-05, "loss": 0.1479, "step": 4749 }, { "epoch": 2.51, "learning_rate": 1.039255532423647e-05, "loss": 0.1318, "step": 4750 }, { "epoch": 2.51, "learning_rate": 1.0389128632595129e-05, "loss": 0.2145, "step": 4751 }, { "epoch": 2.52, "learning_rate": 1.0385701895191558e-05, "loss": 0.1726, "step": 4752 }, { "epoch": 2.52, "learning_rate": 1.038227511242875e-05, "loss": 0.1564, "step": 4753 }, { "epoch": 2.52, "learning_rate": 1.0378848284709704e-05, "loss": 0.1427, "step": 4754 }, { "epoch": 2.52, "learning_rate": 1.0375421412437411e-05, "loss": 0.1724, "step": 4755 }, { "epoch": 2.52, "learning_rate": 1.0371994496014886e-05, "loss": 0.0593, "step": 4756 }, { "epoch": 2.52, "learning_rate": 1.0368567535845137e-05, "loss": 0.0789, "step": 4757 }, { "epoch": 2.52, "learning_rate": 1.0365140532331186e-05, "loss": 0.1134, "step": 4758 }, { "epoch": 2.52, "learning_rate": 1.0361713485876049e-05, "loss": 0.1372, "step": 4759 }, { "epoch": 2.52, "learning_rate": 1.0358286396882753e-05, "loss": 0.0789, "step": 4760 }, { "epoch": 2.52, "learning_rate": 1.0354859265754328e-05, "loss": 0.1675, "step": 4761 }, { "epoch": 2.52, "learning_rate": 1.035143209289382e-05, "loss": 0.1822, "step": 4762 }, { "epoch": 2.52, "learning_rate": 1.0348004878704264e-05, "loss": 0.1114, "step": 4763 }, { "epoch": 2.52, "learning_rate": 1.0344577623588703e-05, "loss": 0.1619, "step": 4764 }, { "epoch": 2.52, "learning_rate": 1.0341150327950195e-05, "loss": 0.1587, "step": 4765 }, { "epoch": 2.52, "learning_rate": 1.0337722992191795e-05, "loss": 0.1541, "step": 4766 }, { "epoch": 2.52, "learning_rate": 1.033429561671656e-05, "loss": 0.1089, "step": 4767 }, { "epoch": 2.52, "learning_rate": 1.0330868201927563e-05, "loss": 0.1486, "step": 4768 }, { "epoch": 2.52, "learning_rate": 1.0327440748227863e-05, "loss": 0.2095, "step": 4769 }, { "epoch": 2.52, "learning_rate": 1.0324013256020549e-05, "loss": 0.0934, "step": 4770 }, { "epoch": 2.53, "learning_rate": 1.0320585725708684e-05, "loss": 0.1327, "step": 4771 }, { "epoch": 2.53, "learning_rate": 1.0317158157695363e-05, "loss": 0.2009, "step": 4772 }, { "epoch": 2.53, "learning_rate": 1.0313730552383674e-05, "loss": 0.093, "step": 4773 }, { "epoch": 2.53, "learning_rate": 1.0310302910176706e-05, "loss": 0.1341, "step": 4774 }, { "epoch": 2.53, "learning_rate": 1.0306875231477556e-05, "loss": 0.1192, "step": 4775 }, { "epoch": 2.53, "learning_rate": 1.0303447516689324e-05, "loss": 0.1318, "step": 4776 }, { "epoch": 2.53, "learning_rate": 1.0300019766215116e-05, "loss": 0.1578, "step": 4777 }, { "epoch": 2.53, "learning_rate": 1.029659198045805e-05, "loss": 0.1249, "step": 4778 }, { "epoch": 2.53, "learning_rate": 1.0293164159821229e-05, "loss": 0.1329, "step": 4779 }, { "epoch": 2.53, "learning_rate": 1.0289736304707774e-05, "loss": 0.1226, "step": 4780 }, { "epoch": 2.53, "learning_rate": 1.0286308415520806e-05, "loss": 0.0741, "step": 4781 }, { "epoch": 2.53, "learning_rate": 1.0282880492663456e-05, "loss": 0.1977, "step": 4782 }, { "epoch": 2.53, "learning_rate": 1.027945253653885e-05, "loss": 0.0706, "step": 4783 }, { "epoch": 2.53, "learning_rate": 1.027602454755012e-05, "loss": 0.1923, "step": 4784 }, { "epoch": 2.53, "learning_rate": 1.0272596526100406e-05, "loss": 0.0873, "step": 4785 }, { "epoch": 2.53, "learning_rate": 1.0269168472592852e-05, "loss": 0.1041, "step": 4786 }, { "epoch": 2.53, "learning_rate": 1.0265740387430598e-05, "loss": 0.1603, "step": 4787 }, { "epoch": 2.53, "learning_rate": 1.0262312271016797e-05, "loss": 0.1293, "step": 4788 }, { "epoch": 2.53, "learning_rate": 1.02588841237546e-05, "loss": 0.1205, "step": 4789 }, { "epoch": 2.54, "learning_rate": 1.0255455946047166e-05, "loss": 0.1147, "step": 4790 }, { "epoch": 2.54, "learning_rate": 1.0252027738297649e-05, "loss": 0.1201, "step": 4791 }, { "epoch": 2.54, "learning_rate": 1.0248599500909218e-05, "loss": 0.1985, "step": 4792 }, { "epoch": 2.54, "learning_rate": 1.0245171234285036e-05, "loss": 0.0956, "step": 4793 }, { "epoch": 2.54, "learning_rate": 1.0241742938828278e-05, "loss": 0.1106, "step": 4794 }, { "epoch": 2.54, "learning_rate": 1.0238314614942113e-05, "loss": 0.2028, "step": 4795 }, { "epoch": 2.54, "learning_rate": 1.023488626302972e-05, "loss": 0.1158, "step": 4796 }, { "epoch": 2.54, "learning_rate": 1.0231457883494281e-05, "loss": 0.0689, "step": 4797 }, { "epoch": 2.54, "learning_rate": 1.0228029476738974e-05, "loss": 0.1754, "step": 4798 }, { "epoch": 2.54, "learning_rate": 1.0224601043166993e-05, "loss": 0.0833, "step": 4799 }, { "epoch": 2.54, "learning_rate": 1.0221172583181525e-05, "loss": 0.1412, "step": 4800 }, { "epoch": 2.54, "learning_rate": 1.021774409718576e-05, "loss": 0.2158, "step": 4801 }, { "epoch": 2.54, "learning_rate": 1.0214315585582898e-05, "loss": 0.2215, "step": 4802 }, { "epoch": 2.54, "learning_rate": 1.0210887048776137e-05, "loss": 0.1282, "step": 4803 }, { "epoch": 2.54, "learning_rate": 1.020745848716868e-05, "loss": 0.0987, "step": 4804 }, { "epoch": 2.54, "learning_rate": 1.020402990116373e-05, "loss": 0.1113, "step": 4805 }, { "epoch": 2.54, "learning_rate": 1.0200601291164499e-05, "loss": 0.1359, "step": 4806 }, { "epoch": 2.54, "learning_rate": 1.019717265757419e-05, "loss": 0.1367, "step": 4807 }, { "epoch": 2.54, "learning_rate": 1.0193744000796024e-05, "loss": 0.1516, "step": 4808 }, { "epoch": 2.55, "learning_rate": 1.0190315321233216e-05, "loss": 0.1535, "step": 4809 }, { "epoch": 2.55, "learning_rate": 1.018688661928898e-05, "loss": 0.1627, "step": 4810 }, { "epoch": 2.55, "learning_rate": 1.0183457895366547e-05, "loss": 0.088, "step": 4811 }, { "epoch": 2.55, "learning_rate": 1.018002914986913e-05, "loss": 0.1359, "step": 4812 }, { "epoch": 2.55, "learning_rate": 1.0176600383199957e-05, "loss": 0.1749, "step": 4813 }, { "epoch": 2.55, "learning_rate": 1.0173171595762265e-05, "loss": 0.1544, "step": 4814 }, { "epoch": 2.55, "learning_rate": 1.0169742787959284e-05, "loss": 0.1311, "step": 4815 }, { "epoch": 2.55, "learning_rate": 1.0166313960194244e-05, "loss": 0.1487, "step": 4816 }, { "epoch": 2.55, "learning_rate": 1.016288511287038e-05, "loss": 0.0774, "step": 4817 }, { "epoch": 2.55, "learning_rate": 1.0159456246390933e-05, "loss": 0.1816, "step": 4818 }, { "epoch": 2.55, "learning_rate": 1.0156027361159146e-05, "loss": 0.1207, "step": 4819 }, { "epoch": 2.55, "learning_rate": 1.015259845757826e-05, "loss": 0.1035, "step": 4820 }, { "epoch": 2.55, "learning_rate": 1.0149169536051522e-05, "loss": 0.1275, "step": 4821 }, { "epoch": 2.55, "learning_rate": 1.0145740596982175e-05, "loss": 0.1502, "step": 4822 }, { "epoch": 2.55, "learning_rate": 1.0142311640773472e-05, "loss": 0.1841, "step": 4823 }, { "epoch": 2.55, "learning_rate": 1.0138882667828665e-05, "loss": 0.2027, "step": 4824 }, { "epoch": 2.55, "learning_rate": 1.0135453678551004e-05, "loss": 0.1966, "step": 4825 }, { "epoch": 2.55, "learning_rate": 1.0132024673343747e-05, "loss": 0.1526, "step": 4826 }, { "epoch": 2.55, "learning_rate": 1.0128595652610154e-05, "loss": 0.1254, "step": 4827 }, { "epoch": 2.56, "learning_rate": 1.0125166616753474e-05, "loss": 0.16, "step": 4828 }, { "epoch": 2.56, "learning_rate": 1.012173756617698e-05, "loss": 0.1501, "step": 4829 }, { "epoch": 2.56, "learning_rate": 1.0118308501283927e-05, "loss": 0.18, "step": 4830 }, { "epoch": 2.56, "learning_rate": 1.0114879422477584e-05, "loss": 0.0841, "step": 4831 }, { "epoch": 2.56, "learning_rate": 1.011145033016121e-05, "loss": 0.1005, "step": 4832 }, { "epoch": 2.56, "learning_rate": 1.0108021224738076e-05, "loss": 0.2019, "step": 4833 }, { "epoch": 2.56, "learning_rate": 1.010459210661145e-05, "loss": 0.0656, "step": 4834 }, { "epoch": 2.56, "learning_rate": 1.0101162976184611e-05, "loss": 0.1446, "step": 4835 }, { "epoch": 2.56, "learning_rate": 1.009773383386082e-05, "loss": 0.12, "step": 4836 }, { "epoch": 2.56, "learning_rate": 1.0094304680043351e-05, "loss": 0.1044, "step": 4837 }, { "epoch": 2.56, "learning_rate": 1.0090875515135486e-05, "loss": 0.0971, "step": 4838 }, { "epoch": 2.56, "learning_rate": 1.0087446339540494e-05, "loss": 0.1585, "step": 4839 }, { "epoch": 2.56, "learning_rate": 1.0084017153661657e-05, "loss": 0.0839, "step": 4840 }, { "epoch": 2.56, "learning_rate": 1.0080587957902248e-05, "loss": 0.1575, "step": 4841 }, { "epoch": 2.56, "learning_rate": 1.0077158752665553e-05, "loss": 0.1718, "step": 4842 }, { "epoch": 2.56, "learning_rate": 1.007372953835485e-05, "loss": 0.0515, "step": 4843 }, { "epoch": 2.56, "learning_rate": 1.0070300315373418e-05, "loss": 0.076, "step": 4844 }, { "epoch": 2.56, "learning_rate": 1.0066871084124543e-05, "loss": 0.0919, "step": 4845 }, { "epoch": 2.57, "learning_rate": 1.0063441845011508e-05, "loss": 0.1934, "step": 4846 }, { "epoch": 2.57, "learning_rate": 1.0060012598437601e-05, "loss": 0.1695, "step": 4847 }, { "epoch": 2.57, "learning_rate": 1.0056583344806101e-05, "loss": 0.1914, "step": 4848 }, { "epoch": 2.57, "learning_rate": 1.0053154084520298e-05, "loss": 0.0965, "step": 4849 }, { "epoch": 2.57, "learning_rate": 1.0049724817983477e-05, "loss": 0.0622, "step": 4850 }, { "epoch": 2.57, "learning_rate": 1.0046295545598931e-05, "loss": 0.1942, "step": 4851 }, { "epoch": 2.57, "learning_rate": 1.0042866267769946e-05, "loss": 0.1019, "step": 4852 }, { "epoch": 2.57, "learning_rate": 1.0039436984899808e-05, "loss": 0.1325, "step": 4853 }, { "epoch": 2.57, "learning_rate": 1.0036007697391812e-05, "loss": 0.0849, "step": 4854 }, { "epoch": 2.57, "learning_rate": 1.0032578405649245e-05, "loss": 0.1127, "step": 4855 }, { "epoch": 2.57, "learning_rate": 1.0029149110075399e-05, "loss": 0.1163, "step": 4856 }, { "epoch": 2.57, "learning_rate": 1.0025719811073563e-05, "loss": 0.143, "step": 4857 }, { "epoch": 2.57, "learning_rate": 1.0022290509047033e-05, "loss": 0.0616, "step": 4858 }, { "epoch": 2.57, "learning_rate": 1.0018861204399098e-05, "loss": 0.1503, "step": 4859 }, { "epoch": 2.57, "learning_rate": 1.001543189753305e-05, "loss": 0.0633, "step": 4860 }, { "epoch": 2.57, "learning_rate": 1.0012002588852186e-05, "loss": 0.1554, "step": 4861 }, { "epoch": 2.57, "learning_rate": 1.0008573278759794e-05, "loss": 0.1507, "step": 4862 }, { "epoch": 2.57, "learning_rate": 1.0005143967659173e-05, "loss": 0.1129, "step": 4863 }, { "epoch": 2.57, "learning_rate": 1.0001714655953606e-05, "loss": 0.107, "step": 4864 }, { "epoch": 2.58, "learning_rate": 9.998285344046395e-06, "loss": 0.159, "step": 4865 }, { "epoch": 2.58, "learning_rate": 9.99485603234083e-06, "loss": 0.1493, "step": 4866 }, { "epoch": 2.58, "learning_rate": 9.991426721240206e-06, "loss": 0.2159, "step": 4867 }, { "epoch": 2.58, "learning_rate": 9.987997411147816e-06, "loss": 0.1175, "step": 4868 }, { "epoch": 2.58, "learning_rate": 9.984568102466951e-06, "loss": 0.1102, "step": 4869 }, { "epoch": 2.58, "learning_rate": 9.981138795600906e-06, "loss": 0.1509, "step": 4870 }, { "epoch": 2.58, "learning_rate": 9.97770949095297e-06, "loss": 0.1461, "step": 4871 }, { "epoch": 2.58, "learning_rate": 9.97428018892644e-06, "loss": 0.1577, "step": 4872 }, { "epoch": 2.58, "learning_rate": 9.970850889924607e-06, "loss": 0.1014, "step": 4873 }, { "epoch": 2.58, "learning_rate": 9.967421594350759e-06, "loss": 0.0779, "step": 4874 }, { "epoch": 2.58, "learning_rate": 9.963992302608192e-06, "loss": 0.0559, "step": 4875 }, { "epoch": 2.58, "learning_rate": 9.960563015100191e-06, "loss": 0.1369, "step": 4876 }, { "epoch": 2.58, "learning_rate": 9.957133732230059e-06, "loss": 0.168, "step": 4877 }, { "epoch": 2.58, "learning_rate": 9.953704454401074e-06, "loss": 0.0747, "step": 4878 }, { "epoch": 2.58, "learning_rate": 9.950275182016525e-06, "loss": 0.1418, "step": 4879 }, { "epoch": 2.58, "learning_rate": 9.946845915479705e-06, "loss": 0.1925, "step": 4880 }, { "epoch": 2.58, "learning_rate": 9.943416655193902e-06, "loss": 0.1744, "step": 4881 }, { "epoch": 2.58, "learning_rate": 9.939987401562402e-06, "loss": 0.191, "step": 4882 }, { "epoch": 2.58, "learning_rate": 9.936558154988492e-06, "loss": 0.1586, "step": 4883 }, { "epoch": 2.59, "learning_rate": 9.933128915875458e-06, "loss": 0.093, "step": 4884 }, { "epoch": 2.59, "learning_rate": 9.929699684626585e-06, "loss": 0.1075, "step": 4885 }, { "epoch": 2.59, "learning_rate": 9.926270461645155e-06, "loss": 0.1165, "step": 4886 }, { "epoch": 2.59, "learning_rate": 9.92284124733445e-06, "loss": 0.1445, "step": 4887 }, { "epoch": 2.59, "learning_rate": 9.919412042097754e-06, "loss": 0.1292, "step": 4888 }, { "epoch": 2.59, "learning_rate": 9.915982846338348e-06, "loss": 0.0582, "step": 4889 }, { "epoch": 2.59, "learning_rate": 9.91255366045951e-06, "loss": 0.1236, "step": 4890 }, { "epoch": 2.59, "learning_rate": 9.909124484864519e-06, "loss": 0.1784, "step": 4891 }, { "epoch": 2.59, "learning_rate": 9.905695319956649e-06, "loss": 0.1585, "step": 4892 }, { "epoch": 2.59, "learning_rate": 9.902266166139186e-06, "loss": 0.1138, "step": 4893 }, { "epoch": 2.59, "learning_rate": 9.898837023815396e-06, "loss": 0.1086, "step": 4894 }, { "epoch": 2.59, "learning_rate": 9.89540789338855e-06, "loss": 0.1262, "step": 4895 }, { "epoch": 2.59, "learning_rate": 9.891978775261926e-06, "loss": 0.1048, "step": 4896 }, { "epoch": 2.59, "learning_rate": 9.888549669838792e-06, "loss": 0.2172, "step": 4897 }, { "epoch": 2.59, "learning_rate": 9.88512057752242e-06, "loss": 0.1706, "step": 4898 }, { "epoch": 2.59, "learning_rate": 9.881691498716073e-06, "loss": 0.1631, "step": 4899 }, { "epoch": 2.59, "learning_rate": 9.87826243382302e-06, "loss": 0.1674, "step": 4900 }, { "epoch": 2.59, "learning_rate": 9.874833383246528e-06, "loss": 0.1515, "step": 4901 }, { "epoch": 2.59, "learning_rate": 9.871404347389851e-06, "loss": 0.0822, "step": 4902 }, { "epoch": 2.6, "learning_rate": 9.867975326656255e-06, "loss": 0.1167, "step": 4903 }, { "epoch": 2.6, "learning_rate": 9.864546321448998e-06, "loss": 0.133, "step": 4904 }, { "epoch": 2.6, "learning_rate": 9.861117332171339e-06, "loss": 0.0923, "step": 4905 }, { "epoch": 2.6, "learning_rate": 9.85768835922653e-06, "loss": 0.1792, "step": 4906 }, { "epoch": 2.6, "learning_rate": 9.854259403017825e-06, "loss": 0.2188, "step": 4907 }, { "epoch": 2.6, "learning_rate": 9.85083046394848e-06, "loss": 0.1095, "step": 4908 }, { "epoch": 2.6, "learning_rate": 9.847401542421745e-06, "loss": 0.1036, "step": 4909 }, { "epoch": 2.6, "learning_rate": 9.843972638840859e-06, "loss": 0.1074, "step": 4910 }, { "epoch": 2.6, "learning_rate": 9.840543753609068e-06, "loss": 0.1195, "step": 4911 }, { "epoch": 2.6, "learning_rate": 9.837114887129623e-06, "loss": 0.193, "step": 4912 }, { "epoch": 2.6, "learning_rate": 9.833686039805762e-06, "loss": 0.1266, "step": 4913 }, { "epoch": 2.6, "learning_rate": 9.83025721204072e-06, "loss": 0.1294, "step": 4914 }, { "epoch": 2.6, "learning_rate": 9.826828404237735e-06, "loss": 0.1336, "step": 4915 }, { "epoch": 2.6, "learning_rate": 9.823399616800043e-06, "loss": 0.1755, "step": 4916 }, { "epoch": 2.6, "learning_rate": 9.819970850130876e-06, "loss": 0.0798, "step": 4917 }, { "epoch": 2.6, "learning_rate": 9.81654210463346e-06, "loss": 0.2065, "step": 4918 }, { "epoch": 2.6, "learning_rate": 9.813113380711023e-06, "loss": 0.1046, "step": 4919 }, { "epoch": 2.6, "learning_rate": 9.809684678766788e-06, "loss": 0.0975, "step": 4920 }, { "epoch": 2.6, "learning_rate": 9.806255999203978e-06, "loss": 0.2046, "step": 4921 }, { "epoch": 2.61, "learning_rate": 9.802827342425811e-06, "loss": 0.195, "step": 4922 }, { "epoch": 2.61, "learning_rate": 9.799398708835503e-06, "loss": 0.157, "step": 4923 }, { "epoch": 2.61, "learning_rate": 9.79597009883627e-06, "loss": 0.1803, "step": 4924 }, { "epoch": 2.61, "learning_rate": 9.792541512831324e-06, "loss": 0.1398, "step": 4925 }, { "epoch": 2.61, "learning_rate": 9.789112951223867e-06, "loss": 0.1294, "step": 4926 }, { "epoch": 2.61, "learning_rate": 9.785684414417105e-06, "loss": 0.0944, "step": 4927 }, { "epoch": 2.61, "learning_rate": 9.782255902814242e-06, "loss": 0.1313, "step": 4928 }, { "epoch": 2.61, "learning_rate": 9.77882741681848e-06, "loss": 0.1682, "step": 4929 }, { "epoch": 2.61, "learning_rate": 9.775398956833009e-06, "loss": 0.1079, "step": 4930 }, { "epoch": 2.61, "learning_rate": 9.771970523261027e-06, "loss": 0.1321, "step": 4931 }, { "epoch": 2.61, "learning_rate": 9.768542116505722e-06, "loss": 0.0985, "step": 4932 }, { "epoch": 2.61, "learning_rate": 9.765113736970282e-06, "loss": 0.1079, "step": 4933 }, { "epoch": 2.61, "learning_rate": 9.761685385057892e-06, "loss": 0.1399, "step": 4934 }, { "epoch": 2.61, "learning_rate": 9.758257061171727e-06, "loss": 0.1443, "step": 4935 }, { "epoch": 2.61, "learning_rate": 9.754828765714967e-06, "loss": 0.0717, "step": 4936 }, { "epoch": 2.61, "learning_rate": 9.751400499090786e-06, "loss": 0.1101, "step": 4937 }, { "epoch": 2.61, "learning_rate": 9.747972261702353e-06, "loss": 0.1242, "step": 4938 }, { "epoch": 2.61, "learning_rate": 9.744544053952838e-06, "loss": 0.1079, "step": 4939 }, { "epoch": 2.61, "learning_rate": 9.741115876245401e-06, "loss": 0.1978, "step": 4940 }, { "epoch": 2.62, "learning_rate": 9.737687728983203e-06, "loss": 0.1923, "step": 4941 }, { "epoch": 2.62, "learning_rate": 9.734259612569405e-06, "loss": 0.1567, "step": 4942 }, { "epoch": 2.62, "learning_rate": 9.730831527407152e-06, "loss": 0.132, "step": 4943 }, { "epoch": 2.62, "learning_rate": 9.727403473899595e-06, "loss": 0.1982, "step": 4944 }, { "epoch": 2.62, "learning_rate": 9.723975452449883e-06, "loss": 0.1169, "step": 4945 }, { "epoch": 2.62, "learning_rate": 9.720547463461155e-06, "loss": 0.108, "step": 4946 }, { "epoch": 2.62, "learning_rate": 9.717119507336546e-06, "loss": 0.1893, "step": 4947 }, { "epoch": 2.62, "learning_rate": 9.713691584479195e-06, "loss": 0.1495, "step": 4948 }, { "epoch": 2.62, "learning_rate": 9.710263695292227e-06, "loss": 0.1154, "step": 4949 }, { "epoch": 2.62, "learning_rate": 9.706835840178776e-06, "loss": 0.1551, "step": 4950 }, { "epoch": 2.62, "learning_rate": 9.703408019541956e-06, "loss": 0.112, "step": 4951 }, { "epoch": 2.62, "learning_rate": 9.699980233784885e-06, "loss": 0.1271, "step": 4952 }, { "epoch": 2.62, "learning_rate": 9.696552483310677e-06, "loss": 0.125, "step": 4953 }, { "epoch": 2.62, "learning_rate": 9.693124768522447e-06, "loss": 0.1102, "step": 4954 }, { "epoch": 2.62, "learning_rate": 9.689697089823297e-06, "loss": 0.1472, "step": 4955 }, { "epoch": 2.62, "learning_rate": 9.686269447616327e-06, "loss": 0.1025, "step": 4956 }, { "epoch": 2.62, "learning_rate": 9.682841842304637e-06, "loss": 0.1253, "step": 4957 }, { "epoch": 2.62, "learning_rate": 9.679414274291319e-06, "loss": 0.1441, "step": 4958 }, { "epoch": 2.62, "learning_rate": 9.675986743979458e-06, "loss": 0.071, "step": 4959 }, { "epoch": 2.63, "learning_rate": 9.672559251772138e-06, "loss": 0.1314, "step": 4960 }, { "epoch": 2.63, "learning_rate": 9.669131798072442e-06, "loss": 0.1084, "step": 4961 }, { "epoch": 2.63, "learning_rate": 9.665704383283443e-06, "loss": 0.1077, "step": 4962 }, { "epoch": 2.63, "learning_rate": 9.662277007808208e-06, "loss": 0.1533, "step": 4963 }, { "epoch": 2.63, "learning_rate": 9.658849672049807e-06, "loss": 0.0678, "step": 4964 }, { "epoch": 2.63, "learning_rate": 9.655422376411297e-06, "loss": 0.1136, "step": 4965 }, { "epoch": 2.63, "learning_rate": 9.651995121295743e-06, "loss": 0.163, "step": 4966 }, { "epoch": 2.63, "learning_rate": 9.648567907106185e-06, "loss": 0.1404, "step": 4967 }, { "epoch": 2.63, "learning_rate": 9.645140734245673e-06, "loss": 0.0631, "step": 4968 }, { "epoch": 2.63, "learning_rate": 9.64171360311725e-06, "loss": 0.0933, "step": 4969 }, { "epoch": 2.63, "learning_rate": 9.638286514123954e-06, "loss": 0.0843, "step": 4970 }, { "epoch": 2.63, "learning_rate": 9.634859467668816e-06, "loss": 0.1728, "step": 4971 }, { "epoch": 2.63, "learning_rate": 9.631432464154861e-06, "loss": 0.1695, "step": 4972 }, { "epoch": 2.63, "learning_rate": 9.628005503985114e-06, "loss": 0.1096, "step": 4973 }, { "epoch": 2.63, "learning_rate": 9.624578587562592e-06, "loss": 0.0925, "step": 4974 }, { "epoch": 2.63, "learning_rate": 9.6211517152903e-06, "loss": 0.2176, "step": 4975 }, { "epoch": 2.63, "learning_rate": 9.617724887571251e-06, "loss": 0.1025, "step": 4976 }, { "epoch": 2.63, "learning_rate": 9.614298104808443e-06, "loss": 0.1982, "step": 4977 }, { "epoch": 2.63, "learning_rate": 9.610871367404873e-06, "loss": 0.1527, "step": 4978 }, { "epoch": 2.64, "learning_rate": 9.607444675763532e-06, "loss": 0.1329, "step": 4979 }, { "epoch": 2.64, "learning_rate": 9.604018030287402e-06, "loss": 0.1674, "step": 4980 }, { "epoch": 2.64, "learning_rate": 9.600591431379466e-06, "loss": 0.1001, "step": 4981 }, { "epoch": 2.64, "learning_rate": 9.597164879442703e-06, "loss": 0.141, "step": 4982 }, { "epoch": 2.64, "learning_rate": 9.59373837488007e-06, "loss": 0.0666, "step": 4983 }, { "epoch": 2.64, "learning_rate": 9.590311918094536e-06, "loss": 0.1259, "step": 4984 }, { "epoch": 2.64, "learning_rate": 9.58688550948906e-06, "loss": 0.096, "step": 4985 }, { "epoch": 2.64, "learning_rate": 9.58345914946659e-06, "loss": 0.0896, "step": 4986 }, { "epoch": 2.64, "learning_rate": 9.580032838430076e-06, "loss": 0.1305, "step": 4987 }, { "epoch": 2.64, "learning_rate": 9.576606576782458e-06, "loss": 0.0915, "step": 4988 }, { "epoch": 2.64, "learning_rate": 9.573180364926668e-06, "loss": 0.1414, "step": 4989 }, { "epoch": 2.64, "learning_rate": 9.569754203265639e-06, "loss": 0.1473, "step": 4990 }, { "epoch": 2.64, "learning_rate": 9.566328092202288e-06, "loss": 0.102, "step": 4991 }, { "epoch": 2.64, "learning_rate": 9.562902032139535e-06, "loss": 0.0832, "step": 4992 }, { "epoch": 2.64, "learning_rate": 9.55947602348029e-06, "loss": 0.1749, "step": 4993 }, { "epoch": 2.64, "learning_rate": 9.556050066627457e-06, "loss": 0.121, "step": 4994 }, { "epoch": 2.64, "learning_rate": 9.55262416198394e-06, "loss": 0.0757, "step": 4995 }, { "epoch": 2.64, "learning_rate": 9.549198309952621e-06, "loss": 0.1688, "step": 4996 }, { "epoch": 2.64, "learning_rate": 9.545772510936397e-06, "loss": 0.1584, "step": 4997 }, { "epoch": 2.65, "learning_rate": 9.54234676533815e-06, "loss": 0.1526, "step": 4998 }, { "epoch": 2.65, "learning_rate": 9.53892107356074e-06, "loss": 0.1245, "step": 4999 }, { "epoch": 2.65, "learning_rate": 9.535495436007043e-06, "loss": 0.0843, "step": 5000 }, { "epoch": 2.65, "learning_rate": 9.532069853079922e-06, "loss": 0.1006, "step": 5001 }, { "epoch": 2.65, "learning_rate": 9.528644325182228e-06, "loss": 0.123, "step": 5002 }, { "epoch": 2.65, "learning_rate": 9.52521885271681e-06, "loss": 0.1078, "step": 5003 }, { "epoch": 2.65, "learning_rate": 9.52179343608651e-06, "loss": 0.1157, "step": 5004 }, { "epoch": 2.65, "learning_rate": 9.518368075694165e-06, "loss": 0.1731, "step": 5005 }, { "epoch": 2.65, "learning_rate": 9.514942771942603e-06, "loss": 0.1346, "step": 5006 }, { "epoch": 2.65, "learning_rate": 9.511517525234644e-06, "loss": 0.2658, "step": 5007 }, { "epoch": 2.65, "learning_rate": 9.508092335973104e-06, "loss": 0.1289, "step": 5008 }, { "epoch": 2.65, "learning_rate": 9.50466720456079e-06, "loss": 0.1187, "step": 5009 }, { "epoch": 2.65, "learning_rate": 9.501242131400507e-06, "loss": 0.1435, "step": 5010 }, { "epoch": 2.65, "learning_rate": 9.497817116895043e-06, "loss": 0.1375, "step": 5011 }, { "epoch": 2.65, "learning_rate": 9.494392161447195e-06, "loss": 0.1194, "step": 5012 }, { "epoch": 2.65, "learning_rate": 9.49096726545974e-06, "loss": 0.1968, "step": 5013 }, { "epoch": 2.65, "learning_rate": 9.487542429335456e-06, "loss": 0.0618, "step": 5014 }, { "epoch": 2.65, "learning_rate": 9.484117653477097e-06, "loss": 0.15, "step": 5015 }, { "epoch": 2.66, "learning_rate": 9.480692938287436e-06, "loss": 0.2172, "step": 5016 }, { "epoch": 2.66, "learning_rate": 9.47726828416922e-06, "loss": 0.1343, "step": 5017 }, { "epoch": 2.66, "learning_rate": 9.473843691525196e-06, "loss": 0.1655, "step": 5018 }, { "epoch": 2.66, "learning_rate": 9.4704191607581e-06, "loss": 0.1352, "step": 5019 }, { "epoch": 2.66, "learning_rate": 9.466994692270666e-06, "loss": 0.078, "step": 5020 }, { "epoch": 2.66, "learning_rate": 9.463570286465616e-06, "loss": 0.1793, "step": 5021 }, { "epoch": 2.66, "learning_rate": 9.46014594374567e-06, "loss": 0.1048, "step": 5022 }, { "epoch": 2.66, "learning_rate": 9.45672166451353e-06, "loss": 0.1431, "step": 5023 }, { "epoch": 2.66, "learning_rate": 9.453297449171902e-06, "loss": 0.1048, "step": 5024 }, { "epoch": 2.66, "learning_rate": 9.449873298123478e-06, "loss": 0.0736, "step": 5025 }, { "epoch": 2.66, "learning_rate": 9.446449211770948e-06, "loss": 0.176, "step": 5026 }, { "epoch": 2.66, "learning_rate": 9.443025190516982e-06, "loss": 0.064, "step": 5027 }, { "epoch": 2.66, "learning_rate": 9.439601234764261e-06, "loss": 0.1756, "step": 5028 }, { "epoch": 2.66, "learning_rate": 9.436177344915445e-06, "loss": 0.1323, "step": 5029 }, { "epoch": 2.66, "learning_rate": 9.432753521373193e-06, "loss": 0.1025, "step": 5030 }, { "epoch": 2.66, "learning_rate": 9.42932976454014e-06, "loss": 0.1858, "step": 5031 }, { "epoch": 2.66, "learning_rate": 9.42590607481894e-06, "loss": 0.1376, "step": 5032 }, { "epoch": 2.66, "learning_rate": 9.422482452612219e-06, "loss": 0.1546, "step": 5033 }, { "epoch": 2.66, "learning_rate": 9.419058898322601e-06, "loss": 0.1012, "step": 5034 }, { "epoch": 2.67, "learning_rate": 9.415635412352704e-06, "loss": 0.112, "step": 5035 }, { "epoch": 2.67, "learning_rate": 9.412211995105134e-06, "loss": 0.1263, "step": 5036 }, { "epoch": 2.67, "learning_rate": 9.408788646982496e-06, "loss": 0.1213, "step": 5037 }, { "epoch": 2.67, "learning_rate": 9.405365368387377e-06, "loss": 0.0846, "step": 5038 }, { "epoch": 2.67, "learning_rate": 9.40194215972236e-06, "loss": 0.1283, "step": 5039 }, { "epoch": 2.67, "learning_rate": 9.398519021390023e-06, "loss": 0.0925, "step": 5040 }, { "epoch": 2.67, "learning_rate": 9.395095953792933e-06, "loss": 0.1367, "step": 5041 }, { "epoch": 2.67, "learning_rate": 9.391672957333646e-06, "loss": 0.1333, "step": 5042 }, { "epoch": 2.67, "learning_rate": 9.388250032414719e-06, "loss": 0.0968, "step": 5043 }, { "epoch": 2.67, "learning_rate": 9.384827179438688e-06, "loss": 0.1995, "step": 5044 }, { "epoch": 2.67, "learning_rate": 9.38140439880809e-06, "loss": 0.1486, "step": 5045 }, { "epoch": 2.67, "learning_rate": 9.377981690925451e-06, "loss": 0.1872, "step": 5046 }, { "epoch": 2.67, "learning_rate": 9.374559056193285e-06, "loss": 0.1467, "step": 5047 }, { "epoch": 2.67, "learning_rate": 9.371136495014099e-06, "loss": 0.1772, "step": 5048 }, { "epoch": 2.67, "learning_rate": 9.367714007790394e-06, "loss": 0.1268, "step": 5049 }, { "epoch": 2.67, "learning_rate": 9.364291594924661e-06, "loss": 0.1694, "step": 5050 }, { "epoch": 2.67, "learning_rate": 9.360869256819382e-06, "loss": 0.1231, "step": 5051 }, { "epoch": 2.67, "learning_rate": 9.35744699387703e-06, "loss": 0.1126, "step": 5052 }, { "epoch": 2.67, "learning_rate": 9.354024806500067e-06, "loss": 0.1052, "step": 5053 }, { "epoch": 2.68, "learning_rate": 9.350602695090953e-06, "loss": 0.1983, "step": 5054 }, { "epoch": 2.68, "learning_rate": 9.347180660052136e-06, "loss": 0.0952, "step": 5055 }, { "epoch": 2.68, "learning_rate": 9.343758701786045e-06, "loss": 0.1598, "step": 5056 }, { "epoch": 2.68, "learning_rate": 9.340336820695111e-06, "loss": 0.1298, "step": 5057 }, { "epoch": 2.68, "learning_rate": 9.336915017181756e-06, "loss": 0.0617, "step": 5058 }, { "epoch": 2.68, "learning_rate": 9.333493291648393e-06, "loss": 0.1291, "step": 5059 }, { "epoch": 2.68, "learning_rate": 9.330071644497417e-06, "loss": 0.1157, "step": 5060 }, { "epoch": 2.68, "learning_rate": 9.326650076131223e-06, "loss": 0.1376, "step": 5061 }, { "epoch": 2.68, "learning_rate": 9.323228586952193e-06, "loss": 0.1729, "step": 5062 }, { "epoch": 2.68, "learning_rate": 9.319807177362703e-06, "loss": 0.0882, "step": 5063 }, { "epoch": 2.68, "learning_rate": 9.316385847765112e-06, "loss": 0.1271, "step": 5064 }, { "epoch": 2.68, "learning_rate": 9.312964598561777e-06, "loss": 0.0562, "step": 5065 }, { "epoch": 2.68, "learning_rate": 9.309543430155042e-06, "loss": 0.2074, "step": 5066 }, { "epoch": 2.68, "learning_rate": 9.306122342947246e-06, "loss": 0.0875, "step": 5067 }, { "epoch": 2.68, "learning_rate": 9.30270133734071e-06, "loss": 0.2443, "step": 5068 }, { "epoch": 2.68, "learning_rate": 9.299280413737752e-06, "loss": 0.1124, "step": 5069 }, { "epoch": 2.68, "learning_rate": 9.295859572540682e-06, "loss": 0.1381, "step": 5070 }, { "epoch": 2.68, "learning_rate": 9.292438814151799e-06, "loss": 0.1203, "step": 5071 }, { "epoch": 2.68, "learning_rate": 9.289018138973381e-06, "loss": 0.1228, "step": 5072 }, { "epoch": 2.69, "learning_rate": 9.28559754740771e-06, "loss": 0.1655, "step": 5073 }, { "epoch": 2.69, "learning_rate": 9.282177039857055e-06, "loss": 0.1633, "step": 5074 }, { "epoch": 2.69, "learning_rate": 9.278756616723674e-06, "loss": 0.1336, "step": 5075 }, { "epoch": 2.69, "learning_rate": 9.275336278409814e-06, "loss": 0.097, "step": 5076 }, { "epoch": 2.69, "learning_rate": 9.271916025317714e-06, "loss": 0.1479, "step": 5077 }, { "epoch": 2.69, "learning_rate": 9.268495857849601e-06, "loss": 0.1101, "step": 5078 }, { "epoch": 2.69, "learning_rate": 9.265075776407695e-06, "loss": 0.0784, "step": 5079 }, { "epoch": 2.69, "learning_rate": 9.2616557813942e-06, "loss": 0.1089, "step": 5080 }, { "epoch": 2.69, "learning_rate": 9.258235873211315e-06, "loss": 0.1202, "step": 5081 }, { "epoch": 2.69, "learning_rate": 9.254816052261227e-06, "loss": 0.0991, "step": 5082 }, { "epoch": 2.69, "learning_rate": 9.251396318946113e-06, "loss": 0.1927, "step": 5083 }, { "epoch": 2.69, "learning_rate": 9.24797667366814e-06, "loss": 0.055, "step": 5084 }, { "epoch": 2.69, "learning_rate": 9.244557116829467e-06, "loss": 0.1622, "step": 5085 }, { "epoch": 2.69, "learning_rate": 9.241137648832237e-06, "loss": 0.1023, "step": 5086 }, { "epoch": 2.69, "learning_rate": 9.237718270078592e-06, "loss": 0.0897, "step": 5087 }, { "epoch": 2.69, "learning_rate": 9.234298980970644e-06, "loss": 0.1291, "step": 5088 }, { "epoch": 2.69, "learning_rate": 9.23087978191052e-06, "loss": 0.0915, "step": 5089 }, { "epoch": 2.69, "learning_rate": 9.227460673300317e-06, "loss": 0.1648, "step": 5090 }, { "epoch": 2.69, "learning_rate": 9.224041655542131e-06, "loss": 0.1347, "step": 5091 }, { "epoch": 2.7, "learning_rate": 9.220622729038045e-06, "loss": 0.1613, "step": 5092 }, { "epoch": 2.7, "learning_rate": 9.217203894190128e-06, "loss": 0.1896, "step": 5093 }, { "epoch": 2.7, "learning_rate": 9.213785151400446e-06, "loss": 0.158, "step": 5094 }, { "epoch": 2.7, "learning_rate": 9.210366501071047e-06, "loss": 0.0895, "step": 5095 }, { "epoch": 2.7, "learning_rate": 9.20694794360397e-06, "loss": 0.1292, "step": 5096 }, { "epoch": 2.7, "learning_rate": 9.203529479401241e-06, "loss": 0.0584, "step": 5097 }, { "epoch": 2.7, "learning_rate": 9.200111108864883e-06, "loss": 0.1128, "step": 5098 }, { "epoch": 2.7, "learning_rate": 9.196692832396897e-06, "loss": 0.0741, "step": 5099 }, { "epoch": 2.7, "learning_rate": 9.19327465039928e-06, "loss": 0.0834, "step": 5100 }, { "epoch": 2.7, "learning_rate": 9.18985656327402e-06, "loss": 0.1583, "step": 5101 }, { "epoch": 2.7, "learning_rate": 9.186438571423086e-06, "loss": 0.1498, "step": 5102 }, { "epoch": 2.7, "learning_rate": 9.183020675248448e-06, "loss": 0.1199, "step": 5103 }, { "epoch": 2.7, "learning_rate": 9.179602875152042e-06, "loss": 0.1369, "step": 5104 }, { "epoch": 2.7, "learning_rate": 9.176185171535821e-06, "loss": 0.0854, "step": 5105 }, { "epoch": 2.7, "learning_rate": 9.172767564801706e-06, "loss": 0.1647, "step": 5106 }, { "epoch": 2.7, "learning_rate": 9.169350055351617e-06, "loss": 0.1498, "step": 5107 }, { "epoch": 2.7, "learning_rate": 9.165932643587455e-06, "loss": 0.1109, "step": 5108 }, { "epoch": 2.7, "learning_rate": 9.162515329911119e-06, "loss": 0.151, "step": 5109 }, { "epoch": 2.7, "learning_rate": 9.15909811472449e-06, "loss": 0.1137, "step": 5110 }, { "epoch": 2.71, "learning_rate": 9.155680998429437e-06, "loss": 0.1227, "step": 5111 }, { "epoch": 2.71, "learning_rate": 9.152263981427819e-06, "loss": 0.2145, "step": 5112 }, { "epoch": 2.71, "learning_rate": 9.148847064121485e-06, "loss": 0.1269, "step": 5113 }, { "epoch": 2.71, "learning_rate": 9.145430246912268e-06, "loss": 0.122, "step": 5114 }, { "epoch": 2.71, "learning_rate": 9.142013530201992e-06, "loss": 0.1316, "step": 5115 }, { "epoch": 2.71, "learning_rate": 9.138596914392473e-06, "loss": 0.1317, "step": 5116 }, { "epoch": 2.71, "learning_rate": 9.135180399885509e-06, "loss": 0.0673, "step": 5117 }, { "epoch": 2.71, "learning_rate": 9.131763987082886e-06, "loss": 0.1657, "step": 5118 }, { "epoch": 2.71, "learning_rate": 9.128347676386387e-06, "loss": 0.1722, "step": 5119 }, { "epoch": 2.71, "learning_rate": 9.12493146819777e-06, "loss": 0.1561, "step": 5120 }, { "epoch": 2.71, "learning_rate": 9.121515362918786e-06, "loss": 0.0925, "step": 5121 }, { "epoch": 2.71, "learning_rate": 9.118099360951178e-06, "loss": 0.1485, "step": 5122 }, { "epoch": 2.71, "learning_rate": 9.114683462696677e-06, "loss": 0.0808, "step": 5123 }, { "epoch": 2.71, "learning_rate": 9.111267668556993e-06, "loss": 0.2135, "step": 5124 }, { "epoch": 2.71, "learning_rate": 9.107851978933833e-06, "loss": 0.1344, "step": 5125 }, { "epoch": 2.71, "learning_rate": 9.104436394228885e-06, "loss": 0.1453, "step": 5126 }, { "epoch": 2.71, "learning_rate": 9.10102091484384e-06, "loss": 0.0803, "step": 5127 }, { "epoch": 2.71, "learning_rate": 9.097605541180347e-06, "loss": 0.1589, "step": 5128 }, { "epoch": 2.71, "learning_rate": 9.09419027364007e-06, "loss": 0.2448, "step": 5129 }, { "epoch": 2.72, "learning_rate": 9.090775112624646e-06, "loss": 0.2049, "step": 5130 }, { "epoch": 2.72, "learning_rate": 9.087360058535709e-06, "loss": 0.1064, "step": 5131 }, { "epoch": 2.72, "learning_rate": 9.083945111774872e-06, "loss": 0.1545, "step": 5132 }, { "epoch": 2.72, "learning_rate": 9.080530272743741e-06, "loss": 0.2503, "step": 5133 }, { "epoch": 2.72, "learning_rate": 9.077115541843908e-06, "loss": 0.1177, "step": 5134 }, { "epoch": 2.72, "learning_rate": 9.07370091947695e-06, "loss": 0.1077, "step": 5135 }, { "epoch": 2.72, "learning_rate": 9.070286406044432e-06, "loss": 0.1639, "step": 5136 }, { "epoch": 2.72, "learning_rate": 9.066872001947906e-06, "loss": 0.1441, "step": 5137 }, { "epoch": 2.72, "learning_rate": 9.063457707588915e-06, "loss": 0.1696, "step": 5138 }, { "epoch": 2.72, "learning_rate": 9.060043523368984e-06, "loss": 0.1713, "step": 5139 }, { "epoch": 2.72, "learning_rate": 9.056629449689629e-06, "loss": 0.1419, "step": 5140 }, { "epoch": 2.72, "learning_rate": 9.05321548695235e-06, "loss": 0.0651, "step": 5141 }, { "epoch": 2.72, "learning_rate": 9.049801635558633e-06, "loss": 0.2005, "step": 5142 }, { "epoch": 2.72, "learning_rate": 9.046387895909962e-06, "loss": 0.0744, "step": 5143 }, { "epoch": 2.72, "learning_rate": 9.042974268407789e-06, "loss": 0.1396, "step": 5144 }, { "epoch": 2.72, "learning_rate": 9.039560753453565e-06, "loss": 0.1297, "step": 5145 }, { "epoch": 2.72, "learning_rate": 9.036147351448723e-06, "loss": 0.1616, "step": 5146 }, { "epoch": 2.72, "learning_rate": 9.032734062794693e-06, "loss": 0.1118, "step": 5147 }, { "epoch": 2.72, "learning_rate": 9.029320887892881e-06, "loss": 0.0784, "step": 5148 }, { "epoch": 2.73, "learning_rate": 9.02590782714468e-06, "loss": 0.1094, "step": 5149 }, { "epoch": 2.73, "learning_rate": 9.022494880951473e-06, "loss": 0.0904, "step": 5150 }, { "epoch": 2.73, "learning_rate": 9.019082049714631e-06, "loss": 0.1523, "step": 5151 }, { "epoch": 2.73, "learning_rate": 9.015669333835506e-06, "loss": 0.1784, "step": 5152 }, { "epoch": 2.73, "learning_rate": 9.012256733715441e-06, "loss": 0.1498, "step": 5153 }, { "epoch": 2.73, "learning_rate": 9.008844249755763e-06, "loss": 0.2083, "step": 5154 }, { "epoch": 2.73, "learning_rate": 9.005431882357785e-06, "loss": 0.1082, "step": 5155 }, { "epoch": 2.73, "learning_rate": 9.002019631922813e-06, "loss": 0.0942, "step": 5156 }, { "epoch": 2.73, "learning_rate": 8.998607498852126e-06, "loss": 0.1231, "step": 5157 }, { "epoch": 2.73, "learning_rate": 8.995195483547004e-06, "loss": 0.1641, "step": 5158 }, { "epoch": 2.73, "learning_rate": 8.991783586408707e-06, "loss": 0.2234, "step": 5159 }, { "epoch": 2.73, "learning_rate": 8.988371807838472e-06, "loss": 0.136, "step": 5160 }, { "epoch": 2.73, "learning_rate": 8.984960148237534e-06, "loss": 0.1119, "step": 5161 }, { "epoch": 2.73, "learning_rate": 8.981548608007112e-06, "loss": 0.176, "step": 5162 }, { "epoch": 2.73, "learning_rate": 8.97813718754841e-06, "loss": 0.1289, "step": 5163 }, { "epoch": 2.73, "learning_rate": 8.974725887262613e-06, "loss": 0.1094, "step": 5164 }, { "epoch": 2.73, "learning_rate": 8.971314707550902e-06, "loss": 0.141, "step": 5165 }, { "epoch": 2.73, "learning_rate": 8.967903648814433e-06, "loss": 0.1173, "step": 5166 }, { "epoch": 2.73, "learning_rate": 8.964492711454358e-06, "loss": 0.0843, "step": 5167 }, { "epoch": 2.74, "learning_rate": 8.9610818958718e-06, "loss": 0.1048, "step": 5168 }, { "epoch": 2.74, "learning_rate": 8.957671202467888e-06, "loss": 0.109, "step": 5169 }, { "epoch": 2.74, "learning_rate": 8.954260631643716e-06, "loss": 0.1951, "step": 5170 }, { "epoch": 2.74, "learning_rate": 8.950850183800379e-06, "loss": 0.086, "step": 5171 }, { "epoch": 2.74, "learning_rate": 8.94743985933895e-06, "loss": 0.2152, "step": 5172 }, { "epoch": 2.74, "learning_rate": 8.94402965866049e-06, "loss": 0.1338, "step": 5173 }, { "epoch": 2.74, "learning_rate": 8.940619582166045e-06, "loss": 0.1317, "step": 5174 }, { "epoch": 2.74, "learning_rate": 8.93720963025665e-06, "loss": 0.1618, "step": 5175 }, { "epoch": 2.74, "learning_rate": 8.933799803333311e-06, "loss": 0.0967, "step": 5176 }, { "epoch": 2.74, "learning_rate": 8.930390101797037e-06, "loss": 0.1618, "step": 5177 }, { "epoch": 2.74, "learning_rate": 8.926980526048815e-06, "loss": 0.098, "step": 5178 }, { "epoch": 2.74, "learning_rate": 8.923571076489616e-06, "loss": 0.1125, "step": 5179 }, { "epoch": 2.74, "learning_rate": 8.920161753520398e-06, "loss": 0.0918, "step": 5180 }, { "epoch": 2.74, "learning_rate": 8.916752557542103e-06, "loss": 0.1417, "step": 5181 }, { "epoch": 2.74, "learning_rate": 8.913343488955658e-06, "loss": 0.1246, "step": 5182 }, { "epoch": 2.74, "learning_rate": 8.909934548161977e-06, "loss": 0.0977, "step": 5183 }, { "epoch": 2.74, "learning_rate": 8.90652573556196e-06, "loss": 0.1336, "step": 5184 }, { "epoch": 2.74, "learning_rate": 8.903117051556484e-06, "loss": 0.145, "step": 5185 }, { "epoch": 2.75, "learning_rate": 8.899708496546417e-06, "loss": 0.15, "step": 5186 }, { "epoch": 2.75, "learning_rate": 8.896300070932616e-06, "loss": 0.1913, "step": 5187 }, { "epoch": 2.75, "learning_rate": 8.892891775115911e-06, "loss": 0.0972, "step": 5188 }, { "epoch": 2.75, "learning_rate": 8.88948360949713e-06, "loss": 0.1505, "step": 5189 }, { "epoch": 2.75, "learning_rate": 8.886075574477077e-06, "loss": 0.1337, "step": 5190 }, { "epoch": 2.75, "learning_rate": 8.882667670456544e-06, "loss": 0.1071, "step": 5191 }, { "epoch": 2.75, "learning_rate": 8.879259897836307e-06, "loss": 0.1415, "step": 5192 }, { "epoch": 2.75, "learning_rate": 8.875852257017124e-06, "loss": 0.0984, "step": 5193 }, { "epoch": 2.75, "learning_rate": 8.87244474839974e-06, "loss": 0.1408, "step": 5194 }, { "epoch": 2.75, "learning_rate": 8.869037372384886e-06, "loss": 0.1269, "step": 5195 }, { "epoch": 2.75, "learning_rate": 8.865630129373275e-06, "loss": 0.1285, "step": 5196 }, { "epoch": 2.75, "learning_rate": 8.862223019765603e-06, "loss": 0.128, "step": 5197 }, { "epoch": 2.75, "learning_rate": 8.858816043962554e-06, "loss": 0.1119, "step": 5198 }, { "epoch": 2.75, "learning_rate": 8.855409202364792e-06, "loss": 0.1017, "step": 5199 }, { "epoch": 2.75, "learning_rate": 8.852002495372977e-06, "loss": 0.1501, "step": 5200 }, { "epoch": 2.75, "learning_rate": 8.848595923387731e-06, "loss": 0.1252, "step": 5201 }, { "epoch": 2.75, "learning_rate": 8.84518948680968e-06, "loss": 0.1109, "step": 5202 }, { "epoch": 2.75, "learning_rate": 8.841783186039424e-06, "loss": 0.0916, "step": 5203 }, { "epoch": 2.75, "learning_rate": 8.838377021477552e-06, "loss": 0.1355, "step": 5204 }, { "epoch": 2.76, "learning_rate": 8.834970993524638e-06, "loss": 0.1106, "step": 5205 }, { "epoch": 2.76, "learning_rate": 8.831565102581232e-06, "loss": 0.1948, "step": 5206 }, { "epoch": 2.76, "learning_rate": 8.828159349047874e-06, "loss": 0.1338, "step": 5207 }, { "epoch": 2.76, "learning_rate": 8.824753733325092e-06, "loss": 0.1202, "step": 5208 }, { "epoch": 2.76, "learning_rate": 8.821348255813385e-06, "loss": 0.1811, "step": 5209 }, { "epoch": 2.76, "learning_rate": 8.817942916913247e-06, "loss": 0.1415, "step": 5210 }, { "epoch": 2.76, "learning_rate": 8.814537717025151e-06, "loss": 0.1005, "step": 5211 }, { "epoch": 2.76, "learning_rate": 8.811132656549556e-06, "loss": 0.1324, "step": 5212 }, { "epoch": 2.76, "learning_rate": 8.807727735886901e-06, "loss": 0.1026, "step": 5213 }, { "epoch": 2.76, "learning_rate": 8.804322955437615e-06, "loss": 0.1188, "step": 5214 }, { "epoch": 2.76, "learning_rate": 8.8009183156021e-06, "loss": 0.129, "step": 5215 }, { "epoch": 2.76, "learning_rate": 8.797513816780756e-06, "loss": 0.0665, "step": 5216 }, { "epoch": 2.76, "learning_rate": 8.79410945937395e-06, "loss": 0.1594, "step": 5217 }, { "epoch": 2.76, "learning_rate": 8.790705243782046e-06, "loss": 0.1047, "step": 5218 }, { "epoch": 2.76, "learning_rate": 8.78730117040538e-06, "loss": 0.0924, "step": 5219 }, { "epoch": 2.76, "learning_rate": 8.783897239644283e-06, "loss": 0.1201, "step": 5220 }, { "epoch": 2.76, "learning_rate": 8.780493451899063e-06, "loss": 0.1025, "step": 5221 }, { "epoch": 2.76, "learning_rate": 8.77708980757001e-06, "loss": 0.1463, "step": 5222 }, { "epoch": 2.76, "learning_rate": 8.773686307057397e-06, "loss": 0.1006, "step": 5223 }, { "epoch": 2.77, "learning_rate": 8.770282950761487e-06, "loss": 0.0553, "step": 5224 }, { "epoch": 2.77, "learning_rate": 8.766879739082515e-06, "loss": 0.1345, "step": 5225 }, { "epoch": 2.77, "learning_rate": 8.763476672420708e-06, "loss": 0.158, "step": 5226 }, { "epoch": 2.77, "learning_rate": 8.76007375117627e-06, "loss": 0.1518, "step": 5227 }, { "epoch": 2.77, "learning_rate": 8.756670975749395e-06, "loss": 0.1058, "step": 5228 }, { "epoch": 2.77, "learning_rate": 8.753268346540253e-06, "loss": 0.1052, "step": 5229 }, { "epoch": 2.77, "learning_rate": 8.749865863948996e-06, "loss": 0.1179, "step": 5230 }, { "epoch": 2.77, "learning_rate": 8.746463528375768e-06, "loss": 0.1158, "step": 5231 }, { "epoch": 2.77, "learning_rate": 8.743061340220693e-06, "loss": 0.1647, "step": 5232 }, { "epoch": 2.77, "learning_rate": 8.739659299883863e-06, "loss": 0.1546, "step": 5233 }, { "epoch": 2.77, "learning_rate": 8.736257407765368e-06, "loss": 0.0849, "step": 5234 }, { "epoch": 2.77, "learning_rate": 8.732855664265281e-06, "loss": 0.1288, "step": 5235 }, { "epoch": 2.77, "learning_rate": 8.72945406978365e-06, "loss": 0.1075, "step": 5236 }, { "epoch": 2.77, "learning_rate": 8.726052624720512e-06, "loss": 0.1967, "step": 5237 }, { "epoch": 2.77, "learning_rate": 8.722651329475877e-06, "loss": 0.1075, "step": 5238 }, { "epoch": 2.77, "learning_rate": 8.719250184449748e-06, "loss": 0.1124, "step": 5239 }, { "epoch": 2.77, "learning_rate": 8.715849190042109e-06, "loss": 0.1246, "step": 5240 }, { "epoch": 2.77, "learning_rate": 8.712448346652914e-06, "loss": 0.1768, "step": 5241 }, { "epoch": 2.77, "learning_rate": 8.709047654682115e-06, "loss": 0.1979, "step": 5242 }, { "epoch": 2.78, "learning_rate": 8.705647114529636e-06, "loss": 0.0965, "step": 5243 }, { "epoch": 2.78, "learning_rate": 8.70224672659539e-06, "loss": 0.1259, "step": 5244 }, { "epoch": 2.78, "learning_rate": 8.698846491279265e-06, "loss": 0.1573, "step": 5245 }, { "epoch": 2.78, "learning_rate": 8.695446408981135e-06, "loss": 0.1456, "step": 5246 }, { "epoch": 2.78, "learning_rate": 8.69204648010086e-06, "loss": 0.119, "step": 5247 }, { "epoch": 2.78, "learning_rate": 8.68864670503828e-06, "loss": 0.1393, "step": 5248 }, { "epoch": 2.78, "learning_rate": 8.685247084193204e-06, "loss": 0.1196, "step": 5249 }, { "epoch": 2.78, "learning_rate": 8.681847617965439e-06, "loss": 0.1346, "step": 5250 }, { "epoch": 2.78, "learning_rate": 8.67844830675477e-06, "loss": 0.1513, "step": 5251 }, { "epoch": 2.78, "learning_rate": 8.675049150960961e-06, "loss": 0.1771, "step": 5252 }, { "epoch": 2.78, "learning_rate": 8.67165015098376e-06, "loss": 0.1597, "step": 5253 }, { "epoch": 2.78, "learning_rate": 8.668251307222894e-06, "loss": 0.0783, "step": 5254 }, { "epoch": 2.78, "learning_rate": 8.664852620078073e-06, "loss": 0.1517, "step": 5255 }, { "epoch": 2.78, "learning_rate": 8.661454089948993e-06, "loss": 0.1103, "step": 5256 }, { "epoch": 2.78, "learning_rate": 8.658055717235319e-06, "loss": 0.0937, "step": 5257 }, { "epoch": 2.78, "learning_rate": 8.654657502336712e-06, "loss": 0.2239, "step": 5258 }, { "epoch": 2.78, "learning_rate": 8.651259445652805e-06, "loss": 0.14, "step": 5259 }, { "epoch": 2.78, "learning_rate": 8.647861547583218e-06, "loss": 0.114, "step": 5260 }, { "epoch": 2.78, "learning_rate": 8.644463808527547e-06, "loss": 0.1472, "step": 5261 }, { "epoch": 2.79, "learning_rate": 8.641066228885375e-06, "loss": 0.177, "step": 5262 }, { "epoch": 2.79, "learning_rate": 8.637668809056265e-06, "loss": 0.114, "step": 5263 }, { "epoch": 2.79, "learning_rate": 8.63427154943976e-06, "loss": 0.1195, "step": 5264 }, { "epoch": 2.79, "learning_rate": 8.630874450435375e-06, "loss": 0.1249, "step": 5265 }, { "epoch": 2.79, "learning_rate": 8.627477512442623e-06, "loss": 0.1286, "step": 5266 }, { "epoch": 2.79, "learning_rate": 8.624080735860989e-06, "loss": 0.1662, "step": 5267 }, { "epoch": 2.79, "learning_rate": 8.62068412108994e-06, "loss": 0.1325, "step": 5268 }, { "epoch": 2.79, "learning_rate": 8.617287668528921e-06, "loss": 0.1688, "step": 5269 }, { "epoch": 2.79, "learning_rate": 8.613891378577365e-06, "loss": 0.1605, "step": 5270 }, { "epoch": 2.79, "learning_rate": 8.610495251634679e-06, "loss": 0.1174, "step": 5271 }, { "epoch": 2.79, "learning_rate": 8.607099288100257e-06, "loss": 0.1654, "step": 5272 }, { "epoch": 2.79, "learning_rate": 8.603703488373466e-06, "loss": 0.1137, "step": 5273 }, { "epoch": 2.79, "learning_rate": 8.600307852853662e-06, "loss": 0.1209, "step": 5274 }, { "epoch": 2.79, "learning_rate": 8.596912381940175e-06, "loss": 0.1059, "step": 5275 }, { "epoch": 2.79, "learning_rate": 8.593517076032317e-06, "loss": 0.1029, "step": 5276 }, { "epoch": 2.79, "learning_rate": 8.590121935529388e-06, "loss": 0.113, "step": 5277 }, { "epoch": 2.79, "learning_rate": 8.586726960830658e-06, "loss": 0.1306, "step": 5278 }, { "epoch": 2.79, "learning_rate": 8.583332152335385e-06, "loss": 0.126, "step": 5279 }, { "epoch": 2.79, "learning_rate": 8.579937510442807e-06, "loss": 0.1326, "step": 5280 }, { "epoch": 2.8, "learning_rate": 8.576543035552127e-06, "loss": 0.0863, "step": 5281 }, { "epoch": 2.8, "learning_rate": 8.573148728062556e-06, "loss": 0.1192, "step": 5282 }, { "epoch": 2.8, "learning_rate": 8.569754588373264e-06, "loss": 0.1537, "step": 5283 }, { "epoch": 2.8, "learning_rate": 8.56636061688341e-06, "loss": 0.1476, "step": 5284 }, { "epoch": 2.8, "learning_rate": 8.562966813992128e-06, "loss": 0.0961, "step": 5285 }, { "epoch": 2.8, "learning_rate": 8.559573180098539e-06, "loss": 0.1899, "step": 5286 }, { "epoch": 2.8, "learning_rate": 8.556179715601739e-06, "loss": 0.1756, "step": 5287 }, { "epoch": 2.8, "learning_rate": 8.552786420900808e-06, "loss": 0.1107, "step": 5288 }, { "epoch": 2.8, "learning_rate": 8.5493932963948e-06, "loss": 0.1394, "step": 5289 }, { "epoch": 2.8, "learning_rate": 8.54600034248275e-06, "loss": 0.1601, "step": 5290 }, { "epoch": 2.8, "learning_rate": 8.542607559563683e-06, "loss": 0.059, "step": 5291 }, { "epoch": 2.8, "learning_rate": 8.539214948036588e-06, "loss": 0.1113, "step": 5292 }, { "epoch": 2.8, "learning_rate": 8.53582250830045e-06, "loss": 0.1196, "step": 5293 }, { "epoch": 2.8, "learning_rate": 8.532430240754223e-06, "loss": 0.101, "step": 5294 }, { "epoch": 2.8, "learning_rate": 8.529038145796843e-06, "loss": 0.167, "step": 5295 }, { "epoch": 2.8, "learning_rate": 8.52564622382723e-06, "loss": 0.0854, "step": 5296 }, { "epoch": 2.8, "learning_rate": 8.522254475244277e-06, "loss": 0.2225, "step": 5297 }, { "epoch": 2.8, "learning_rate": 8.518862900446855e-06, "loss": 0.2275, "step": 5298 }, { "epoch": 2.8, "learning_rate": 8.515471499833828e-06, "loss": 0.1901, "step": 5299 }, { "epoch": 2.81, "learning_rate": 8.512080273804027e-06, "loss": 0.088, "step": 5300 }, { "epoch": 2.81, "learning_rate": 8.508689222756265e-06, "loss": 0.0932, "step": 5301 }, { "epoch": 2.81, "learning_rate": 8.505298347089337e-06, "loss": 0.1241, "step": 5302 }, { "epoch": 2.81, "learning_rate": 8.501907647202015e-06, "loss": 0.081, "step": 5303 }, { "epoch": 2.81, "learning_rate": 8.498517123493055e-06, "loss": 0.1532, "step": 5304 }, { "epoch": 2.81, "learning_rate": 8.49512677636119e-06, "loss": 0.1381, "step": 5305 }, { "epoch": 2.81, "learning_rate": 8.491736606205123e-06, "loss": 0.0858, "step": 5306 }, { "epoch": 2.81, "learning_rate": 8.488346613423546e-06, "loss": 0.085, "step": 5307 }, { "epoch": 2.81, "learning_rate": 8.484956798415134e-06, "loss": 0.1493, "step": 5308 }, { "epoch": 2.81, "learning_rate": 8.48156716157853e-06, "loss": 0.1189, "step": 5309 }, { "epoch": 2.81, "learning_rate": 8.478177703312364e-06, "loss": 0.092, "step": 5310 }, { "epoch": 2.81, "learning_rate": 8.474788424015242e-06, "loss": 0.136, "step": 5311 }, { "epoch": 2.81, "learning_rate": 8.47139932408575e-06, "loss": 0.1482, "step": 5312 }, { "epoch": 2.81, "learning_rate": 8.468010403922452e-06, "loss": 0.0743, "step": 5313 }, { "epoch": 2.81, "learning_rate": 8.464621663923888e-06, "loss": 0.1893, "step": 5314 }, { "epoch": 2.81, "learning_rate": 8.461233104488585e-06, "loss": 0.1822, "step": 5315 }, { "epoch": 2.81, "learning_rate": 8.457844726015039e-06, "loss": 0.1537, "step": 5316 }, { "epoch": 2.81, "learning_rate": 8.454456528901733e-06, "loss": 0.131, "step": 5317 }, { "epoch": 2.81, "learning_rate": 8.451068513547121e-06, "loss": 0.2161, "step": 5318 }, { "epoch": 2.82, "learning_rate": 8.447680680349642e-06, "loss": 0.1378, "step": 5319 }, { "epoch": 2.82, "learning_rate": 8.444293029707714e-06, "loss": 0.1044, "step": 5320 }, { "epoch": 2.82, "learning_rate": 8.440905562019733e-06, "loss": 0.1357, "step": 5321 }, { "epoch": 2.82, "learning_rate": 8.43751827768406e-06, "loss": 0.0979, "step": 5322 }, { "epoch": 2.82, "learning_rate": 8.434131177099052e-06, "loss": 0.0635, "step": 5323 }, { "epoch": 2.82, "learning_rate": 8.430744260663042e-06, "loss": 0.1221, "step": 5324 }, { "epoch": 2.82, "learning_rate": 8.427357528774333e-06, "loss": 0.1004, "step": 5325 }, { "epoch": 2.82, "learning_rate": 8.423970981831211e-06, "loss": 0.1048, "step": 5326 }, { "epoch": 2.82, "learning_rate": 8.420584620231942e-06, "loss": 0.1114, "step": 5327 }, { "epoch": 2.82, "learning_rate": 8.417198444374768e-06, "loss": 0.2088, "step": 5328 }, { "epoch": 2.82, "learning_rate": 8.413812454657908e-06, "loss": 0.0957, "step": 5329 }, { "epoch": 2.82, "learning_rate": 8.41042665147956e-06, "loss": 0.1276, "step": 5330 }, { "epoch": 2.82, "learning_rate": 8.407041035237902e-06, "loss": 0.1888, "step": 5331 }, { "epoch": 2.82, "learning_rate": 8.403655606331087e-06, "loss": 0.1503, "step": 5332 }, { "epoch": 2.82, "learning_rate": 8.40027036515725e-06, "loss": 0.1846, "step": 5333 }, { "epoch": 2.82, "learning_rate": 8.396885312114497e-06, "loss": 0.104, "step": 5334 }, { "epoch": 2.82, "learning_rate": 8.393500447600922e-06, "loss": 0.1311, "step": 5335 }, { "epoch": 2.82, "learning_rate": 8.390115772014587e-06, "loss": 0.1805, "step": 5336 }, { "epoch": 2.82, "learning_rate": 8.386731285753545e-06, "loss": 0.1339, "step": 5337 }, { "epoch": 2.83, "learning_rate": 8.383346989215802e-06, "loss": 0.1843, "step": 5338 }, { "epoch": 2.83, "learning_rate": 8.379962882799366e-06, "loss": 0.1902, "step": 5339 }, { "epoch": 2.83, "learning_rate": 8.376578966902216e-06, "loss": 0.1399, "step": 5340 }, { "epoch": 2.83, "learning_rate": 8.373195241922303e-06, "loss": 0.1619, "step": 5341 }, { "epoch": 2.83, "learning_rate": 8.369811708257562e-06, "loss": 0.1313, "step": 5342 }, { "epoch": 2.83, "learning_rate": 8.366428366305898e-06, "loss": 0.1638, "step": 5343 }, { "epoch": 2.83, "learning_rate": 8.363045216465205e-06, "loss": 0.1093, "step": 5344 }, { "epoch": 2.83, "learning_rate": 8.359662259133344e-06, "loss": 0.1468, "step": 5345 }, { "epoch": 2.83, "learning_rate": 8.356279494708152e-06, "loss": 0.1426, "step": 5346 }, { "epoch": 2.83, "learning_rate": 8.352896923587456e-06, "loss": 0.0968, "step": 5347 }, { "epoch": 2.83, "learning_rate": 8.349514546169047e-06, "loss": 0.1549, "step": 5348 }, { "epoch": 2.83, "learning_rate": 8.346132362850702e-06, "loss": 0.1179, "step": 5349 }, { "epoch": 2.83, "learning_rate": 8.342750374030169e-06, "loss": 0.1096, "step": 5350 }, { "epoch": 2.83, "learning_rate": 8.339368580105179e-06, "loss": 0.1392, "step": 5351 }, { "epoch": 2.83, "learning_rate": 8.335986981473435e-06, "loss": 0.1102, "step": 5352 }, { "epoch": 2.83, "learning_rate": 8.332605578532625e-06, "loss": 0.1135, "step": 5353 }, { "epoch": 2.83, "learning_rate": 8.329224371680397e-06, "loss": 0.1128, "step": 5354 }, { "epoch": 2.83, "learning_rate": 8.325843361314394e-06, "loss": 0.147, "step": 5355 }, { "epoch": 2.83, "learning_rate": 8.322462547832228e-06, "loss": 0.2141, "step": 5356 }, { "epoch": 2.84, "learning_rate": 8.31908193163149e-06, "loss": 0.1316, "step": 5357 }, { "epoch": 2.84, "learning_rate": 8.315701513109742e-06, "loss": 0.1897, "step": 5358 }, { "epoch": 2.84, "learning_rate": 8.312321292664534e-06, "loss": 0.0987, "step": 5359 }, { "epoch": 2.84, "learning_rate": 8.30894127069338e-06, "loss": 0.123, "step": 5360 }, { "epoch": 2.84, "learning_rate": 8.305561447593782e-06, "loss": 0.1361, "step": 5361 }, { "epoch": 2.84, "learning_rate": 8.302181823763209e-06, "loss": 0.0521, "step": 5362 }, { "epoch": 2.84, "learning_rate": 8.298802399599112e-06, "loss": 0.0743, "step": 5363 }, { "epoch": 2.84, "learning_rate": 8.295423175498916e-06, "loss": 0.1146, "step": 5364 }, { "epoch": 2.84, "learning_rate": 8.292044151860026e-06, "loss": 0.162, "step": 5365 }, { "epoch": 2.84, "learning_rate": 8.288665329079822e-06, "loss": 0.1542, "step": 5366 }, { "epoch": 2.84, "learning_rate": 8.285286707555657e-06, "loss": 0.0868, "step": 5367 }, { "epoch": 2.84, "learning_rate": 8.281908287684865e-06, "loss": 0.11, "step": 5368 }, { "epoch": 2.84, "learning_rate": 8.278530069864756e-06, "loss": 0.2083, "step": 5369 }, { "epoch": 2.84, "learning_rate": 8.27515205449261e-06, "loss": 0.1282, "step": 5370 }, { "epoch": 2.84, "learning_rate": 8.27177424196569e-06, "loss": 0.1837, "step": 5371 }, { "epoch": 2.84, "learning_rate": 8.268396632681232e-06, "loss": 0.1145, "step": 5372 }, { "epoch": 2.84, "learning_rate": 8.265019227036451e-06, "loss": 0.1256, "step": 5373 }, { "epoch": 2.84, "learning_rate": 8.261642025428533e-06, "loss": 0.0904, "step": 5374 }, { "epoch": 2.85, "learning_rate": 8.258265028254647e-06, "loss": 0.1231, "step": 5375 }, { "epoch": 2.85, "learning_rate": 8.254888235911927e-06, "loss": 0.172, "step": 5376 }, { "epoch": 2.85, "learning_rate": 8.2515116487975e-06, "loss": 0.1146, "step": 5377 }, { "epoch": 2.85, "learning_rate": 8.248135267308452e-06, "loss": 0.1423, "step": 5378 }, { "epoch": 2.85, "learning_rate": 8.24475909184185e-06, "loss": 0.0783, "step": 5379 }, { "epoch": 2.85, "learning_rate": 8.241383122794741e-06, "loss": 0.1047, "step": 5380 }, { "epoch": 2.85, "learning_rate": 8.238007360564145e-06, "loss": 0.1353, "step": 5381 }, { "epoch": 2.85, "learning_rate": 8.234631805547058e-06, "loss": 0.1198, "step": 5382 }, { "epoch": 2.85, "learning_rate": 8.231256458140453e-06, "loss": 0.0526, "step": 5383 }, { "epoch": 2.85, "learning_rate": 8.227881318741274e-06, "loss": 0.1402, "step": 5384 }, { "epoch": 2.85, "learning_rate": 8.224506387746447e-06, "loss": 0.086, "step": 5385 }, { "epoch": 2.85, "learning_rate": 8.221131665552863e-06, "loss": 0.1008, "step": 5386 }, { "epoch": 2.85, "learning_rate": 8.217757152557403e-06, "loss": 0.0815, "step": 5387 }, { "epoch": 2.85, "learning_rate": 8.214382849156912e-06, "loss": 0.0902, "step": 5388 }, { "epoch": 2.85, "learning_rate": 8.211008755748215e-06, "loss": 0.1653, "step": 5389 }, { "epoch": 2.85, "learning_rate": 8.207634872728109e-06, "loss": 0.1818, "step": 5390 }, { "epoch": 2.85, "learning_rate": 8.204261200493373e-06, "loss": 0.101, "step": 5391 }, { "epoch": 2.85, "learning_rate": 8.200887739440753e-06, "loss": 0.1511, "step": 5392 }, { "epoch": 2.85, "learning_rate": 8.197514489966982e-06, "loss": 0.0976, "step": 5393 }, { "epoch": 2.86, "learning_rate": 8.19414145246875e-06, "loss": 0.1172, "step": 5394 }, { "epoch": 2.86, "learning_rate": 8.190768627342737e-06, "loss": 0.2051, "step": 5395 }, { "epoch": 2.86, "learning_rate": 8.18739601498559e-06, "loss": 0.1417, "step": 5396 }, { "epoch": 2.86, "learning_rate": 8.18402361579394e-06, "loss": 0.1274, "step": 5397 }, { "epoch": 2.86, "learning_rate": 8.180651430164383e-06, "loss": 0.1109, "step": 5398 }, { "epoch": 2.86, "learning_rate": 8.177279458493496e-06, "loss": 0.0661, "step": 5399 }, { "epoch": 2.86, "learning_rate": 8.17390770117783e-06, "loss": 0.125, "step": 5400 }, { "epoch": 2.86, "learning_rate": 8.170536158613908e-06, "loss": 0.0825, "step": 5401 }, { "epoch": 2.86, "learning_rate": 8.16716483119823e-06, "loss": 0.1406, "step": 5402 }, { "epoch": 2.86, "learning_rate": 8.16379371932727e-06, "loss": 0.1509, "step": 5403 }, { "epoch": 2.86, "learning_rate": 8.160422823397473e-06, "loss": 0.1238, "step": 5404 }, { "epoch": 2.86, "learning_rate": 8.157052143805268e-06, "loss": 0.0601, "step": 5405 }, { "epoch": 2.86, "learning_rate": 8.153681680947051e-06, "loss": 0.1072, "step": 5406 }, { "epoch": 2.86, "learning_rate": 8.150311435219195e-06, "loss": 0.1065, "step": 5407 }, { "epoch": 2.86, "learning_rate": 8.146941407018046e-06, "loss": 0.1028, "step": 5408 }, { "epoch": 2.86, "learning_rate": 8.143571596739931e-06, "loss": 0.1964, "step": 5409 }, { "epoch": 2.86, "learning_rate": 8.140202004781136e-06, "loss": 0.1276, "step": 5410 }, { "epoch": 2.86, "learning_rate": 8.136832631537934e-06, "loss": 0.1461, "step": 5411 }, { "epoch": 2.86, "learning_rate": 8.133463477406573e-06, "loss": 0.1545, "step": 5412 }, { "epoch": 2.87, "learning_rate": 8.13009454278327e-06, "loss": 0.1462, "step": 5413 }, { "epoch": 2.87, "learning_rate": 8.126725828064218e-06, "loss": 0.072, "step": 5414 }, { "epoch": 2.87, "learning_rate": 8.123357333645583e-06, "loss": 0.1408, "step": 5415 }, { "epoch": 2.87, "learning_rate": 8.119989059923505e-06, "loss": 0.0995, "step": 5416 }, { "epoch": 2.87, "learning_rate": 8.116621007294106e-06, "loss": 0.1229, "step": 5417 }, { "epoch": 2.87, "learning_rate": 8.113253176153465e-06, "loss": 0.0784, "step": 5418 }, { "epoch": 2.87, "learning_rate": 8.10988556689765e-06, "loss": 0.1287, "step": 5419 }, { "epoch": 2.87, "learning_rate": 8.106518179922697e-06, "loss": 0.0958, "step": 5420 }, { "epoch": 2.87, "learning_rate": 8.103151015624619e-06, "loss": 0.1141, "step": 5421 }, { "epoch": 2.87, "learning_rate": 8.099784074399398e-06, "loss": 0.1612, "step": 5422 }, { "epoch": 2.87, "learning_rate": 8.09641735664299e-06, "loss": 0.1466, "step": 5423 }, { "epoch": 2.87, "learning_rate": 8.093050862751333e-06, "loss": 0.1133, "step": 5424 }, { "epoch": 2.87, "learning_rate": 8.089684593120335e-06, "loss": 0.09, "step": 5425 }, { "epoch": 2.87, "learning_rate": 8.086318548145866e-06, "loss": 0.0993, "step": 5426 }, { "epoch": 2.87, "learning_rate": 8.08295272822378e-06, "loss": 0.1259, "step": 5427 }, { "epoch": 2.87, "learning_rate": 8.079587133749912e-06, "loss": 0.0534, "step": 5428 }, { "epoch": 2.87, "learning_rate": 8.076221765120053e-06, "loss": 0.152, "step": 5429 }, { "epoch": 2.87, "learning_rate": 8.072856622729981e-06, "loss": 0.1701, "step": 5430 }, { "epoch": 2.87, "learning_rate": 8.069491706975445e-06, "loss": 0.0819, "step": 5431 }, { "epoch": 2.88, "learning_rate": 8.066127018252158e-06, "loss": 0.1175, "step": 5432 }, { "epoch": 2.88, "learning_rate": 8.062762556955821e-06, "loss": 0.1296, "step": 5433 }, { "epoch": 2.88, "learning_rate": 8.0593983234821e-06, "loss": 0.174, "step": 5434 }, { "epoch": 2.88, "learning_rate": 8.056034318226628e-06, "loss": 0.1662, "step": 5435 }, { "epoch": 2.88, "learning_rate": 8.052670541585023e-06, "loss": 0.1001, "step": 5436 }, { "epoch": 2.88, "learning_rate": 8.04930699395287e-06, "loss": 0.0612, "step": 5437 }, { "epoch": 2.88, "learning_rate": 8.045943675725727e-06, "loss": 0.1281, "step": 5438 }, { "epoch": 2.88, "learning_rate": 8.04258058729913e-06, "loss": 0.1786, "step": 5439 }, { "epoch": 2.88, "learning_rate": 8.039217729068584e-06, "loss": 0.1223, "step": 5440 }, { "epoch": 2.88, "learning_rate": 8.035855101429564e-06, "loss": 0.1114, "step": 5441 }, { "epoch": 2.88, "learning_rate": 8.032492704777524e-06, "loss": 0.0613, "step": 5442 }, { "epoch": 2.88, "learning_rate": 8.029130539507886e-06, "loss": 0.1056, "step": 5443 }, { "epoch": 2.88, "learning_rate": 8.025768606016047e-06, "loss": 0.1812, "step": 5444 }, { "epoch": 2.88, "learning_rate": 8.022406904697376e-06, "loss": 0.1204, "step": 5445 }, { "epoch": 2.88, "learning_rate": 8.019045435947216e-06, "loss": 0.1227, "step": 5446 }, { "epoch": 2.88, "learning_rate": 8.015684200160883e-06, "loss": 0.1089, "step": 5447 }, { "epoch": 2.88, "learning_rate": 8.01232319773366e-06, "loss": 0.1442, "step": 5448 }, { "epoch": 2.88, "learning_rate": 8.008962429060811e-06, "loss": 0.1746, "step": 5449 }, { "epoch": 2.88, "learning_rate": 8.005601894537572e-06, "loss": 0.1369, "step": 5450 }, { "epoch": 2.89, "learning_rate": 8.002241594559138e-06, "loss": 0.1629, "step": 5451 }, { "epoch": 2.89, "learning_rate": 7.998881529520694e-06, "loss": 0.138, "step": 5452 }, { "epoch": 2.89, "learning_rate": 7.995521699817383e-06, "loss": 0.1842, "step": 5453 }, { "epoch": 2.89, "learning_rate": 7.992162105844335e-06, "loss": 0.1262, "step": 5454 }, { "epoch": 2.89, "learning_rate": 7.98880274799664e-06, "loss": 0.1757, "step": 5455 }, { "epoch": 2.89, "learning_rate": 7.985443626669365e-06, "loss": 0.1508, "step": 5456 }, { "epoch": 2.89, "learning_rate": 7.982084742257547e-06, "loss": 0.1079, "step": 5457 }, { "epoch": 2.89, "learning_rate": 7.978726095156205e-06, "loss": 0.0976, "step": 5458 }, { "epoch": 2.89, "learning_rate": 7.97536768576031e-06, "loss": 0.1169, "step": 5459 }, { "epoch": 2.89, "learning_rate": 7.972009514464824e-06, "loss": 0.1412, "step": 5460 }, { "epoch": 2.89, "learning_rate": 7.968651581664672e-06, "loss": 0.1465, "step": 5461 }, { "epoch": 2.89, "learning_rate": 7.965293887754755e-06, "loss": 0.1595, "step": 5462 }, { "epoch": 2.89, "learning_rate": 7.961936433129943e-06, "loss": 0.1924, "step": 5463 }, { "epoch": 2.89, "learning_rate": 7.958579218185075e-06, "loss": 0.1625, "step": 5464 }, { "epoch": 2.89, "learning_rate": 7.95522224331497e-06, "loss": 0.0838, "step": 5465 }, { "epoch": 2.89, "learning_rate": 7.951865508914417e-06, "loss": 0.1557, "step": 5466 }, { "epoch": 2.89, "learning_rate": 7.948509015378166e-06, "loss": 0.1658, "step": 5467 }, { "epoch": 2.89, "learning_rate": 7.945152763100952e-06, "loss": 0.1466, "step": 5468 }, { "epoch": 2.89, "learning_rate": 7.941796752477472e-06, "loss": 0.1243, "step": 5469 }, { "epoch": 2.9, "learning_rate": 7.938440983902404e-06, "loss": 0.1576, "step": 5470 }, { "epoch": 2.9, "learning_rate": 7.935085457770391e-06, "loss": 0.1128, "step": 5471 }, { "epoch": 2.9, "learning_rate": 7.931730174476047e-06, "loss": 0.1759, "step": 5472 }, { "epoch": 2.9, "learning_rate": 7.928375134413961e-06, "loss": 0.1341, "step": 5473 }, { "epoch": 2.9, "learning_rate": 7.925020337978695e-06, "loss": 0.1414, "step": 5474 }, { "epoch": 2.9, "learning_rate": 7.92166578556477e-06, "loss": 0.1137, "step": 5475 }, { "epoch": 2.9, "learning_rate": 7.918311477566696e-06, "loss": 0.1152, "step": 5476 }, { "epoch": 2.9, "learning_rate": 7.914957414378942e-06, "loss": 0.1422, "step": 5477 }, { "epoch": 2.9, "learning_rate": 7.911603596395953e-06, "loss": 0.1275, "step": 5478 }, { "epoch": 2.9, "learning_rate": 7.908250024012142e-06, "loss": 0.154, "step": 5479 }, { "epoch": 2.9, "learning_rate": 7.904896697621896e-06, "loss": 0.0922, "step": 5480 }, { "epoch": 2.9, "learning_rate": 7.901543617619572e-06, "loss": 0.1648, "step": 5481 }, { "epoch": 2.9, "learning_rate": 7.898190784399507e-06, "loss": 0.176, "step": 5482 }, { "epoch": 2.9, "learning_rate": 7.894838198355985e-06, "loss": 0.1029, "step": 5483 }, { "epoch": 2.9, "learning_rate": 7.891485859883284e-06, "loss": 0.1008, "step": 5484 }, { "epoch": 2.9, "learning_rate": 7.888133769375644e-06, "loss": 0.0972, "step": 5485 }, { "epoch": 2.9, "learning_rate": 7.88478192722728e-06, "loss": 0.1453, "step": 5486 }, { "epoch": 2.9, "learning_rate": 7.88143033383237e-06, "loss": 0.112, "step": 5487 }, { "epoch": 2.9, "learning_rate": 7.87807898958507e-06, "loss": 0.2215, "step": 5488 }, { "epoch": 2.91, "learning_rate": 7.874727894879504e-06, "loss": 0.136, "step": 5489 }, { "epoch": 2.91, "learning_rate": 7.871377050109769e-06, "loss": 0.1296, "step": 5490 }, { "epoch": 2.91, "learning_rate": 7.868026455669925e-06, "loss": 0.1126, "step": 5491 }, { "epoch": 2.91, "learning_rate": 7.86467611195401e-06, "loss": 0.1589, "step": 5492 }, { "epoch": 2.91, "learning_rate": 7.861326019356033e-06, "loss": 0.1236, "step": 5493 }, { "epoch": 2.91, "learning_rate": 7.857976178269969e-06, "loss": 0.0991, "step": 5494 }, { "epoch": 2.91, "learning_rate": 7.854626589089763e-06, "loss": 0.1255, "step": 5495 }, { "epoch": 2.91, "learning_rate": 7.851277252209335e-06, "loss": 0.1632, "step": 5496 }, { "epoch": 2.91, "learning_rate": 7.847928168022575e-06, "loss": 0.118, "step": 5497 }, { "epoch": 2.91, "learning_rate": 7.844579336923344e-06, "loss": 0.0872, "step": 5498 }, { "epoch": 2.91, "learning_rate": 7.841230759305461e-06, "loss": 0.1676, "step": 5499 }, { "epoch": 2.91, "learning_rate": 7.837882435562728e-06, "loss": 0.1388, "step": 5500 }, { "epoch": 2.91, "learning_rate": 7.834534366088918e-06, "loss": 0.1554, "step": 5501 }, { "epoch": 2.91, "learning_rate": 7.831186551277769e-06, "loss": 0.1267, "step": 5502 }, { "epoch": 2.91, "learning_rate": 7.827838991522987e-06, "loss": 0.0808, "step": 5503 }, { "epoch": 2.91, "learning_rate": 7.824491687218254e-06, "loss": 0.1668, "step": 5504 }, { "epoch": 2.91, "learning_rate": 7.821144638757218e-06, "loss": 0.154, "step": 5505 }, { "epoch": 2.91, "learning_rate": 7.8177978465335e-06, "loss": 0.0644, "step": 5506 }, { "epoch": 2.91, "learning_rate": 7.814451310940683e-06, "loss": 0.1945, "step": 5507 }, { "epoch": 2.92, "learning_rate": 7.81110503237233e-06, "loss": 0.1504, "step": 5508 }, { "epoch": 2.92, "learning_rate": 7.80775901122197e-06, "loss": 0.1596, "step": 5509 }, { "epoch": 2.92, "learning_rate": 7.804413247883098e-06, "loss": 0.1073, "step": 5510 }, { "epoch": 2.92, "learning_rate": 7.801067742749182e-06, "loss": 0.1073, "step": 5511 }, { "epoch": 2.92, "learning_rate": 7.797722496213663e-06, "loss": 0.1169, "step": 5512 }, { "epoch": 2.92, "learning_rate": 7.794377508669948e-06, "loss": 0.104, "step": 5513 }, { "epoch": 2.92, "learning_rate": 7.791032780511413e-06, "loss": 0.1029, "step": 5514 }, { "epoch": 2.92, "learning_rate": 7.787688312131398e-06, "loss": 0.1225, "step": 5515 }, { "epoch": 2.92, "learning_rate": 7.784344103923224e-06, "loss": 0.1657, "step": 5516 }, { "epoch": 2.92, "learning_rate": 7.781000156280175e-06, "loss": 0.1488, "step": 5517 }, { "epoch": 2.92, "learning_rate": 7.777656469595508e-06, "loss": 0.1197, "step": 5518 }, { "epoch": 2.92, "learning_rate": 7.77431304426244e-06, "loss": 0.1738, "step": 5519 }, { "epoch": 2.92, "learning_rate": 7.77096988067417e-06, "loss": 0.1396, "step": 5520 }, { "epoch": 2.92, "learning_rate": 7.767626979223856e-06, "loss": 0.1289, "step": 5521 }, { "epoch": 2.92, "learning_rate": 7.764284340304635e-06, "loss": 0.1306, "step": 5522 }, { "epoch": 2.92, "learning_rate": 7.760941964309601e-06, "loss": 0.1323, "step": 5523 }, { "epoch": 2.92, "learning_rate": 7.757599851631825e-06, "loss": 0.0954, "step": 5524 }, { "epoch": 2.92, "learning_rate": 7.754258002664347e-06, "loss": 0.109, "step": 5525 }, { "epoch": 2.92, "learning_rate": 7.750916417800171e-06, "loss": 0.1481, "step": 5526 }, { "epoch": 2.93, "learning_rate": 7.747575097432279e-06, "loss": 0.1369, "step": 5527 }, { "epoch": 2.93, "learning_rate": 7.744234041953614e-06, "loss": 0.0956, "step": 5528 }, { "epoch": 2.93, "learning_rate": 7.740893251757089e-06, "loss": 0.1558, "step": 5529 }, { "epoch": 2.93, "learning_rate": 7.73755272723559e-06, "loss": 0.1708, "step": 5530 }, { "epoch": 2.93, "learning_rate": 7.734212468781963e-06, "loss": 0.1262, "step": 5531 }, { "epoch": 2.93, "learning_rate": 7.730872476789031e-06, "loss": 0.0912, "step": 5532 }, { "epoch": 2.93, "learning_rate": 7.727532751649586e-06, "loss": 0.1528, "step": 5533 }, { "epoch": 2.93, "learning_rate": 7.724193293756382e-06, "loss": 0.1194, "step": 5534 }, { "epoch": 2.93, "learning_rate": 7.720854103502148e-06, "loss": 0.0877, "step": 5535 }, { "epoch": 2.93, "learning_rate": 7.717515181279575e-06, "loss": 0.131, "step": 5536 }, { "epoch": 2.93, "learning_rate": 7.714176527481329e-06, "loss": 0.1463, "step": 5537 }, { "epoch": 2.93, "learning_rate": 7.710838142500045e-06, "loss": 0.1899, "step": 5538 }, { "epoch": 2.93, "learning_rate": 7.707500026728318e-06, "loss": 0.1647, "step": 5539 }, { "epoch": 2.93, "learning_rate": 7.704162180558718e-06, "loss": 0.0685, "step": 5540 }, { "epoch": 2.93, "learning_rate": 7.70082460438378e-06, "loss": 0.1663, "step": 5541 }, { "epoch": 2.93, "learning_rate": 7.697487298596009e-06, "loss": 0.1274, "step": 5542 }, { "epoch": 2.93, "learning_rate": 7.694150263587883e-06, "loss": 0.1109, "step": 5543 }, { "epoch": 2.93, "learning_rate": 7.69081349975184e-06, "loss": 0.1042, "step": 5544 }, { "epoch": 2.94, "learning_rate": 7.687477007480288e-06, "loss": 0.1256, "step": 5545 }, { "epoch": 2.94, "learning_rate": 7.684140787165608e-06, "loss": 0.0934, "step": 5546 }, { "epoch": 2.94, "learning_rate": 7.680804839200143e-06, "loss": 0.1279, "step": 5547 }, { "epoch": 2.94, "learning_rate": 7.677469163976208e-06, "loss": 0.1516, "step": 5548 }, { "epoch": 2.94, "learning_rate": 7.674133761886082e-06, "loss": 0.1328, "step": 5549 }, { "epoch": 2.94, "learning_rate": 7.670798633322016e-06, "loss": 0.1478, "step": 5550 }, { "epoch": 2.94, "learning_rate": 7.667463778676226e-06, "loss": 0.1665, "step": 5551 }, { "epoch": 2.94, "learning_rate": 7.664129198340899e-06, "loss": 0.1554, "step": 5552 }, { "epoch": 2.94, "learning_rate": 7.660794892708184e-06, "loss": 0.164, "step": 5553 }, { "epoch": 2.94, "learning_rate": 7.657460862170212e-06, "loss": 0.0793, "step": 5554 }, { "epoch": 2.94, "learning_rate": 7.654127107119057e-06, "loss": 0.1512, "step": 5555 }, { "epoch": 2.94, "learning_rate": 7.65079362794678e-06, "loss": 0.1114, "step": 5556 }, { "epoch": 2.94, "learning_rate": 7.647460425045403e-06, "loss": 0.1379, "step": 5557 }, { "epoch": 2.94, "learning_rate": 7.64412749880692e-06, "loss": 0.1869, "step": 5558 }, { "epoch": 2.94, "learning_rate": 7.64079484962329e-06, "loss": 0.1114, "step": 5559 }, { "epoch": 2.94, "learning_rate": 7.637462477886433e-06, "loss": 0.129, "step": 5560 }, { "epoch": 2.94, "learning_rate": 7.634130383988245e-06, "loss": 0.097, "step": 5561 }, { "epoch": 2.94, "learning_rate": 7.630798568320587e-06, "loss": 0.1242, "step": 5562 }, { "epoch": 2.94, "learning_rate": 7.627467031275287e-06, "loss": 0.1378, "step": 5563 }, { "epoch": 2.95, "learning_rate": 7.6241357732441365e-06, "loss": 0.1164, "step": 5564 }, { "epoch": 2.95, "learning_rate": 7.6208047946189e-06, "loss": 0.1267, "step": 5565 }, { "epoch": 2.95, "learning_rate": 7.617474095791306e-06, "loss": 0.123, "step": 5566 }, { "epoch": 2.95, "learning_rate": 7.61414367715305e-06, "loss": 0.1205, "step": 5567 }, { "epoch": 2.95, "learning_rate": 7.610813539095796e-06, "loss": 0.1046, "step": 5568 }, { "epoch": 2.95, "learning_rate": 7.607483682011172e-06, "loss": 0.1825, "step": 5569 }, { "epoch": 2.95, "learning_rate": 7.60415410629078e-06, "loss": 0.1493, "step": 5570 }, { "epoch": 2.95, "learning_rate": 7.6008248123261855e-06, "loss": 0.1429, "step": 5571 }, { "epoch": 2.95, "learning_rate": 7.597495800508911e-06, "loss": 0.0894, "step": 5572 }, { "epoch": 2.95, "learning_rate": 7.594167071230456e-06, "loss": 0.1347, "step": 5573 }, { "epoch": 2.95, "learning_rate": 7.59083862488229e-06, "loss": 0.1007, "step": 5574 }, { "epoch": 2.95, "learning_rate": 7.587510461855842e-06, "loss": 0.1727, "step": 5575 }, { "epoch": 2.95, "learning_rate": 7.58418258254251e-06, "loss": 0.1254, "step": 5576 }, { "epoch": 2.95, "learning_rate": 7.580854987333658e-06, "loss": 0.1694, "step": 5577 }, { "epoch": 2.95, "learning_rate": 7.577527676620618e-06, "loss": 0.1031, "step": 5578 }, { "epoch": 2.95, "learning_rate": 7.5742006507946896e-06, "loss": 0.1007, "step": 5579 }, { "epoch": 2.95, "learning_rate": 7.570873910247134e-06, "loss": 0.106, "step": 5580 }, { "epoch": 2.95, "learning_rate": 7.567547455369182e-06, "loss": 0.1445, "step": 5581 }, { "epoch": 2.95, "learning_rate": 7.5642212865520316e-06, "loss": 0.128, "step": 5582 }, { "epoch": 2.96, "learning_rate": 7.560895404186845e-06, "loss": 0.1304, "step": 5583 }, { "epoch": 2.96, "learning_rate": 7.557569808664752e-06, "loss": 0.0776, "step": 5584 }, { "epoch": 2.96, "learning_rate": 7.554244500376851e-06, "loss": 0.1088, "step": 5585 }, { "epoch": 2.96, "learning_rate": 7.550919479714205e-06, "loss": 0.1109, "step": 5586 }, { "epoch": 2.96, "learning_rate": 7.547594747067843e-06, "loss": 0.1758, "step": 5587 }, { "epoch": 2.96, "learning_rate": 7.544270302828751e-06, "loss": 0.133, "step": 5588 }, { "epoch": 2.96, "learning_rate": 7.540946147387898e-06, "loss": 0.144, "step": 5589 }, { "epoch": 2.96, "learning_rate": 7.537622281136207e-06, "loss": 0.1459, "step": 5590 }, { "epoch": 2.96, "learning_rate": 7.534298704464572e-06, "loss": 0.165, "step": 5591 }, { "epoch": 2.96, "learning_rate": 7.530975417763853e-06, "loss": 0.0874, "step": 5592 }, { "epoch": 2.96, "learning_rate": 7.527652421424872e-06, "loss": 0.0631, "step": 5593 }, { "epoch": 2.96, "learning_rate": 7.52432971583842e-06, "loss": 0.0915, "step": 5594 }, { "epoch": 2.96, "learning_rate": 7.521007301395256e-06, "loss": 0.1496, "step": 5595 }, { "epoch": 2.96, "learning_rate": 7.517685178486097e-06, "loss": 0.1739, "step": 5596 }, { "epoch": 2.96, "learning_rate": 7.514363347501634e-06, "loss": 0.1433, "step": 5597 }, { "epoch": 2.96, "learning_rate": 7.511041808832519e-06, "loss": 0.1338, "step": 5598 }, { "epoch": 2.96, "learning_rate": 7.507720562869371e-06, "loss": 0.1441, "step": 5599 }, { "epoch": 2.96, "learning_rate": 7.504399610002773e-06, "loss": 0.141, "step": 5600 }, { "epoch": 2.96, "learning_rate": 7.501078950623278e-06, "loss": 0.0999, "step": 5601 }, { "epoch": 2.97, "learning_rate": 7.4977585851214015e-06, "loss": 0.0881, "step": 5602 }, { "epoch": 2.97, "learning_rate": 7.494438513887627e-06, "loss": 0.1379, "step": 5603 }, { "epoch": 2.97, "learning_rate": 7.4911187373123915e-06, "loss": 0.2015, "step": 5604 }, { "epoch": 2.97, "learning_rate": 7.487799255786114e-06, "loss": 0.0999, "step": 5605 }, { "epoch": 2.97, "learning_rate": 7.48448006969917e-06, "loss": 0.1776, "step": 5606 }, { "epoch": 2.97, "learning_rate": 7.4811611794419025e-06, "loss": 0.144, "step": 5607 }, { "epoch": 2.97, "learning_rate": 7.477842585404618e-06, "loss": 0.1065, "step": 5608 }, { "epoch": 2.97, "learning_rate": 7.474524287977587e-06, "loss": 0.127, "step": 5609 }, { "epoch": 2.97, "learning_rate": 7.471206287551052e-06, "loss": 0.147, "step": 5610 }, { "epoch": 2.97, "learning_rate": 7.467888584515215e-06, "loss": 0.1142, "step": 5611 }, { "epoch": 2.97, "learning_rate": 7.46457117926024e-06, "loss": 0.1525, "step": 5612 }, { "epoch": 2.97, "learning_rate": 7.461254072176262e-06, "loss": 0.1395, "step": 5613 }, { "epoch": 2.97, "learning_rate": 7.457937263653379e-06, "loss": 0.1207, "step": 5614 }, { "epoch": 2.97, "learning_rate": 7.454620754081651e-06, "loss": 0.0935, "step": 5615 }, { "epoch": 2.97, "learning_rate": 7.45130454385111e-06, "loss": 0.1266, "step": 5616 }, { "epoch": 2.97, "learning_rate": 7.447988633351746e-06, "loss": 0.0771, "step": 5617 }, { "epoch": 2.97, "learning_rate": 7.444673022973518e-06, "loss": 0.1865, "step": 5618 }, { "epoch": 2.97, "learning_rate": 7.441357713106347e-06, "loss": 0.1285, "step": 5619 }, { "epoch": 2.97, "learning_rate": 7.438042704140116e-06, "loss": 0.1378, "step": 5620 }, { "epoch": 2.98, "learning_rate": 7.434727996464679e-06, "loss": 0.1736, "step": 5621 }, { "epoch": 2.98, "learning_rate": 7.431413590469852e-06, "loss": 0.1136, "step": 5622 }, { "epoch": 2.98, "learning_rate": 7.428099486545412e-06, "loss": 0.0837, "step": 5623 }, { "epoch": 2.98, "learning_rate": 7.424785685081109e-06, "loss": 0.1235, "step": 5624 }, { "epoch": 2.98, "learning_rate": 7.421472186466647e-06, "loss": 0.2035, "step": 5625 }, { "epoch": 2.98, "learning_rate": 7.418158991091699e-06, "loss": 0.1137, "step": 5626 }, { "epoch": 2.98, "learning_rate": 7.4148460993459115e-06, "loss": 0.1309, "step": 5627 }, { "epoch": 2.98, "learning_rate": 7.4115335116188755e-06, "loss": 0.0868, "step": 5628 }, { "epoch": 2.98, "learning_rate": 7.408221228300163e-06, "loss": 0.1634, "step": 5629 }, { "epoch": 2.98, "learning_rate": 7.404909249779301e-06, "loss": 0.1214, "step": 5630 }, { "epoch": 2.98, "learning_rate": 7.401597576445787e-06, "loss": 0.1707, "step": 5631 }, { "epoch": 2.98, "learning_rate": 7.398286208689081e-06, "loss": 0.1459, "step": 5632 }, { "epoch": 2.98, "learning_rate": 7.394975146898604e-06, "loss": 0.1187, "step": 5633 }, { "epoch": 2.98, "learning_rate": 7.391664391463742e-06, "loss": 0.1689, "step": 5634 }, { "epoch": 2.98, "learning_rate": 7.38835394277385e-06, "loss": 0.1248, "step": 5635 }, { "epoch": 2.98, "learning_rate": 7.385043801218234e-06, "loss": 0.1273, "step": 5636 }, { "epoch": 2.98, "learning_rate": 7.381733967186182e-06, "loss": 0.1615, "step": 5637 }, { "epoch": 2.98, "learning_rate": 7.3784244410669296e-06, "loss": 0.1017, "step": 5638 }, { "epoch": 2.98, "learning_rate": 7.375115223249687e-06, "loss": 0.1413, "step": 5639 }, { "epoch": 2.99, "learning_rate": 7.371806314123624e-06, "loss": 0.157, "step": 5640 }, { "epoch": 2.99, "learning_rate": 7.368497714077871e-06, "loss": 0.1174, "step": 5641 }, { "epoch": 2.99, "learning_rate": 7.365189423501528e-06, "loss": 0.1232, "step": 5642 }, { "epoch": 2.99, "learning_rate": 7.3618814427836605e-06, "loss": 0.1619, "step": 5643 }, { "epoch": 2.99, "learning_rate": 7.358573772313285e-06, "loss": 0.1159, "step": 5644 }, { "epoch": 2.99, "learning_rate": 7.355266412479391e-06, "loss": 0.1264, "step": 5645 }, { "epoch": 2.99, "learning_rate": 7.351959363670931e-06, "loss": 0.1024, "step": 5646 }, { "epoch": 2.99, "learning_rate": 7.348652626276822e-06, "loss": 0.0866, "step": 5647 }, { "epoch": 2.99, "learning_rate": 7.34534620068594e-06, "loss": 0.1563, "step": 5648 }, { "epoch": 2.99, "learning_rate": 7.342040087287128e-06, "loss": 0.1051, "step": 5649 }, { "epoch": 2.99, "learning_rate": 7.33873428646919e-06, "loss": 0.1177, "step": 5650 }, { "epoch": 2.99, "learning_rate": 7.335428798620896e-06, "loss": 0.1273, "step": 5651 }, { "epoch": 2.99, "learning_rate": 7.332123624130975e-06, "loss": 0.1236, "step": 5652 }, { "epoch": 2.99, "learning_rate": 7.328818763388121e-06, "loss": 0.1738, "step": 5653 }, { "epoch": 2.99, "learning_rate": 7.3255142167809935e-06, "loss": 0.1176, "step": 5654 }, { "epoch": 2.99, "learning_rate": 7.322209984698212e-06, "loss": 0.1134, "step": 5655 }, { "epoch": 2.99, "learning_rate": 7.318906067528359e-06, "loss": 0.0997, "step": 5656 }, { "epoch": 2.99, "learning_rate": 7.315602465659983e-06, "loss": 0.1093, "step": 5657 }, { "epoch": 2.99, "learning_rate": 7.312299179481595e-06, "loss": 0.144, "step": 5658 }, { "epoch": 3.0, "learning_rate": 7.308996209381669e-06, "loss": 0.0723, "step": 5659 }, { "epoch": 3.0, "learning_rate": 7.305693555748632e-06, "loss": 0.0837, "step": 5660 }, { "epoch": 3.0, "learning_rate": 7.302391218970886e-06, "loss": 0.099, "step": 5661 }, { "epoch": 3.0, "learning_rate": 7.299089199436794e-06, "loss": 0.1401, "step": 5662 }, { "epoch": 3.0, "learning_rate": 7.295787497534679e-06, "loss": 0.1708, "step": 5663 }, { "epoch": 3.0, "learning_rate": 7.292486113652824e-06, "loss": 0.2638, "step": 5664 }, { "epoch": 3.0, "learning_rate": 7.289185048179482e-06, "loss": 0.0992, "step": 5665 }, { "epoch": 3.0, "learning_rate": 7.285884301502861e-06, "loss": 0.1263, "step": 5666 }, { "epoch": 3.0, "learning_rate": 7.282583874011138e-06, "loss": 0.0925, "step": 5667 }, { "epoch": 3.0, "learning_rate": 7.279283766092444e-06, "loss": 0.0956, "step": 5668 }, { "epoch": 3.0, "learning_rate": 7.275983978134882e-06, "loss": 0.0658, "step": 5669 }, { "epoch": 3.0, "learning_rate": 7.272684510526511e-06, "loss": 0.0944, "step": 5670 }, { "epoch": 3.0, "learning_rate": 7.269385363655356e-06, "loss": 0.1308, "step": 5671 }, { "epoch": 3.0, "learning_rate": 7.2660865379094e-06, "loss": 0.0651, "step": 5672 }, { "epoch": 3.0, "learning_rate": 7.262788033676592e-06, "loss": 0.0493, "step": 5673 }, { "epoch": 3.0, "learning_rate": 7.259489851344844e-06, "loss": 0.084, "step": 5674 }, { "epoch": 3.0, "learning_rate": 7.25619199130203e-06, "loss": 0.109, "step": 5675 }, { "epoch": 3.0, "learning_rate": 7.252894453935979e-06, "loss": 0.1117, "step": 5676 }, { "epoch": 3.0, "learning_rate": 7.249597239634485e-06, "loss": 0.0746, "step": 5677 }, { "epoch": 3.01, "learning_rate": 7.2463003487853134e-06, "loss": 0.1075, "step": 5678 }, { "epoch": 3.01, "learning_rate": 7.243003781776182e-06, "loss": 0.0969, "step": 5679 }, { "epoch": 3.01, "learning_rate": 7.239707538994774e-06, "loss": 0.0672, "step": 5680 }, { "epoch": 3.01, "learning_rate": 7.2364116208287295e-06, "loss": 0.0997, "step": 5681 }, { "epoch": 3.01, "learning_rate": 7.233116027665659e-06, "loss": 0.0668, "step": 5682 }, { "epoch": 3.01, "learning_rate": 7.2298207598931296e-06, "loss": 0.1221, "step": 5683 }, { "epoch": 3.01, "learning_rate": 7.226525817898668e-06, "loss": 0.1008, "step": 5684 }, { "epoch": 3.01, "learning_rate": 7.223231202069767e-06, "loss": 0.0522, "step": 5685 }, { "epoch": 3.01, "learning_rate": 7.219936912793879e-06, "loss": 0.1319, "step": 5686 }, { "epoch": 3.01, "learning_rate": 7.216642950458419e-06, "loss": 0.0772, "step": 5687 }, { "epoch": 3.01, "learning_rate": 7.213349315450759e-06, "loss": 0.1078, "step": 5688 }, { "epoch": 3.01, "learning_rate": 7.210056008158244e-06, "loss": 0.0814, "step": 5689 }, { "epoch": 3.01, "learning_rate": 7.206763028968167e-06, "loss": 0.1597, "step": 5690 }, { "epoch": 3.01, "learning_rate": 7.203470378267791e-06, "loss": 0.1037, "step": 5691 }, { "epoch": 3.01, "learning_rate": 7.200178056444338e-06, "loss": 0.1073, "step": 5692 }, { "epoch": 3.01, "learning_rate": 7.196886063884988e-06, "loss": 0.1132, "step": 5693 }, { "epoch": 3.01, "learning_rate": 7.1935944009768865e-06, "loss": 0.1159, "step": 5694 }, { "epoch": 3.01, "learning_rate": 7.19030306810714e-06, "loss": 0.1323, "step": 5695 }, { "epoch": 3.01, "learning_rate": 7.187012065662814e-06, "loss": 0.1021, "step": 5696 }, { "epoch": 3.02, "learning_rate": 7.183721394030937e-06, "loss": 0.1467, "step": 5697 }, { "epoch": 3.02, "learning_rate": 7.180431053598498e-06, "loss": 0.0869, "step": 5698 }, { "epoch": 3.02, "learning_rate": 7.1771410447524444e-06, "loss": 0.1196, "step": 5699 }, { "epoch": 3.02, "learning_rate": 7.1738513678796966e-06, "loss": 0.0734, "step": 5700 }, { "epoch": 3.02, "learning_rate": 7.170562023367114e-06, "loss": 0.0673, "step": 5701 }, { "epoch": 3.02, "learning_rate": 7.1672730116015344e-06, "loss": 0.0768, "step": 5702 }, { "epoch": 3.02, "learning_rate": 7.163984332969751e-06, "loss": 0.0542, "step": 5703 }, { "epoch": 3.02, "learning_rate": 7.160695987858521e-06, "loss": 0.088, "step": 5704 }, { "epoch": 3.02, "learning_rate": 7.157407976654559e-06, "loss": 0.0902, "step": 5705 }, { "epoch": 3.02, "learning_rate": 7.154120299744539e-06, "loss": 0.1, "step": 5706 }, { "epoch": 3.02, "learning_rate": 7.1508329575151e-06, "loss": 0.099, "step": 5707 }, { "epoch": 3.02, "learning_rate": 7.147545950352839e-06, "loss": 0.0995, "step": 5708 }, { "epoch": 3.02, "learning_rate": 7.144259278644312e-06, "loss": 0.1309, "step": 5709 }, { "epoch": 3.02, "learning_rate": 7.140972942776037e-06, "loss": 0.1054, "step": 5710 }, { "epoch": 3.02, "learning_rate": 7.137686943134496e-06, "loss": 0.0857, "step": 5711 }, { "epoch": 3.02, "learning_rate": 7.134401280106127e-06, "loss": 0.1014, "step": 5712 }, { "epoch": 3.02, "learning_rate": 7.13111595407733e-06, "loss": 0.1059, "step": 5713 }, { "epoch": 3.02, "learning_rate": 7.127830965434465e-06, "loss": 0.0662, "step": 5714 }, { "epoch": 3.03, "learning_rate": 7.124546314563851e-06, "loss": 0.1045, "step": 5715 }, { "epoch": 3.03, "learning_rate": 7.121262001851778e-06, "loss": 0.0956, "step": 5716 }, { "epoch": 3.03, "learning_rate": 7.117978027684475e-06, "loss": 0.0966, "step": 5717 }, { "epoch": 3.03, "learning_rate": 7.114694392448147e-06, "loss": 0.1584, "step": 5718 }, { "epoch": 3.03, "learning_rate": 7.111411096528955e-06, "loss": 0.0514, "step": 5719 }, { "epoch": 3.03, "learning_rate": 7.108128140313024e-06, "loss": 0.0956, "step": 5720 }, { "epoch": 3.03, "learning_rate": 7.104845524186434e-06, "loss": 0.1156, "step": 5721 }, { "epoch": 3.03, "learning_rate": 7.101563248535227e-06, "loss": 0.0787, "step": 5722 }, { "epoch": 3.03, "learning_rate": 7.098281313745403e-06, "loss": 0.0771, "step": 5723 }, { "epoch": 3.03, "learning_rate": 7.094999720202927e-06, "loss": 0.0665, "step": 5724 }, { "epoch": 3.03, "learning_rate": 7.091718468293714e-06, "loss": 0.0765, "step": 5725 }, { "epoch": 3.03, "learning_rate": 7.088437558403649e-06, "loss": 0.0842, "step": 5726 }, { "epoch": 3.03, "learning_rate": 7.085156990918572e-06, "loss": 0.099, "step": 5727 }, { "epoch": 3.03, "learning_rate": 7.081876766224285e-06, "loss": 0.0823, "step": 5728 }, { "epoch": 3.03, "learning_rate": 7.078596884706548e-06, "loss": 0.1102, "step": 5729 }, { "epoch": 3.03, "learning_rate": 7.075317346751076e-06, "loss": 0.0998, "step": 5730 }, { "epoch": 3.03, "learning_rate": 7.072038152743557e-06, "loss": 0.1287, "step": 5731 }, { "epoch": 3.03, "learning_rate": 7.0687593030696315e-06, "loss": 0.0748, "step": 5732 }, { "epoch": 3.03, "learning_rate": 7.065480798114887e-06, "loss": 0.0879, "step": 5733 }, { "epoch": 3.04, "learning_rate": 7.062202638264885e-06, "loss": 0.1047, "step": 5734 }, { "epoch": 3.04, "learning_rate": 7.058924823905148e-06, "loss": 0.1096, "step": 5735 }, { "epoch": 3.04, "learning_rate": 7.0556473554211505e-06, "loss": 0.1091, "step": 5736 }, { "epoch": 3.04, "learning_rate": 7.052370233198328e-06, "loss": 0.0634, "step": 5737 }, { "epoch": 3.04, "learning_rate": 7.049093457622076e-06, "loss": 0.1479, "step": 5738 }, { "epoch": 3.04, "learning_rate": 7.04581702907775e-06, "loss": 0.0917, "step": 5739 }, { "epoch": 3.04, "learning_rate": 7.042540947950667e-06, "loss": 0.1062, "step": 5740 }, { "epoch": 3.04, "learning_rate": 7.039265214626093e-06, "loss": 0.1543, "step": 5741 }, { "epoch": 3.04, "learning_rate": 7.035989829489265e-06, "loss": 0.081, "step": 5742 }, { "epoch": 3.04, "learning_rate": 7.0327147929253715e-06, "loss": 0.0716, "step": 5743 }, { "epoch": 3.04, "learning_rate": 7.029440105319565e-06, "loss": 0.0607, "step": 5744 }, { "epoch": 3.04, "learning_rate": 7.026165767056954e-06, "loss": 0.0618, "step": 5745 }, { "epoch": 3.04, "learning_rate": 7.022891778522602e-06, "loss": 0.0815, "step": 5746 }, { "epoch": 3.04, "learning_rate": 7.019618140101545e-06, "loss": 0.0827, "step": 5747 }, { "epoch": 3.04, "learning_rate": 7.016344852178767e-06, "loss": 0.082, "step": 5748 }, { "epoch": 3.04, "learning_rate": 7.013071915139207e-06, "loss": 0.0765, "step": 5749 }, { "epoch": 3.04, "learning_rate": 7.009799329367768e-06, "loss": 0.1393, "step": 5750 }, { "epoch": 3.04, "learning_rate": 7.006527095249316e-06, "loss": 0.1123, "step": 5751 }, { "epoch": 3.04, "learning_rate": 7.003255213168672e-06, "loss": 0.0706, "step": 5752 }, { "epoch": 3.05, "learning_rate": 6.999983683510612e-06, "loss": 0.1228, "step": 5753 }, { "epoch": 3.05, "learning_rate": 6.996712506659877e-06, "loss": 0.1155, "step": 5754 }, { "epoch": 3.05, "learning_rate": 6.993441683001161e-06, "loss": 0.0869, "step": 5755 }, { "epoch": 3.05, "learning_rate": 6.990171212919124e-06, "loss": 0.0885, "step": 5756 }, { "epoch": 3.05, "learning_rate": 6.98690109679837e-06, "loss": 0.045, "step": 5757 }, { "epoch": 3.05, "learning_rate": 6.983631335023475e-06, "loss": 0.0745, "step": 5758 }, { "epoch": 3.05, "learning_rate": 6.980361927978972e-06, "loss": 0.0842, "step": 5759 }, { "epoch": 3.05, "learning_rate": 6.9770928760493436e-06, "loss": 0.124, "step": 5760 }, { "epoch": 3.05, "learning_rate": 6.973824179619038e-06, "loss": 0.1538, "step": 5761 }, { "epoch": 3.05, "learning_rate": 6.970555839072464e-06, "loss": 0.0595, "step": 5762 }, { "epoch": 3.05, "learning_rate": 6.967287854793979e-06, "loss": 0.1127, "step": 5763 }, { "epoch": 3.05, "learning_rate": 6.964020227167911e-06, "loss": 0.0731, "step": 5764 }, { "epoch": 3.05, "learning_rate": 6.960752956578527e-06, "loss": 0.1004, "step": 5765 }, { "epoch": 3.05, "learning_rate": 6.957486043410073e-06, "loss": 0.0589, "step": 5766 }, { "epoch": 3.05, "learning_rate": 6.9542194880467425e-06, "loss": 0.1203, "step": 5767 }, { "epoch": 3.05, "learning_rate": 6.950953290872688e-06, "loss": 0.0926, "step": 5768 }, { "epoch": 3.05, "learning_rate": 6.947687452272018e-06, "loss": 0.1235, "step": 5769 }, { "epoch": 3.05, "learning_rate": 6.944421972628803e-06, "loss": 0.078, "step": 5770 }, { "epoch": 3.05, "learning_rate": 6.9411568523270685e-06, "loss": 0.0946, "step": 5771 }, { "epoch": 3.06, "learning_rate": 6.937892091750802e-06, "loss": 0.0982, "step": 5772 }, { "epoch": 3.06, "learning_rate": 6.934627691283938e-06, "loss": 0.0463, "step": 5773 }, { "epoch": 3.06, "learning_rate": 6.931363651310382e-06, "loss": 0.0645, "step": 5774 }, { "epoch": 3.06, "learning_rate": 6.928099972213989e-06, "loss": 0.1402, "step": 5775 }, { "epoch": 3.06, "learning_rate": 6.924836654378571e-06, "loss": 0.0732, "step": 5776 }, { "epoch": 3.06, "learning_rate": 6.921573698187903e-06, "loss": 0.1512, "step": 5777 }, { "epoch": 3.06, "learning_rate": 6.918311104025716e-06, "loss": 0.0784, "step": 5778 }, { "epoch": 3.06, "learning_rate": 6.915048872275693e-06, "loss": 0.1039, "step": 5779 }, { "epoch": 3.06, "learning_rate": 6.911787003321487e-06, "loss": 0.0818, "step": 5780 }, { "epoch": 3.06, "learning_rate": 6.908525497546686e-06, "loss": 0.1363, "step": 5781 }, { "epoch": 3.06, "learning_rate": 6.905264355334858e-06, "loss": 0.1344, "step": 5782 }, { "epoch": 3.06, "learning_rate": 6.902003577069517e-06, "loss": 0.0787, "step": 5783 }, { "epoch": 3.06, "learning_rate": 6.898743163134136e-06, "loss": 0.064, "step": 5784 }, { "epoch": 3.06, "learning_rate": 6.895483113912147e-06, "loss": 0.1209, "step": 5785 }, { "epoch": 3.06, "learning_rate": 6.892223429786938e-06, "loss": 0.0848, "step": 5786 }, { "epoch": 3.06, "learning_rate": 6.8889641111418505e-06, "loss": 0.131, "step": 5787 }, { "epoch": 3.06, "learning_rate": 6.885705158360191e-06, "loss": 0.0965, "step": 5788 }, { "epoch": 3.06, "learning_rate": 6.882446571825213e-06, "loss": 0.1083, "step": 5789 }, { "epoch": 3.06, "learning_rate": 6.879188351920135e-06, "loss": 0.1256, "step": 5790 }, { "epoch": 3.07, "learning_rate": 6.875930499028129e-06, "loss": 0.0865, "step": 5791 }, { "epoch": 3.07, "learning_rate": 6.872673013532323e-06, "loss": 0.123, "step": 5792 }, { "epoch": 3.07, "learning_rate": 6.8694158958158075e-06, "loss": 0.1387, "step": 5793 }, { "epoch": 3.07, "learning_rate": 6.866159146261621e-06, "loss": 0.0605, "step": 5794 }, { "epoch": 3.07, "learning_rate": 6.862902765252765e-06, "loss": 0.0871, "step": 5795 }, { "epoch": 3.07, "learning_rate": 6.859646753172198e-06, "loss": 0.1322, "step": 5796 }, { "epoch": 3.07, "learning_rate": 6.856391110402829e-06, "loss": 0.094, "step": 5797 }, { "epoch": 3.07, "learning_rate": 6.853135837327527e-06, "loss": 0.0744, "step": 5798 }, { "epoch": 3.07, "learning_rate": 6.8498809343291205e-06, "loss": 0.0565, "step": 5799 }, { "epoch": 3.07, "learning_rate": 6.846626401790393e-06, "loss": 0.0912, "step": 5800 }, { "epoch": 3.07, "learning_rate": 6.84337224009408e-06, "loss": 0.0873, "step": 5801 }, { "epoch": 3.07, "learning_rate": 6.840118449622879e-06, "loss": 0.1003, "step": 5802 }, { "epoch": 3.07, "learning_rate": 6.836865030759439e-06, "loss": 0.0888, "step": 5803 }, { "epoch": 3.07, "learning_rate": 6.833611983886376e-06, "loss": 0.0999, "step": 5804 }, { "epoch": 3.07, "learning_rate": 6.830359309386243e-06, "loss": 0.0789, "step": 5805 }, { "epoch": 3.07, "learning_rate": 6.827107007641567e-06, "loss": 0.1398, "step": 5806 }, { "epoch": 3.07, "learning_rate": 6.82385507903482e-06, "loss": 0.1149, "step": 5807 }, { "epoch": 3.07, "learning_rate": 6.820603523948441e-06, "loss": 0.0965, "step": 5808 }, { "epoch": 3.07, "learning_rate": 6.817352342764816e-06, "loss": 0.0687, "step": 5809 }, { "epoch": 3.08, "learning_rate": 6.814101535866287e-06, "loss": 0.0379, "step": 5810 }, { "epoch": 3.08, "learning_rate": 6.810851103635159e-06, "loss": 0.0851, "step": 5811 }, { "epoch": 3.08, "learning_rate": 6.807601046453685e-06, "loss": 0.0608, "step": 5812 }, { "epoch": 3.08, "learning_rate": 6.804351364704082e-06, "loss": 0.1161, "step": 5813 }, { "epoch": 3.08, "learning_rate": 6.801102058768514e-06, "loss": 0.0953, "step": 5814 }, { "epoch": 3.08, "learning_rate": 6.797853129029107e-06, "loss": 0.1368, "step": 5815 }, { "epoch": 3.08, "learning_rate": 6.794604575867941e-06, "loss": 0.0958, "step": 5816 }, { "epoch": 3.08, "learning_rate": 6.791356399667051e-06, "loss": 0.0751, "step": 5817 }, { "epoch": 3.08, "learning_rate": 6.7881086008084296e-06, "loss": 0.1403, "step": 5818 }, { "epoch": 3.08, "learning_rate": 6.7848611796740205e-06, "loss": 0.1379, "step": 5819 }, { "epoch": 3.08, "learning_rate": 6.781614136645731e-06, "loss": 0.111, "step": 5820 }, { "epoch": 3.08, "learning_rate": 6.7783674721054225e-06, "loss": 0.0486, "step": 5821 }, { "epoch": 3.08, "learning_rate": 6.775121186434897e-06, "loss": 0.0569, "step": 5822 }, { "epoch": 3.08, "learning_rate": 6.771875280015929e-06, "loss": 0.1499, "step": 5823 }, { "epoch": 3.08, "learning_rate": 6.768629753230247e-06, "loss": 0.0863, "step": 5824 }, { "epoch": 3.08, "learning_rate": 6.765384606459526e-06, "loss": 0.1082, "step": 5825 }, { "epoch": 3.08, "learning_rate": 6.762139840085402e-06, "loss": 0.1365, "step": 5826 }, { "epoch": 3.08, "learning_rate": 6.758895454489468e-06, "loss": 0.1141, "step": 5827 }, { "epoch": 3.08, "learning_rate": 6.755651450053267e-06, "loss": 0.1022, "step": 5828 }, { "epoch": 3.09, "learning_rate": 6.7524078271583025e-06, "loss": 0.0734, "step": 5829 }, { "epoch": 3.09, "learning_rate": 6.749164586186025e-06, "loss": 0.0879, "step": 5830 }, { "epoch": 3.09, "learning_rate": 6.74592172751785e-06, "loss": 0.1304, "step": 5831 }, { "epoch": 3.09, "learning_rate": 6.742679251535143e-06, "loss": 0.0945, "step": 5832 }, { "epoch": 3.09, "learning_rate": 6.739437158619222e-06, "loss": 0.1222, "step": 5833 }, { "epoch": 3.09, "learning_rate": 6.736195449151365e-06, "loss": 0.1008, "step": 5834 }, { "epoch": 3.09, "learning_rate": 6.732954123512805e-06, "loss": 0.1712, "step": 5835 }, { "epoch": 3.09, "learning_rate": 6.729713182084725e-06, "loss": 0.073, "step": 5836 }, { "epoch": 3.09, "learning_rate": 6.726472625248271e-06, "loss": 0.0879, "step": 5837 }, { "epoch": 3.09, "learning_rate": 6.723232453384527e-06, "loss": 0.0908, "step": 5838 }, { "epoch": 3.09, "learning_rate": 6.719992666874553e-06, "loss": 0.1091, "step": 5839 }, { "epoch": 3.09, "learning_rate": 6.7167532660993505e-06, "loss": 0.0853, "step": 5840 }, { "epoch": 3.09, "learning_rate": 6.713514251439879e-06, "loss": 0.0979, "step": 5841 }, { "epoch": 3.09, "learning_rate": 6.710275623277052e-06, "loss": 0.1006, "step": 5842 }, { "epoch": 3.09, "learning_rate": 6.707037381991738e-06, "loss": 0.1355, "step": 5843 }, { "epoch": 3.09, "learning_rate": 6.70379952796476e-06, "loss": 0.0813, "step": 5844 }, { "epoch": 3.09, "learning_rate": 6.700562061576898e-06, "loss": 0.1043, "step": 5845 }, { "epoch": 3.09, "learning_rate": 6.697324983208882e-06, "loss": 0.0924, "step": 5846 }, { "epoch": 3.09, "learning_rate": 6.694088293241396e-06, "loss": 0.1431, "step": 5847 }, { "epoch": 3.1, "learning_rate": 6.690851992055083e-06, "loss": 0.1097, "step": 5848 }, { "epoch": 3.1, "learning_rate": 6.687616080030536e-06, "loss": 0.1055, "step": 5849 }, { "epoch": 3.1, "learning_rate": 6.684380557548304e-06, "loss": 0.1223, "step": 5850 }, { "epoch": 3.1, "learning_rate": 6.681145424988895e-06, "loss": 0.0886, "step": 5851 }, { "epoch": 3.1, "learning_rate": 6.677910682732762e-06, "loss": 0.1286, "step": 5852 }, { "epoch": 3.1, "learning_rate": 6.6746763311603235e-06, "loss": 0.1426, "step": 5853 }, { "epoch": 3.1, "learning_rate": 6.671442370651932e-06, "loss": 0.0928, "step": 5854 }, { "epoch": 3.1, "learning_rate": 6.668208801587915e-06, "loss": 0.0898, "step": 5855 }, { "epoch": 3.1, "learning_rate": 6.6649756243485475e-06, "loss": 0.0878, "step": 5856 }, { "epoch": 3.1, "learning_rate": 6.661742839314054e-06, "loss": 0.0949, "step": 5857 }, { "epoch": 3.1, "learning_rate": 6.658510446864616e-06, "loss": 0.0616, "step": 5858 }, { "epoch": 3.1, "learning_rate": 6.655278447380371e-06, "loss": 0.084, "step": 5859 }, { "epoch": 3.1, "learning_rate": 6.652046841241405e-06, "loss": 0.1088, "step": 5860 }, { "epoch": 3.1, "learning_rate": 6.648815628827765e-06, "loss": 0.1013, "step": 5861 }, { "epoch": 3.1, "learning_rate": 6.6455848105194405e-06, "loss": 0.1035, "step": 5862 }, { "epoch": 3.1, "learning_rate": 6.642354386696388e-06, "loss": 0.1159, "step": 5863 }, { "epoch": 3.1, "learning_rate": 6.6391243577385075e-06, "loss": 0.0873, "step": 5864 }, { "epoch": 3.1, "learning_rate": 6.635894724025655e-06, "loss": 0.1141, "step": 5865 }, { "epoch": 3.1, "learning_rate": 6.632665485937646e-06, "loss": 0.0445, "step": 5866 }, { "epoch": 3.11, "learning_rate": 6.629436643854243e-06, "loss": 0.0958, "step": 5867 }, { "epoch": 3.11, "learning_rate": 6.626208198155162e-06, "loss": 0.1123, "step": 5868 }, { "epoch": 3.11, "learning_rate": 6.622980149220078e-06, "loss": 0.1158, "step": 5869 }, { "epoch": 3.11, "learning_rate": 6.61975249742861e-06, "loss": 0.0649, "step": 5870 }, { "epoch": 3.11, "learning_rate": 6.616525243160338e-06, "loss": 0.0951, "step": 5871 }, { "epoch": 3.11, "learning_rate": 6.613298386794793e-06, "loss": 0.1138, "step": 5872 }, { "epoch": 3.11, "learning_rate": 6.610071928711459e-06, "loss": 0.0876, "step": 5873 }, { "epoch": 3.11, "learning_rate": 6.606845869289773e-06, "loss": 0.1192, "step": 5874 }, { "epoch": 3.11, "learning_rate": 6.603620208909125e-06, "loss": 0.119, "step": 5875 }, { "epoch": 3.11, "learning_rate": 6.600394947948858e-06, "loss": 0.0953, "step": 5876 }, { "epoch": 3.11, "learning_rate": 6.597170086788277e-06, "loss": 0.1061, "step": 5877 }, { "epoch": 3.11, "learning_rate": 6.59394562580662e-06, "loss": 0.067, "step": 5878 }, { "epoch": 3.11, "learning_rate": 6.590721565383092e-06, "loss": 0.0998, "step": 5879 }, { "epoch": 3.11, "learning_rate": 6.587497905896849e-06, "loss": 0.0745, "step": 5880 }, { "epoch": 3.11, "learning_rate": 6.584274647727002e-06, "loss": 0.0913, "step": 5881 }, { "epoch": 3.11, "learning_rate": 6.58105179125261e-06, "loss": 0.0908, "step": 5882 }, { "epoch": 3.11, "learning_rate": 6.577829336852688e-06, "loss": 0.0441, "step": 5883 }, { "epoch": 3.11, "learning_rate": 6.5746072849062e-06, "loss": 0.0572, "step": 5884 }, { "epoch": 3.11, "learning_rate": 6.571385635792068e-06, "loss": 0.096, "step": 5885 }, { "epoch": 3.12, "learning_rate": 6.568164389889161e-06, "loss": 0.1481, "step": 5886 }, { "epoch": 3.12, "learning_rate": 6.5649435475763036e-06, "loss": 0.1, "step": 5887 }, { "epoch": 3.12, "learning_rate": 6.561723109232274e-06, "loss": 0.0858, "step": 5888 }, { "epoch": 3.12, "learning_rate": 6.5585030752358015e-06, "loss": 0.1189, "step": 5889 }, { "epoch": 3.12, "learning_rate": 6.555283445965568e-06, "loss": 0.0736, "step": 5890 }, { "epoch": 3.12, "learning_rate": 6.552064221800205e-06, "loss": 0.1221, "step": 5891 }, { "epoch": 3.12, "learning_rate": 6.5488454031183e-06, "loss": 0.0699, "step": 5892 }, { "epoch": 3.12, "learning_rate": 6.545626990298401e-06, "loss": 0.1002, "step": 5893 }, { "epoch": 3.12, "learning_rate": 6.542408983718986e-06, "loss": 0.1084, "step": 5894 }, { "epoch": 3.12, "learning_rate": 6.539191383758503e-06, "loss": 0.0875, "step": 5895 }, { "epoch": 3.12, "learning_rate": 6.535974190795346e-06, "loss": 0.085, "step": 5896 }, { "epoch": 3.12, "learning_rate": 6.532757405207868e-06, "loss": 0.2744, "step": 5897 }, { "epoch": 3.12, "learning_rate": 6.529541027374365e-06, "loss": 0.0942, "step": 5898 }, { "epoch": 3.12, "learning_rate": 6.526325057673088e-06, "loss": 0.1423, "step": 5899 }, { "epoch": 3.12, "learning_rate": 6.523109496482244e-06, "loss": 0.1189, "step": 5900 }, { "epoch": 3.12, "learning_rate": 6.519894344179989e-06, "loss": 0.109, "step": 5901 }, { "epoch": 3.12, "learning_rate": 6.5166796011444265e-06, "loss": 0.0692, "step": 5902 }, { "epoch": 3.12, "learning_rate": 6.513465267753618e-06, "loss": 0.1159, "step": 5903 }, { "epoch": 3.13, "learning_rate": 6.510251344385576e-06, "loss": 0.1091, "step": 5904 }, { "epoch": 3.13, "learning_rate": 6.5070378314182626e-06, "loss": 0.0971, "step": 5905 }, { "epoch": 3.13, "learning_rate": 6.503824729229593e-06, "loss": 0.0853, "step": 5906 }, { "epoch": 3.13, "learning_rate": 6.5006120381974314e-06, "loss": 0.115, "step": 5907 }, { "epoch": 3.13, "learning_rate": 6.497399758699601e-06, "loss": 0.0729, "step": 5908 }, { "epoch": 3.13, "learning_rate": 6.494187891113873e-06, "loss": 0.1007, "step": 5909 }, { "epoch": 3.13, "learning_rate": 6.4909764358179615e-06, "loss": 0.0975, "step": 5910 }, { "epoch": 3.13, "learning_rate": 6.48776539318954e-06, "loss": 0.1231, "step": 5911 }, { "epoch": 3.13, "learning_rate": 6.484554763606239e-06, "loss": 0.1448, "step": 5912 }, { "epoch": 3.13, "learning_rate": 6.481344547445629e-06, "loss": 0.0727, "step": 5913 }, { "epoch": 3.13, "learning_rate": 6.478134745085242e-06, "loss": 0.136, "step": 5914 }, { "epoch": 3.13, "learning_rate": 6.474925356902553e-06, "loss": 0.0826, "step": 5915 }, { "epoch": 3.13, "learning_rate": 6.471716383274992e-06, "loss": 0.0863, "step": 5916 }, { "epoch": 3.13, "learning_rate": 6.468507824579943e-06, "loss": 0.116, "step": 5917 }, { "epoch": 3.13, "learning_rate": 6.465299681194734e-06, "loss": 0.1004, "step": 5918 }, { "epoch": 3.13, "learning_rate": 6.46209195349665e-06, "loss": 0.1027, "step": 5919 }, { "epoch": 3.13, "learning_rate": 6.458884641862927e-06, "loss": 0.1244, "step": 5920 }, { "epoch": 3.13, "learning_rate": 6.4556777466707495e-06, "loss": 0.112, "step": 5921 }, { "epoch": 3.13, "learning_rate": 6.4524712682972535e-06, "loss": 0.1034, "step": 5922 }, { "epoch": 3.14, "learning_rate": 6.4492652071195264e-06, "loss": 0.0827, "step": 5923 }, { "epoch": 3.14, "learning_rate": 6.446059563514611e-06, "loss": 0.0943, "step": 5924 }, { "epoch": 3.14, "learning_rate": 6.442854337859494e-06, "loss": 0.0966, "step": 5925 }, { "epoch": 3.14, "learning_rate": 6.439649530531114e-06, "loss": 0.1255, "step": 5926 }, { "epoch": 3.14, "learning_rate": 6.436445141906361e-06, "loss": 0.088, "step": 5927 }, { "epoch": 3.14, "learning_rate": 6.4332411723620815e-06, "loss": 0.1404, "step": 5928 }, { "epoch": 3.14, "learning_rate": 6.430037622275065e-06, "loss": 0.1302, "step": 5929 }, { "epoch": 3.14, "learning_rate": 6.426834492022057e-06, "loss": 0.1688, "step": 5930 }, { "epoch": 3.14, "learning_rate": 6.423631781979749e-06, "loss": 0.1565, "step": 5931 }, { "epoch": 3.14, "learning_rate": 6.420429492524787e-06, "loss": 0.0814, "step": 5932 }, { "epoch": 3.14, "learning_rate": 6.417227624033766e-06, "loss": 0.1254, "step": 5933 }, { "epoch": 3.14, "learning_rate": 6.414026176883233e-06, "loss": 0.092, "step": 5934 }, { "epoch": 3.14, "learning_rate": 6.410825151449678e-06, "loss": 0.0811, "step": 5935 }, { "epoch": 3.14, "learning_rate": 6.407624548109553e-06, "loss": 0.1017, "step": 5936 }, { "epoch": 3.14, "learning_rate": 6.404424367239253e-06, "loss": 0.0656, "step": 5937 }, { "epoch": 3.14, "learning_rate": 6.401224609215122e-06, "loss": 0.0999, "step": 5938 }, { "epoch": 3.14, "learning_rate": 6.398025274413464e-06, "loss": 0.1145, "step": 5939 }, { "epoch": 3.14, "learning_rate": 6.394826363210523e-06, "loss": 0.1736, "step": 5940 }, { "epoch": 3.14, "learning_rate": 6.391627875982494e-06, "loss": 0.1258, "step": 5941 }, { "epoch": 3.15, "learning_rate": 6.388429813105532e-06, "loss": 0.0597, "step": 5942 }, { "epoch": 3.15, "learning_rate": 6.385232174955728e-06, "loss": 0.1283, "step": 5943 }, { "epoch": 3.15, "learning_rate": 6.382034961909132e-06, "loss": 0.1071, "step": 5944 }, { "epoch": 3.15, "learning_rate": 6.378838174341743e-06, "loss": 0.1735, "step": 5945 }, { "epoch": 3.15, "learning_rate": 6.375641812629508e-06, "loss": 0.0749, "step": 5946 }, { "epoch": 3.15, "learning_rate": 6.372445877148325e-06, "loss": 0.1304, "step": 5947 }, { "epoch": 3.15, "learning_rate": 6.369250368274044e-06, "loss": 0.1025, "step": 5948 }, { "epoch": 3.15, "learning_rate": 6.366055286382458e-06, "loss": 0.1157, "step": 5949 }, { "epoch": 3.15, "learning_rate": 6.362860631849321e-06, "loss": 0.0892, "step": 5950 }, { "epoch": 3.15, "learning_rate": 6.359666405050324e-06, "loss": 0.0531, "step": 5951 }, { "epoch": 3.15, "learning_rate": 6.356472606361117e-06, "loss": 0.1171, "step": 5952 }, { "epoch": 3.15, "learning_rate": 6.353279236157293e-06, "loss": 0.1065, "step": 5953 }, { "epoch": 3.15, "learning_rate": 6.350086294814402e-06, "loss": 0.1288, "step": 5954 }, { "epoch": 3.15, "learning_rate": 6.346893782707937e-06, "loss": 0.1206, "step": 5955 }, { "epoch": 3.15, "learning_rate": 6.343701700213347e-06, "loss": 0.087, "step": 5956 }, { "epoch": 3.15, "learning_rate": 6.340510047706023e-06, "loss": 0.1037, "step": 5957 }, { "epoch": 3.15, "learning_rate": 6.337318825561311e-06, "loss": 0.0837, "step": 5958 }, { "epoch": 3.15, "learning_rate": 6.334128034154504e-06, "loss": 0.0807, "step": 5959 }, { "epoch": 3.15, "learning_rate": 6.330937673860844e-06, "loss": 0.0647, "step": 5960 }, { "epoch": 3.16, "learning_rate": 6.3277477450555225e-06, "loss": 0.1133, "step": 5961 }, { "epoch": 3.16, "learning_rate": 6.3245582481136825e-06, "loss": 0.0701, "step": 5962 }, { "epoch": 3.16, "learning_rate": 6.321369183410415e-06, "loss": 0.1369, "step": 5963 }, { "epoch": 3.16, "learning_rate": 6.318180551320757e-06, "loss": 0.1188, "step": 5964 }, { "epoch": 3.16, "learning_rate": 6.314992352219699e-06, "loss": 0.1014, "step": 5965 }, { "epoch": 3.16, "learning_rate": 6.311804586482184e-06, "loss": 0.1051, "step": 5966 }, { "epoch": 3.16, "learning_rate": 6.308617254483092e-06, "loss": 0.1064, "step": 5967 }, { "epoch": 3.16, "learning_rate": 6.305430356597258e-06, "loss": 0.0783, "step": 5968 }, { "epoch": 3.16, "learning_rate": 6.30224389319947e-06, "loss": 0.0777, "step": 5969 }, { "epoch": 3.16, "learning_rate": 6.299057864664464e-06, "loss": 0.0849, "step": 5970 }, { "epoch": 3.16, "learning_rate": 6.29587227136692e-06, "loss": 0.1031, "step": 5971 }, { "epoch": 3.16, "learning_rate": 6.29268711368147e-06, "loss": 0.0958, "step": 5972 }, { "epoch": 3.16, "learning_rate": 6.289502391982696e-06, "loss": 0.0721, "step": 5973 }, { "epoch": 3.16, "learning_rate": 6.286318106645126e-06, "loss": 0.0955, "step": 5974 }, { "epoch": 3.16, "learning_rate": 6.2831342580432355e-06, "loss": 0.0754, "step": 5975 }, { "epoch": 3.16, "learning_rate": 6.279950846551452e-06, "loss": 0.0665, "step": 5976 }, { "epoch": 3.16, "learning_rate": 6.276767872544152e-06, "loss": 0.084, "step": 5977 }, { "epoch": 3.16, "learning_rate": 6.273585336395658e-06, "loss": 0.0834, "step": 5978 }, { "epoch": 3.16, "learning_rate": 6.270403238480241e-06, "loss": 0.084, "step": 5979 }, { "epoch": 3.17, "learning_rate": 6.267221579172119e-06, "loss": 0.049, "step": 5980 }, { "epoch": 3.17, "learning_rate": 6.264040358845468e-06, "loss": 0.1291, "step": 5981 }, { "epoch": 3.17, "learning_rate": 6.2608595778744055e-06, "loss": 0.0578, "step": 5982 }, { "epoch": 3.17, "learning_rate": 6.257679236632988e-06, "loss": 0.0832, "step": 5983 }, { "epoch": 3.17, "learning_rate": 6.254499335495232e-06, "loss": 0.1084, "step": 5984 }, { "epoch": 3.17, "learning_rate": 6.251319874835105e-06, "loss": 0.1083, "step": 5985 }, { "epoch": 3.17, "learning_rate": 6.248140855026513e-06, "loss": 0.0961, "step": 5986 }, { "epoch": 3.17, "learning_rate": 6.244962276443316e-06, "loss": 0.0959, "step": 5987 }, { "epoch": 3.17, "learning_rate": 6.241784139459322e-06, "loss": 0.1141, "step": 5988 }, { "epoch": 3.17, "learning_rate": 6.238606444448282e-06, "loss": 0.1354, "step": 5989 }, { "epoch": 3.17, "learning_rate": 6.235429191783903e-06, "loss": 0.1187, "step": 5990 }, { "epoch": 3.17, "learning_rate": 6.232252381839832e-06, "loss": 0.0987, "step": 5991 }, { "epoch": 3.17, "learning_rate": 6.229076014989668e-06, "loss": 0.1096, "step": 5992 }, { "epoch": 3.17, "learning_rate": 6.225900091606958e-06, "loss": 0.0816, "step": 5993 }, { "epoch": 3.17, "learning_rate": 6.222724612065197e-06, "loss": 0.0979, "step": 5994 }, { "epoch": 3.17, "learning_rate": 6.219549576737826e-06, "loss": 0.0746, "step": 5995 }, { "epoch": 3.17, "learning_rate": 6.216374985998234e-06, "loss": 0.0906, "step": 5996 }, { "epoch": 3.17, "learning_rate": 6.213200840219762e-06, "loss": 0.0709, "step": 5997 }, { "epoch": 3.17, "learning_rate": 6.210027139775696e-06, "loss": 0.0672, "step": 5998 }, { "epoch": 3.18, "learning_rate": 6.206853885039264e-06, "loss": 0.082, "step": 5999 }, { "epoch": 3.18, "learning_rate": 6.2036810763836474e-06, "loss": 0.0745, "step": 6000 }, { "epoch": 3.18, "learning_rate": 6.200508714181977e-06, "loss": 0.1123, "step": 6001 }, { "epoch": 3.18, "learning_rate": 6.1973367988073255e-06, "loss": 0.0931, "step": 6002 }, { "epoch": 3.18, "learning_rate": 6.194165330632718e-06, "loss": 0.0814, "step": 6003 }, { "epoch": 3.18, "learning_rate": 6.190994310031125e-06, "loss": 0.0606, "step": 6004 }, { "epoch": 3.18, "learning_rate": 6.187823737375462e-06, "loss": 0.1114, "step": 6005 }, { "epoch": 3.18, "learning_rate": 6.184653613038598e-06, "loss": 0.1085, "step": 6006 }, { "epoch": 3.18, "learning_rate": 6.18148393739334e-06, "loss": 0.1154, "step": 6007 }, { "epoch": 3.18, "learning_rate": 6.178314710812452e-06, "loss": 0.141, "step": 6008 }, { "epoch": 3.18, "learning_rate": 6.1751459336686385e-06, "loss": 0.1051, "step": 6009 }, { "epoch": 3.18, "learning_rate": 6.171977606334552e-06, "loss": 0.0782, "step": 6010 }, { "epoch": 3.18, "learning_rate": 6.168809729182795e-06, "loss": 0.1178, "step": 6011 }, { "epoch": 3.18, "learning_rate": 6.165642302585918e-06, "loss": 0.0722, "step": 6012 }, { "epoch": 3.18, "learning_rate": 6.162475326916415e-06, "loss": 0.0923, "step": 6013 }, { "epoch": 3.18, "learning_rate": 6.159308802546729e-06, "loss": 0.0975, "step": 6014 }, { "epoch": 3.18, "learning_rate": 6.1561427298492426e-06, "loss": 0.1173, "step": 6015 }, { "epoch": 3.18, "learning_rate": 6.152977109196297e-06, "loss": 0.0804, "step": 6016 }, { "epoch": 3.18, "learning_rate": 6.149811940960175e-06, "loss": 0.0815, "step": 6017 }, { "epoch": 3.19, "learning_rate": 6.146647225513105e-06, "loss": 0.117, "step": 6018 }, { "epoch": 3.19, "learning_rate": 6.143482963227263e-06, "loss": 0.1167, "step": 6019 }, { "epoch": 3.19, "learning_rate": 6.140319154474773e-06, "loss": 0.0896, "step": 6020 }, { "epoch": 3.19, "learning_rate": 6.137155799627703e-06, "loss": 0.0814, "step": 6021 }, { "epoch": 3.19, "learning_rate": 6.133992899058073e-06, "loss": 0.0867, "step": 6022 }, { "epoch": 3.19, "learning_rate": 6.130830453137841e-06, "loss": 0.0625, "step": 6023 }, { "epoch": 3.19, "learning_rate": 6.127668462238918e-06, "loss": 0.0676, "step": 6024 }, { "epoch": 3.19, "learning_rate": 6.12450692673316e-06, "loss": 0.0977, "step": 6025 }, { "epoch": 3.19, "learning_rate": 6.121345846992367e-06, "loss": 0.0933, "step": 6026 }, { "epoch": 3.19, "learning_rate": 6.118185223388292e-06, "loss": 0.0987, "step": 6027 }, { "epoch": 3.19, "learning_rate": 6.11502505629263e-06, "loss": 0.0853, "step": 6028 }, { "epoch": 3.19, "learning_rate": 6.1118653460770175e-06, "loss": 0.0662, "step": 6029 }, { "epoch": 3.19, "learning_rate": 6.1087060931130506e-06, "loss": 0.0898, "step": 6030 }, { "epoch": 3.19, "learning_rate": 6.105547297772252e-06, "loss": 0.0811, "step": 6031 }, { "epoch": 3.19, "learning_rate": 6.102388960426109e-06, "loss": 0.0738, "step": 6032 }, { "epoch": 3.19, "learning_rate": 6.099231081446045e-06, "loss": 0.0714, "step": 6033 }, { "epoch": 3.19, "learning_rate": 6.0960736612034345e-06, "loss": 0.0764, "step": 6034 }, { "epoch": 3.19, "learning_rate": 6.0929167000695945e-06, "loss": 0.0907, "step": 6035 }, { "epoch": 3.19, "learning_rate": 6.0897601984157885e-06, "loss": 0.1167, "step": 6036 }, { "epoch": 3.2, "learning_rate": 6.0866041566132295e-06, "loss": 0.065, "step": 6037 }, { "epoch": 3.2, "learning_rate": 6.083448575033073e-06, "loss": 0.095, "step": 6038 }, { "epoch": 3.2, "learning_rate": 6.080293454046418e-06, "loss": 0.1508, "step": 6039 }, { "epoch": 3.2, "learning_rate": 6.077138794024313e-06, "loss": 0.0703, "step": 6040 }, { "epoch": 3.2, "learning_rate": 6.073984595337755e-06, "loss": 0.0615, "step": 6041 }, { "epoch": 3.2, "learning_rate": 6.0708308583576794e-06, "loss": 0.0976, "step": 6042 }, { "epoch": 3.2, "learning_rate": 6.067677583454975e-06, "loss": 0.0603, "step": 6043 }, { "epoch": 3.2, "learning_rate": 6.06452477100047e-06, "loss": 0.0773, "step": 6044 }, { "epoch": 3.2, "learning_rate": 6.061372421364944e-06, "loss": 0.0741, "step": 6045 }, { "epoch": 3.2, "learning_rate": 6.058220534919119e-06, "loss": 0.1365, "step": 6046 }, { "epoch": 3.2, "learning_rate": 6.055069112033658e-06, "loss": 0.1309, "step": 6047 }, { "epoch": 3.2, "learning_rate": 6.051918153079176e-06, "loss": 0.096, "step": 6048 }, { "epoch": 3.2, "learning_rate": 6.048767658426232e-06, "loss": 0.091, "step": 6049 }, { "epoch": 3.2, "learning_rate": 6.04561762844533e-06, "loss": 0.0885, "step": 6050 }, { "epoch": 3.2, "learning_rate": 6.042468063506919e-06, "loss": 0.1153, "step": 6051 }, { "epoch": 3.2, "learning_rate": 6.039318963981393e-06, "loss": 0.1024, "step": 6052 }, { "epoch": 3.2, "learning_rate": 6.036170330239091e-06, "loss": 0.125, "step": 6053 }, { "epoch": 3.2, "learning_rate": 6.033022162650305e-06, "loss": 0.1723, "step": 6054 }, { "epoch": 3.2, "learning_rate": 6.029874461585256e-06, "loss": 0.1591, "step": 6055 }, { "epoch": 3.21, "learning_rate": 6.026727227414122e-06, "loss": 0.1017, "step": 6056 }, { "epoch": 3.21, "learning_rate": 6.023580460507022e-06, "loss": 0.1023, "step": 6057 }, { "epoch": 3.21, "learning_rate": 6.020434161234026e-06, "loss": 0.0947, "step": 6058 }, { "epoch": 3.21, "learning_rate": 6.017288329965143e-06, "loss": 0.0873, "step": 6059 }, { "epoch": 3.21, "learning_rate": 6.014142967070327e-06, "loss": 0.0817, "step": 6060 }, { "epoch": 3.21, "learning_rate": 6.010998072919478e-06, "loss": 0.0889, "step": 6061 }, { "epoch": 3.21, "learning_rate": 6.007853647882442e-06, "loss": 0.0867, "step": 6062 }, { "epoch": 3.21, "learning_rate": 6.004709692329012e-06, "loss": 0.092, "step": 6063 }, { "epoch": 3.21, "learning_rate": 6.0015662066289184e-06, "loss": 0.0706, "step": 6064 }, { "epoch": 3.21, "learning_rate": 5.998423191151841e-06, "loss": 0.0713, "step": 6065 }, { "epoch": 3.21, "learning_rate": 5.995280646267406e-06, "loss": 0.0867, "step": 6066 }, { "epoch": 3.21, "learning_rate": 5.992138572345183e-06, "loss": 0.1142, "step": 6067 }, { "epoch": 3.21, "learning_rate": 5.9889969697546836e-06, "loss": 0.1125, "step": 6068 }, { "epoch": 3.21, "learning_rate": 5.985855838865363e-06, "loss": 0.0613, "step": 6069 }, { "epoch": 3.21, "learning_rate": 5.982715180046631e-06, "loss": 0.1055, "step": 6070 }, { "epoch": 3.21, "learning_rate": 5.979574993667836e-06, "loss": 0.0897, "step": 6071 }, { "epoch": 3.21, "learning_rate": 5.976435280098259e-06, "loss": 0.1234, "step": 6072 }, { "epoch": 3.21, "learning_rate": 5.973296039707142e-06, "loss": 0.0889, "step": 6073 }, { "epoch": 3.22, "learning_rate": 5.9701572728636645e-06, "loss": 0.0694, "step": 6074 }, { "epoch": 3.22, "learning_rate": 5.967018979936954e-06, "loss": 0.0627, "step": 6075 }, { "epoch": 3.22, "learning_rate": 5.963881161296076e-06, "loss": 0.1076, "step": 6076 }, { "epoch": 3.22, "learning_rate": 5.960743817310044e-06, "loss": 0.0801, "step": 6077 }, { "epoch": 3.22, "learning_rate": 5.9576069483478186e-06, "loss": 0.0966, "step": 6078 }, { "epoch": 3.22, "learning_rate": 5.954470554778298e-06, "loss": 0.115, "step": 6079 }, { "epoch": 3.22, "learning_rate": 5.951334636970327e-06, "loss": 0.0568, "step": 6080 }, { "epoch": 3.22, "learning_rate": 5.948199195292699e-06, "loss": 0.1303, "step": 6081 }, { "epoch": 3.22, "learning_rate": 5.945064230114143e-06, "loss": 0.0905, "step": 6082 }, { "epoch": 3.22, "learning_rate": 5.941929741803339e-06, "loss": 0.1387, "step": 6083 }, { "epoch": 3.22, "learning_rate": 5.938795730728907e-06, "loss": 0.1134, "step": 6084 }, { "epoch": 3.22, "learning_rate": 5.9356621972594155e-06, "loss": 0.1107, "step": 6085 }, { "epoch": 3.22, "learning_rate": 5.93252914176337e-06, "loss": 0.1135, "step": 6086 }, { "epoch": 3.22, "learning_rate": 5.929396564609231e-06, "loss": 0.0952, "step": 6087 }, { "epoch": 3.22, "learning_rate": 5.926264466165383e-06, "loss": 0.067, "step": 6088 }, { "epoch": 3.22, "learning_rate": 5.923132846800176e-06, "loss": 0.0762, "step": 6089 }, { "epoch": 3.22, "learning_rate": 5.92000170688189e-06, "loss": 0.0802, "step": 6090 }, { "epoch": 3.22, "learning_rate": 5.916871046778754e-06, "loss": 0.1185, "step": 6091 }, { "epoch": 3.22, "learning_rate": 5.913740866858939e-06, "loss": 0.1051, "step": 6092 }, { "epoch": 3.23, "learning_rate": 5.9106111674905585e-06, "loss": 0.1639, "step": 6093 }, { "epoch": 3.23, "learning_rate": 5.9074819490416734e-06, "loss": 0.055, "step": 6094 }, { "epoch": 3.23, "learning_rate": 5.9043532118802845e-06, "loss": 0.0896, "step": 6095 }, { "epoch": 3.23, "learning_rate": 5.901224956374336e-06, "loss": 0.0862, "step": 6096 }, { "epoch": 3.23, "learning_rate": 5.898097182891718e-06, "loss": 0.119, "step": 6097 }, { "epoch": 3.23, "learning_rate": 5.8949698918002586e-06, "loss": 0.084, "step": 6098 }, { "epoch": 3.23, "learning_rate": 5.891843083467734e-06, "loss": 0.0706, "step": 6099 }, { "epoch": 3.23, "learning_rate": 5.888716758261868e-06, "loss": 0.0977, "step": 6100 }, { "epoch": 3.23, "learning_rate": 5.8855909165503145e-06, "loss": 0.0909, "step": 6101 }, { "epoch": 3.23, "learning_rate": 5.882465558700684e-06, "loss": 0.1058, "step": 6102 }, { "epoch": 3.23, "learning_rate": 5.879340685080526e-06, "loss": 0.0814, "step": 6103 }, { "epoch": 3.23, "learning_rate": 5.87621629605732e-06, "loss": 0.1117, "step": 6104 }, { "epoch": 3.23, "learning_rate": 5.873092391998509e-06, "loss": 0.0676, "step": 6105 }, { "epoch": 3.23, "learning_rate": 5.869968973271467e-06, "loss": 0.0816, "step": 6106 }, { "epoch": 3.23, "learning_rate": 5.866846040243516e-06, "loss": 0.0889, "step": 6107 }, { "epoch": 3.23, "learning_rate": 5.863723593281916e-06, "loss": 0.1082, "step": 6108 }, { "epoch": 3.23, "learning_rate": 5.860601632753873e-06, "loss": 0.1126, "step": 6109 }, { "epoch": 3.23, "learning_rate": 5.857480159026536e-06, "loss": 0.0537, "step": 6110 }, { "epoch": 3.23, "learning_rate": 5.854359172466997e-06, "loss": 0.0932, "step": 6111 }, { "epoch": 3.24, "learning_rate": 5.851238673442286e-06, "loss": 0.0649, "step": 6112 }, { "epoch": 3.24, "learning_rate": 5.84811866231938e-06, "loss": 0.0805, "step": 6113 }, { "epoch": 3.24, "learning_rate": 5.844999139465201e-06, "loss": 0.0735, "step": 6114 }, { "epoch": 3.24, "learning_rate": 5.841880105246605e-06, "loss": 0.1022, "step": 6115 }, { "epoch": 3.24, "learning_rate": 5.838761560030404e-06, "loss": 0.0718, "step": 6116 }, { "epoch": 3.24, "learning_rate": 5.835643504183341e-06, "loss": 0.0694, "step": 6117 }, { "epoch": 3.24, "learning_rate": 5.8325259380721e-06, "loss": 0.1096, "step": 6118 }, { "epoch": 3.24, "learning_rate": 5.8294088620633214e-06, "loss": 0.1134, "step": 6119 }, { "epoch": 3.24, "learning_rate": 5.826292276523574e-06, "loss": 0.0808, "step": 6120 }, { "epoch": 3.24, "learning_rate": 5.82317618181937e-06, "loss": 0.0986, "step": 6121 }, { "epoch": 3.24, "learning_rate": 5.820060578317176e-06, "loss": 0.0924, "step": 6122 }, { "epoch": 3.24, "learning_rate": 5.816945466383384e-06, "loss": 0.1098, "step": 6123 }, { "epoch": 3.24, "learning_rate": 5.813830846384342e-06, "loss": 0.0949, "step": 6124 }, { "epoch": 3.24, "learning_rate": 5.810716718686338e-06, "loss": 0.1203, "step": 6125 }, { "epoch": 3.24, "learning_rate": 5.807603083655591e-06, "loss": 0.0415, "step": 6126 }, { "epoch": 3.24, "learning_rate": 5.8044899416582775e-06, "loss": 0.1071, "step": 6127 }, { "epoch": 3.24, "learning_rate": 5.801377293060506e-06, "loss": 0.0693, "step": 6128 }, { "epoch": 3.24, "learning_rate": 5.798265138228325e-06, "loss": 0.1037, "step": 6129 }, { "epoch": 3.24, "learning_rate": 5.795153477527738e-06, "loss": 0.1111, "step": 6130 }, { "epoch": 3.25, "learning_rate": 5.7920423113246724e-06, "loss": 0.0882, "step": 6131 }, { "epoch": 3.25, "learning_rate": 5.788931639985016e-06, "loss": 0.0576, "step": 6132 }, { "epoch": 3.25, "learning_rate": 5.785821463874581e-06, "loss": 0.1212, "step": 6133 }, { "epoch": 3.25, "learning_rate": 5.782711783359134e-06, "loss": 0.1287, "step": 6134 }, { "epoch": 3.25, "learning_rate": 5.779602598804388e-06, "loss": 0.1245, "step": 6135 }, { "epoch": 3.25, "learning_rate": 5.776493910575971e-06, "loss": 0.0685, "step": 6136 }, { "epoch": 3.25, "learning_rate": 5.773385719039484e-06, "loss": 0.0726, "step": 6137 }, { "epoch": 3.25, "learning_rate": 5.7702780245604465e-06, "loss": 0.0766, "step": 6138 }, { "epoch": 3.25, "learning_rate": 5.767170827504333e-06, "loss": 0.0661, "step": 6139 }, { "epoch": 3.25, "learning_rate": 5.764064128236559e-06, "loss": 0.0675, "step": 6140 }, { "epoch": 3.25, "learning_rate": 5.7609579271224724e-06, "loss": 0.1035, "step": 6141 }, { "epoch": 3.25, "learning_rate": 5.757852224527375e-06, "loss": 0.0859, "step": 6142 }, { "epoch": 3.25, "learning_rate": 5.754747020816498e-06, "loss": 0.1191, "step": 6143 }, { "epoch": 3.25, "learning_rate": 5.751642316355015e-06, "loss": 0.0959, "step": 6144 }, { "epoch": 3.25, "learning_rate": 5.748538111508053e-06, "loss": 0.1186, "step": 6145 }, { "epoch": 3.25, "learning_rate": 5.745434406640664e-06, "loss": 0.0803, "step": 6146 }, { "epoch": 3.25, "learning_rate": 5.742331202117859e-06, "loss": 0.1414, "step": 6147 }, { "epoch": 3.25, "learning_rate": 5.739228498304571e-06, "loss": 0.0558, "step": 6148 }, { "epoch": 3.25, "learning_rate": 5.736126295565687e-06, "loss": 0.1656, "step": 6149 }, { "epoch": 3.26, "learning_rate": 5.733024594266037e-06, "loss": 0.0763, "step": 6150 }, { "epoch": 3.26, "learning_rate": 5.7299233947703815e-06, "loss": 0.0832, "step": 6151 }, { "epoch": 3.26, "learning_rate": 5.726822697443427e-06, "loss": 0.0859, "step": 6152 }, { "epoch": 3.26, "learning_rate": 5.723722502649819e-06, "loss": 0.1087, "step": 6153 }, { "epoch": 3.26, "learning_rate": 5.7206228107541484e-06, "loss": 0.0844, "step": 6154 }, { "epoch": 3.26, "learning_rate": 5.717523622120948e-06, "loss": 0.1745, "step": 6155 }, { "epoch": 3.26, "learning_rate": 5.714424937114681e-06, "loss": 0.1174, "step": 6156 }, { "epoch": 3.26, "learning_rate": 5.711326756099767e-06, "loss": 0.1686, "step": 6157 }, { "epoch": 3.26, "learning_rate": 5.7082290794405486e-06, "loss": 0.0527, "step": 6158 }, { "epoch": 3.26, "learning_rate": 5.705131907501325e-06, "loss": 0.1392, "step": 6159 }, { "epoch": 3.26, "learning_rate": 5.702035240646328e-06, "loss": 0.0788, "step": 6160 }, { "epoch": 3.26, "learning_rate": 5.6989390792397245e-06, "loss": 0.0616, "step": 6161 }, { "epoch": 3.26, "learning_rate": 5.695843423645639e-06, "loss": 0.0625, "step": 6162 }, { "epoch": 3.26, "learning_rate": 5.692748274228117e-06, "loss": 0.0877, "step": 6163 }, { "epoch": 3.26, "learning_rate": 5.689653631351162e-06, "loss": 0.0824, "step": 6164 }, { "epoch": 3.26, "learning_rate": 5.686559495378702e-06, "loss": 0.0709, "step": 6165 }, { "epoch": 3.26, "learning_rate": 5.6834658666746155e-06, "loss": 0.0919, "step": 6166 }, { "epoch": 3.26, "learning_rate": 5.680372745602727e-06, "loss": 0.1063, "step": 6167 }, { "epoch": 3.26, "learning_rate": 5.677280132526779e-06, "loss": 0.1021, "step": 6168 }, { "epoch": 3.27, "learning_rate": 5.674188027810475e-06, "loss": 0.0985, "step": 6169 }, { "epoch": 3.27, "learning_rate": 5.671096431817455e-06, "loss": 0.1043, "step": 6170 }, { "epoch": 3.27, "learning_rate": 5.668005344911289e-06, "loss": 0.0878, "step": 6171 }, { "epoch": 3.27, "learning_rate": 5.6649147674555035e-06, "loss": 0.0751, "step": 6172 }, { "epoch": 3.27, "learning_rate": 5.6618246998135485e-06, "loss": 0.0932, "step": 6173 }, { "epoch": 3.27, "learning_rate": 5.658735142348827e-06, "loss": 0.0872, "step": 6174 }, { "epoch": 3.27, "learning_rate": 5.655646095424674e-06, "loss": 0.0666, "step": 6175 }, { "epoch": 3.27, "learning_rate": 5.652557559404364e-06, "loss": 0.0633, "step": 6176 }, { "epoch": 3.27, "learning_rate": 5.64946953465112e-06, "loss": 0.0869, "step": 6177 }, { "epoch": 3.27, "learning_rate": 5.646382021528093e-06, "loss": 0.0989, "step": 6178 }, { "epoch": 3.27, "learning_rate": 5.6432950203983875e-06, "loss": 0.0896, "step": 6179 }, { "epoch": 3.27, "learning_rate": 5.640208531625032e-06, "loss": 0.0853, "step": 6180 }, { "epoch": 3.27, "learning_rate": 5.637122555571008e-06, "loss": 0.1229, "step": 6181 }, { "epoch": 3.27, "learning_rate": 5.634037092599236e-06, "loss": 0.065, "step": 6182 }, { "epoch": 3.27, "learning_rate": 5.630952143072563e-06, "loss": 0.0792, "step": 6183 }, { "epoch": 3.27, "learning_rate": 5.627867707353794e-06, "loss": 0.083, "step": 6184 }, { "epoch": 3.27, "learning_rate": 5.624783785805658e-06, "loss": 0.0677, "step": 6185 }, { "epoch": 3.27, "learning_rate": 5.621700378790827e-06, "loss": 0.0838, "step": 6186 }, { "epoch": 3.27, "learning_rate": 5.618617486671924e-06, "loss": 0.0654, "step": 6187 }, { "epoch": 3.28, "learning_rate": 5.615535109811494e-06, "loss": 0.0617, "step": 6188 }, { "epoch": 3.28, "learning_rate": 5.612453248572038e-06, "loss": 0.0889, "step": 6189 }, { "epoch": 3.28, "learning_rate": 5.60937190331598e-06, "loss": 0.1181, "step": 6190 }, { "epoch": 3.28, "learning_rate": 5.606291074405701e-06, "loss": 0.1591, "step": 6191 }, { "epoch": 3.28, "learning_rate": 5.603210762203506e-06, "loss": 0.0923, "step": 6192 }, { "epoch": 3.28, "learning_rate": 5.600130967071643e-06, "loss": 0.0823, "step": 6193 }, { "epoch": 3.28, "learning_rate": 5.59705168937231e-06, "loss": 0.0557, "step": 6194 }, { "epoch": 3.28, "learning_rate": 5.5939729294676245e-06, "loss": 0.1099, "step": 6195 }, { "epoch": 3.28, "learning_rate": 5.590894687719661e-06, "loss": 0.0827, "step": 6196 }, { "epoch": 3.28, "learning_rate": 5.5878169644904315e-06, "loss": 0.0624, "step": 6197 }, { "epoch": 3.28, "learning_rate": 5.584739760141868e-06, "loss": 0.1237, "step": 6198 }, { "epoch": 3.28, "learning_rate": 5.58166307503587e-06, "loss": 0.0431, "step": 6199 }, { "epoch": 3.28, "learning_rate": 5.578586909534253e-06, "loss": 0.0725, "step": 6200 }, { "epoch": 3.28, "learning_rate": 5.575511263998778e-06, "loss": 0.1196, "step": 6201 }, { "epoch": 3.28, "learning_rate": 5.572436138791153e-06, "loss": 0.0763, "step": 6202 }, { "epoch": 3.28, "learning_rate": 5.5693615342730114e-06, "loss": 0.0698, "step": 6203 }, { "epoch": 3.28, "learning_rate": 5.566287450805939e-06, "loss": 0.0509, "step": 6204 }, { "epoch": 3.28, "learning_rate": 5.563213888751445e-06, "loss": 0.0738, "step": 6205 }, { "epoch": 3.28, "learning_rate": 5.560140848470996e-06, "loss": 0.128, "step": 6206 }, { "epoch": 3.29, "learning_rate": 5.557068330325976e-06, "loss": 0.1236, "step": 6207 }, { "epoch": 3.29, "learning_rate": 5.5539963346777315e-06, "loss": 0.1142, "step": 6208 }, { "epoch": 3.29, "learning_rate": 5.550924861887525e-06, "loss": 0.0892, "step": 6209 }, { "epoch": 3.29, "learning_rate": 5.547853912316569e-06, "loss": 0.1088, "step": 6210 }, { "epoch": 3.29, "learning_rate": 5.544783486326012e-06, "loss": 0.0631, "step": 6211 }, { "epoch": 3.29, "learning_rate": 5.5417135842769475e-06, "loss": 0.1318, "step": 6212 }, { "epoch": 3.29, "learning_rate": 5.538644206530394e-06, "loss": 0.0736, "step": 6213 }, { "epoch": 3.29, "learning_rate": 5.535575353447322e-06, "loss": 0.0827, "step": 6214 }, { "epoch": 3.29, "learning_rate": 5.532507025388628e-06, "loss": 0.113, "step": 6215 }, { "epoch": 3.29, "learning_rate": 5.52943922271516e-06, "loss": 0.0816, "step": 6216 }, { "epoch": 3.29, "learning_rate": 5.526371945787694e-06, "loss": 0.0627, "step": 6217 }, { "epoch": 3.29, "learning_rate": 5.523305194966943e-06, "loss": 0.0999, "step": 6218 }, { "epoch": 3.29, "learning_rate": 5.520238970613569e-06, "loss": 0.1128, "step": 6219 }, { "epoch": 3.29, "learning_rate": 5.517173273088161e-06, "loss": 0.1247, "step": 6220 }, { "epoch": 3.29, "learning_rate": 5.514108102751254e-06, "loss": 0.0765, "step": 6221 }, { "epoch": 3.29, "learning_rate": 5.511043459963312e-06, "loss": 0.1366, "step": 6222 }, { "epoch": 3.29, "learning_rate": 5.507979345084748e-06, "loss": 0.0833, "step": 6223 }, { "epoch": 3.29, "learning_rate": 5.504915758475912e-06, "loss": 0.109, "step": 6224 }, { "epoch": 3.29, "learning_rate": 5.501852700497073e-06, "loss": 0.0918, "step": 6225 }, { "epoch": 3.3, "learning_rate": 5.498790171508464e-06, "loss": 0.0754, "step": 6226 }, { "epoch": 3.3, "learning_rate": 5.495728171870238e-06, "loss": 0.0898, "step": 6227 }, { "epoch": 3.3, "learning_rate": 5.492666701942492e-06, "loss": 0.1318, "step": 6228 }, { "epoch": 3.3, "learning_rate": 5.489605762085266e-06, "loss": 0.1123, "step": 6229 }, { "epoch": 3.3, "learning_rate": 5.486545352658526e-06, "loss": 0.1085, "step": 6230 }, { "epoch": 3.3, "learning_rate": 5.483485474022185e-06, "loss": 0.1266, "step": 6231 }, { "epoch": 3.3, "learning_rate": 5.480426126536092e-06, "loss": 0.0861, "step": 6232 }, { "epoch": 3.3, "learning_rate": 5.477367310560022e-06, "loss": 0.1003, "step": 6233 }, { "epoch": 3.3, "learning_rate": 5.47430902645371e-06, "loss": 0.0654, "step": 6234 }, { "epoch": 3.3, "learning_rate": 5.471251274576805e-06, "loss": 0.0439, "step": 6235 }, { "epoch": 3.3, "learning_rate": 5.468194055288913e-06, "loss": 0.0702, "step": 6236 }, { "epoch": 3.3, "learning_rate": 5.4651373689495625e-06, "loss": 0.0611, "step": 6237 }, { "epoch": 3.3, "learning_rate": 5.462081215918224e-06, "loss": 0.077, "step": 6238 }, { "epoch": 3.3, "learning_rate": 5.459025596554316e-06, "loss": 0.0813, "step": 6239 }, { "epoch": 3.3, "learning_rate": 5.455970511217179e-06, "loss": 0.1008, "step": 6240 }, { "epoch": 3.3, "learning_rate": 5.452915960266094e-06, "loss": 0.1034, "step": 6241 }, { "epoch": 3.3, "learning_rate": 5.449861944060282e-06, "loss": 0.1328, "step": 6242 }, { "epoch": 3.3, "learning_rate": 5.446808462958903e-06, "loss": 0.0848, "step": 6243 }, { "epoch": 3.31, "learning_rate": 5.443755517321054e-06, "loss": 0.0472, "step": 6244 }, { "epoch": 3.31, "learning_rate": 5.4407031075057615e-06, "loss": 0.0918, "step": 6245 }, { "epoch": 3.31, "learning_rate": 5.437651233872001e-06, "loss": 0.1148, "step": 6246 }, { "epoch": 3.31, "learning_rate": 5.4345998967786715e-06, "loss": 0.0829, "step": 6247 }, { "epoch": 3.31, "learning_rate": 5.431549096584623e-06, "loss": 0.101, "step": 6248 }, { "epoch": 3.31, "learning_rate": 5.42849883364863e-06, "loss": 0.0689, "step": 6249 }, { "epoch": 3.31, "learning_rate": 5.425449108329407e-06, "loss": 0.0842, "step": 6250 }, { "epoch": 3.31, "learning_rate": 5.422399920985615e-06, "loss": 0.0976, "step": 6251 }, { "epoch": 3.31, "learning_rate": 5.419351271975834e-06, "loss": 0.1028, "step": 6252 }, { "epoch": 3.31, "learning_rate": 5.416303161658596e-06, "loss": 0.0741, "step": 6253 }, { "epoch": 3.31, "learning_rate": 5.413255590392368e-06, "loss": 0.0848, "step": 6254 }, { "epoch": 3.31, "learning_rate": 5.410208558535541e-06, "loss": 0.1139, "step": 6255 }, { "epoch": 3.31, "learning_rate": 5.407162066446465e-06, "loss": 0.0724, "step": 6256 }, { "epoch": 3.31, "learning_rate": 5.404116114483398e-06, "loss": 0.0601, "step": 6257 }, { "epoch": 3.31, "learning_rate": 5.401070703004554e-06, "loss": 0.0566, "step": 6258 }, { "epoch": 3.31, "learning_rate": 5.398025832368084e-06, "loss": 0.1181, "step": 6259 }, { "epoch": 3.31, "learning_rate": 5.394981502932064e-06, "loss": 0.103, "step": 6260 }, { "epoch": 3.31, "learning_rate": 5.391937715054519e-06, "loss": 0.0577, "step": 6261 }, { "epoch": 3.31, "learning_rate": 5.388894469093398e-06, "loss": 0.0649, "step": 6262 }, { "epoch": 3.32, "learning_rate": 5.385851765406596e-06, "loss": 0.0929, "step": 6263 }, { "epoch": 3.32, "learning_rate": 5.3828096043519406e-06, "loss": 0.0693, "step": 6264 }, { "epoch": 3.32, "learning_rate": 5.37976798628719e-06, "loss": 0.0837, "step": 6265 }, { "epoch": 3.32, "learning_rate": 5.376726911570052e-06, "loss": 0.0999, "step": 6266 }, { "epoch": 3.32, "learning_rate": 5.373686380558154e-06, "loss": 0.0952, "step": 6267 }, { "epoch": 3.32, "learning_rate": 5.370646393609076e-06, "loss": 0.159, "step": 6268 }, { "epoch": 3.32, "learning_rate": 5.367606951080319e-06, "loss": 0.1037, "step": 6269 }, { "epoch": 3.32, "learning_rate": 5.3645680533293284e-06, "loss": 0.1266, "step": 6270 }, { "epoch": 3.32, "learning_rate": 5.361529700713492e-06, "loss": 0.1077, "step": 6271 }, { "epoch": 3.32, "learning_rate": 5.358491893590118e-06, "loss": 0.0965, "step": 6272 }, { "epoch": 3.32, "learning_rate": 5.355454632316457e-06, "loss": 0.1118, "step": 6273 }, { "epoch": 3.32, "learning_rate": 5.3524179172497005e-06, "loss": 0.1255, "step": 6274 }, { "epoch": 3.32, "learning_rate": 5.3493817487469664e-06, "loss": 0.1072, "step": 6275 }, { "epoch": 3.32, "learning_rate": 5.3463461271653214e-06, "loss": 0.1406, "step": 6276 }, { "epoch": 3.32, "learning_rate": 5.343311052861753e-06, "loss": 0.0868, "step": 6277 }, { "epoch": 3.32, "learning_rate": 5.340276526193196e-06, "loss": 0.1013, "step": 6278 }, { "epoch": 3.32, "learning_rate": 5.337242547516511e-06, "loss": 0.0809, "step": 6279 }, { "epoch": 3.32, "learning_rate": 5.334209117188507e-06, "loss": 0.0947, "step": 6280 }, { "epoch": 3.32, "learning_rate": 5.331176235565917e-06, "loss": 0.0767, "step": 6281 }, { "epoch": 3.33, "learning_rate": 5.328143903005409e-06, "loss": 0.0905, "step": 6282 }, { "epoch": 3.33, "learning_rate": 5.325112119863597e-06, "loss": 0.0652, "step": 6283 }, { "epoch": 3.33, "learning_rate": 5.322080886497018e-06, "loss": 0.0831, "step": 6284 }, { "epoch": 3.33, "learning_rate": 5.319050203262157e-06, "loss": 0.0773, "step": 6285 }, { "epoch": 3.33, "learning_rate": 5.316020070515426e-06, "loss": 0.1011, "step": 6286 }, { "epoch": 3.33, "learning_rate": 5.312990488613172e-06, "loss": 0.0859, "step": 6287 }, { "epoch": 3.33, "learning_rate": 5.309961457911683e-06, "loss": 0.0659, "step": 6288 }, { "epoch": 3.33, "learning_rate": 5.306932978767176e-06, "loss": 0.1091, "step": 6289 }, { "epoch": 3.33, "learning_rate": 5.303905051535802e-06, "loss": 0.1168, "step": 6290 }, { "epoch": 3.33, "learning_rate": 5.300877676573657e-06, "loss": 0.0968, "step": 6291 }, { "epoch": 3.33, "learning_rate": 5.2978508542367605e-06, "loss": 0.0811, "step": 6292 }, { "epoch": 3.33, "learning_rate": 5.294824584881078e-06, "loss": 0.1525, "step": 6293 }, { "epoch": 3.33, "learning_rate": 5.291798868862498e-06, "loss": 0.0818, "step": 6294 }, { "epoch": 3.33, "learning_rate": 5.288773706536851e-06, "loss": 0.089, "step": 6295 }, { "epoch": 3.33, "learning_rate": 5.285749098259914e-06, "loss": 0.0578, "step": 6296 }, { "epoch": 3.33, "learning_rate": 5.282725044387367e-06, "loss": 0.0824, "step": 6297 }, { "epoch": 3.33, "learning_rate": 5.2797015452748576e-06, "loss": 0.0889, "step": 6298 }, { "epoch": 3.33, "learning_rate": 5.276678601277947e-06, "loss": 0.0823, "step": 6299 }, { "epoch": 3.33, "learning_rate": 5.2736562127521415e-06, "loss": 0.1489, "step": 6300 }, { "epoch": 3.34, "learning_rate": 5.270634380052883e-06, "loss": 0.046, "step": 6301 }, { "epoch": 3.34, "learning_rate": 5.267613103535541e-06, "loss": 0.1371, "step": 6302 }, { "epoch": 3.34, "learning_rate": 5.2645923835554245e-06, "loss": 0.0647, "step": 6303 }, { "epoch": 3.34, "learning_rate": 5.261572220467777e-06, "loss": 0.1095, "step": 6304 }, { "epoch": 3.34, "learning_rate": 5.258552614627768e-06, "loss": 0.0727, "step": 6305 }, { "epoch": 3.34, "learning_rate": 5.2555335663905185e-06, "loss": 0.096, "step": 6306 }, { "epoch": 3.34, "learning_rate": 5.252515076111064e-06, "loss": 0.1275, "step": 6307 }, { "epoch": 3.34, "learning_rate": 5.249497144144393e-06, "loss": 0.0988, "step": 6308 }, { "epoch": 3.34, "learning_rate": 5.246479770845414e-06, "loss": 0.2385, "step": 6309 }, { "epoch": 3.34, "learning_rate": 5.2434629565689806e-06, "loss": 0.1444, "step": 6310 }, { "epoch": 3.34, "learning_rate": 5.240446701669868e-06, "loss": 0.0959, "step": 6311 }, { "epoch": 3.34, "learning_rate": 5.2374310065028e-06, "loss": 0.0723, "step": 6312 }, { "epoch": 3.34, "learning_rate": 5.234415871422431e-06, "loss": 0.0961, "step": 6313 }, { "epoch": 3.34, "learning_rate": 5.231401296783336e-06, "loss": 0.0641, "step": 6314 }, { "epoch": 3.34, "learning_rate": 5.228387282940037e-06, "loss": 0.0874, "step": 6315 }, { "epoch": 3.34, "learning_rate": 5.225373830246996e-06, "loss": 0.104, "step": 6316 }, { "epoch": 3.34, "learning_rate": 5.2223609390585885e-06, "loss": 0.0622, "step": 6317 }, { "epoch": 3.34, "learning_rate": 5.219348609729147e-06, "loss": 0.1324, "step": 6318 }, { "epoch": 3.34, "learning_rate": 5.216336842612917e-06, "loss": 0.0961, "step": 6319 }, { "epoch": 3.35, "learning_rate": 5.213325638064097e-06, "loss": 0.1562, "step": 6320 }, { "epoch": 3.35, "learning_rate": 5.210314996436805e-06, "loss": 0.0953, "step": 6321 }, { "epoch": 3.35, "learning_rate": 5.207304918085094e-06, "loss": 0.1019, "step": 6322 }, { "epoch": 3.35, "learning_rate": 5.2042954033629645e-06, "loss": 0.0794, "step": 6323 }, { "epoch": 3.35, "learning_rate": 5.20128645262433e-06, "loss": 0.1105, "step": 6324 }, { "epoch": 3.35, "learning_rate": 5.198278066223059e-06, "loss": 0.0753, "step": 6325 }, { "epoch": 3.35, "learning_rate": 5.195270244512934e-06, "loss": 0.1497, "step": 6326 }, { "epoch": 3.35, "learning_rate": 5.1922629878476825e-06, "loss": 0.0987, "step": 6327 }, { "epoch": 3.35, "learning_rate": 5.189256296580971e-06, "loss": 0.072, "step": 6328 }, { "epoch": 3.35, "learning_rate": 5.186250171066385e-06, "loss": 0.1306, "step": 6329 }, { "epoch": 3.35, "learning_rate": 5.183244611657447e-06, "loss": 0.0955, "step": 6330 }, { "epoch": 3.35, "learning_rate": 5.180239618707625e-06, "loss": 0.083, "step": 6331 }, { "epoch": 3.35, "learning_rate": 5.177235192570304e-06, "loss": 0.0946, "step": 6332 }, { "epoch": 3.35, "learning_rate": 5.174231333598815e-06, "loss": 0.0918, "step": 6333 }, { "epoch": 3.35, "learning_rate": 5.171228042146413e-06, "loss": 0.0478, "step": 6334 }, { "epoch": 3.35, "learning_rate": 5.168225318566294e-06, "loss": 0.0558, "step": 6335 }, { "epoch": 3.35, "learning_rate": 5.16522316321158e-06, "loss": 0.0703, "step": 6336 }, { "epoch": 3.35, "learning_rate": 5.162221576435336e-06, "loss": 0.0914, "step": 6337 }, { "epoch": 3.35, "learning_rate": 5.15922055859055e-06, "loss": 0.1564, "step": 6338 }, { "epoch": 3.36, "learning_rate": 5.1562201100301435e-06, "loss": 0.0763, "step": 6339 }, { "epoch": 3.36, "learning_rate": 5.153220231106982e-06, "loss": 0.0809, "step": 6340 }, { "epoch": 3.36, "learning_rate": 5.15022092217385e-06, "loss": 0.106, "step": 6341 }, { "epoch": 3.36, "learning_rate": 5.147222183583474e-06, "loss": 0.0667, "step": 6342 }, { "epoch": 3.36, "learning_rate": 5.144224015688517e-06, "loss": 0.0541, "step": 6343 }, { "epoch": 3.36, "learning_rate": 5.1412264188415585e-06, "loss": 0.0975, "step": 6344 }, { "epoch": 3.36, "learning_rate": 5.1382293933951345e-06, "loss": 0.0797, "step": 6345 }, { "epoch": 3.36, "learning_rate": 5.135232939701685e-06, "loss": 0.078, "step": 6346 }, { "epoch": 3.36, "learning_rate": 5.132237058113608e-06, "loss": 0.0745, "step": 6347 }, { "epoch": 3.36, "learning_rate": 5.1292417489832225e-06, "loss": 0.1328, "step": 6348 }, { "epoch": 3.36, "learning_rate": 5.126247012662781e-06, "loss": 0.067, "step": 6349 }, { "epoch": 3.36, "learning_rate": 5.123252849504474e-06, "loss": 0.1157, "step": 6350 }, { "epoch": 3.36, "learning_rate": 5.120259259860415e-06, "loss": 0.0928, "step": 6351 }, { "epoch": 3.36, "learning_rate": 5.117266244082661e-06, "loss": 0.0699, "step": 6352 }, { "epoch": 3.36, "learning_rate": 5.114273802523195e-06, "loss": 0.0826, "step": 6353 }, { "epoch": 3.36, "learning_rate": 5.111281935533927e-06, "loss": 0.088, "step": 6354 }, { "epoch": 3.36, "learning_rate": 5.108290643466713e-06, "loss": 0.0926, "step": 6355 }, { "epoch": 3.36, "learning_rate": 5.10529992667333e-06, "loss": 0.0696, "step": 6356 }, { "epoch": 3.36, "learning_rate": 5.1023097855054945e-06, "loss": 0.0987, "step": 6357 }, { "epoch": 3.37, "learning_rate": 5.099320220314854e-06, "loss": 0.0694, "step": 6358 }, { "epoch": 3.37, "learning_rate": 5.0963312314529815e-06, "loss": 0.119, "step": 6359 }, { "epoch": 3.37, "learning_rate": 5.093342819271394e-06, "loss": 0.116, "step": 6360 }, { "epoch": 3.37, "learning_rate": 5.09035498412153e-06, "loss": 0.0882, "step": 6361 }, { "epoch": 3.37, "learning_rate": 5.087367726354762e-06, "loss": 0.1215, "step": 6362 }, { "epoch": 3.37, "learning_rate": 5.0843810463224045e-06, "loss": 0.1112, "step": 6363 }, { "epoch": 3.37, "learning_rate": 5.081394944375686e-06, "loss": 0.0923, "step": 6364 }, { "epoch": 3.37, "learning_rate": 5.078409420865788e-06, "loss": 0.118, "step": 6365 }, { "epoch": 3.37, "learning_rate": 5.075424476143804e-06, "loss": 0.0969, "step": 6366 }, { "epoch": 3.37, "learning_rate": 5.072440110560779e-06, "loss": 0.0395, "step": 6367 }, { "epoch": 3.37, "learning_rate": 5.06945632446767e-06, "loss": 0.1078, "step": 6368 }, { "epoch": 3.37, "learning_rate": 5.0664731182153825e-06, "loss": 0.096, "step": 6369 }, { "epoch": 3.37, "learning_rate": 5.063490492154746e-06, "loss": 0.081, "step": 6370 }, { "epoch": 3.37, "learning_rate": 5.060508446636516e-06, "loss": 0.0828, "step": 6371 }, { "epoch": 3.37, "learning_rate": 5.057526982011396e-06, "loss": 0.1434, "step": 6372 }, { "epoch": 3.37, "learning_rate": 5.054546098630004e-06, "loss": 0.0642, "step": 6373 }, { "epoch": 3.37, "learning_rate": 5.051565796842899e-06, "loss": 0.0814, "step": 6374 }, { "epoch": 3.37, "learning_rate": 5.0485860770005764e-06, "loss": 0.1131, "step": 6375 }, { "epoch": 3.37, "learning_rate": 5.045606939453447e-06, "loss": 0.0531, "step": 6376 }, { "epoch": 3.38, "learning_rate": 5.042628384551871e-06, "loss": 0.0805, "step": 6377 }, { "epoch": 3.38, "learning_rate": 5.039650412646127e-06, "loss": 0.0947, "step": 6378 }, { "epoch": 3.38, "learning_rate": 5.0366730240864295e-06, "loss": 0.0807, "step": 6379 }, { "epoch": 3.38, "learning_rate": 5.033696219222928e-06, "loss": 0.0597, "step": 6380 }, { "epoch": 3.38, "learning_rate": 5.030719998405696e-06, "loss": 0.0825, "step": 6381 }, { "epoch": 3.38, "learning_rate": 5.0277443619847466e-06, "loss": 0.1307, "step": 6382 }, { "epoch": 3.38, "learning_rate": 5.0247693103100145e-06, "loss": 0.1196, "step": 6383 }, { "epoch": 3.38, "learning_rate": 5.021794843731376e-06, "loss": 0.0682, "step": 6384 }, { "epoch": 3.38, "learning_rate": 5.018820962598639e-06, "loss": 0.0746, "step": 6385 }, { "epoch": 3.38, "learning_rate": 5.0158476672615245e-06, "loss": 0.07, "step": 6386 }, { "epoch": 3.38, "learning_rate": 5.0128749580697065e-06, "loss": 0.0614, "step": 6387 }, { "epoch": 3.38, "learning_rate": 5.009902835372773e-06, "loss": 0.0988, "step": 6388 }, { "epoch": 3.38, "learning_rate": 5.006931299520258e-06, "loss": 0.0615, "step": 6389 }, { "epoch": 3.38, "learning_rate": 5.003960350861621e-06, "loss": 0.1378, "step": 6390 }, { "epoch": 3.38, "learning_rate": 5.000989989746243e-06, "loss": 0.0714, "step": 6391 }, { "epoch": 3.38, "learning_rate": 4.998020216523453e-06, "loss": 0.1004, "step": 6392 }, { "epoch": 3.38, "learning_rate": 4.995051031542496e-06, "loss": 0.0894, "step": 6393 }, { "epoch": 3.38, "learning_rate": 4.992082435152551e-06, "loss": 0.1535, "step": 6394 }, { "epoch": 3.38, "learning_rate": 4.9891144277027385e-06, "loss": 0.108, "step": 6395 }, { "epoch": 3.39, "learning_rate": 4.986147009542092e-06, "loss": 0.115, "step": 6396 }, { "epoch": 3.39, "learning_rate": 4.9831801810195936e-06, "loss": 0.1295, "step": 6397 }, { "epoch": 3.39, "learning_rate": 4.98021394248414e-06, "loss": 0.1215, "step": 6398 }, { "epoch": 3.39, "learning_rate": 4.9772482942845705e-06, "loss": 0.0943, "step": 6399 }, { "epoch": 3.39, "learning_rate": 4.9742832367696545e-06, "loss": 0.1022, "step": 6400 }, { "epoch": 3.39, "learning_rate": 4.971318770288084e-06, "loss": 0.1141, "step": 6401 }, { "epoch": 3.39, "learning_rate": 4.968354895188485e-06, "loss": 0.1237, "step": 6402 }, { "epoch": 3.39, "learning_rate": 4.965391611819412e-06, "loss": 0.1088, "step": 6403 }, { "epoch": 3.39, "learning_rate": 4.962428920529357e-06, "loss": 0.0925, "step": 6404 }, { "epoch": 3.39, "learning_rate": 4.959466821666739e-06, "loss": 0.1677, "step": 6405 }, { "epoch": 3.39, "learning_rate": 4.956505315579899e-06, "loss": 0.099, "step": 6406 }, { "epoch": 3.39, "learning_rate": 4.9535444026171245e-06, "loss": 0.0521, "step": 6407 }, { "epoch": 3.39, "learning_rate": 4.950584083126616e-06, "loss": 0.1034, "step": 6408 }, { "epoch": 3.39, "learning_rate": 4.947624357456521e-06, "loss": 0.055, "step": 6409 }, { "epoch": 3.39, "learning_rate": 4.944665225954902e-06, "loss": 0.0425, "step": 6410 }, { "epoch": 3.39, "learning_rate": 4.941706688969757e-06, "loss": 0.1156, "step": 6411 }, { "epoch": 3.39, "learning_rate": 4.938748746849023e-06, "loss": 0.1195, "step": 6412 }, { "epoch": 3.39, "learning_rate": 4.93579139994055e-06, "loss": 0.0833, "step": 6413 }, { "epoch": 3.39, "learning_rate": 4.932834648592135e-06, "loss": 0.0418, "step": 6414 }, { "epoch": 3.4, "learning_rate": 4.929878493151491e-06, "loss": 0.1322, "step": 6415 }, { "epoch": 3.4, "learning_rate": 4.926922933966271e-06, "loss": 0.1028, "step": 6416 }, { "epoch": 3.4, "learning_rate": 4.923967971384061e-06, "loss": 0.0674, "step": 6417 }, { "epoch": 3.4, "learning_rate": 4.9210136057523525e-06, "loss": 0.1086, "step": 6418 }, { "epoch": 3.4, "learning_rate": 4.918059837418596e-06, "loss": 0.0696, "step": 6419 }, { "epoch": 3.4, "learning_rate": 4.915106666730162e-06, "loss": 0.0785, "step": 6420 }, { "epoch": 3.4, "learning_rate": 4.91215409403434e-06, "loss": 0.1253, "step": 6421 }, { "epoch": 3.4, "learning_rate": 4.909202119678368e-06, "loss": 0.0952, "step": 6422 }, { "epoch": 3.4, "learning_rate": 4.906250744009394e-06, "loss": 0.0837, "step": 6423 }, { "epoch": 3.4, "learning_rate": 4.903299967374512e-06, "loss": 0.0619, "step": 6424 }, { "epoch": 3.4, "learning_rate": 4.900349790120737e-06, "loss": 0.0667, "step": 6425 }, { "epoch": 3.4, "learning_rate": 4.8974002125950096e-06, "loss": 0.0658, "step": 6426 }, { "epoch": 3.4, "learning_rate": 4.894451235144213e-06, "loss": 0.09, "step": 6427 }, { "epoch": 3.4, "learning_rate": 4.891502858115147e-06, "loss": 0.0766, "step": 6428 }, { "epoch": 3.4, "learning_rate": 4.8885550818545505e-06, "loss": 0.0937, "step": 6429 }, { "epoch": 3.4, "learning_rate": 4.8856079067090826e-06, "loss": 0.072, "step": 6430 }, { "epoch": 3.4, "learning_rate": 4.8826613330253365e-06, "loss": 0.1362, "step": 6431 }, { "epoch": 3.4, "learning_rate": 4.879715361149843e-06, "loss": 0.1074, "step": 6432 }, { "epoch": 3.41, "learning_rate": 4.8767699914290465e-06, "loss": 0.1429, "step": 6433 }, { "epoch": 3.41, "learning_rate": 4.873825224209325e-06, "loss": 0.1099, "step": 6434 }, { "epoch": 3.41, "learning_rate": 4.870881059836997e-06, "loss": 0.1124, "step": 6435 }, { "epoch": 3.41, "learning_rate": 4.867937498658293e-06, "loss": 0.1038, "step": 6436 }, { "epoch": 3.41, "learning_rate": 4.864994541019387e-06, "loss": 0.1488, "step": 6437 }, { "epoch": 3.41, "learning_rate": 4.862052187266371e-06, "loss": 0.1317, "step": 6438 }, { "epoch": 3.41, "learning_rate": 4.859110437745279e-06, "loss": 0.0546, "step": 6439 }, { "epoch": 3.41, "learning_rate": 4.8561692928020565e-06, "loss": 0.0677, "step": 6440 }, { "epoch": 3.41, "learning_rate": 4.853228752782597e-06, "loss": 0.1144, "step": 6441 }, { "epoch": 3.41, "learning_rate": 4.850288818032708e-06, "loss": 0.1169, "step": 6442 }, { "epoch": 3.41, "learning_rate": 4.847349488898127e-06, "loss": 0.0658, "step": 6443 }, { "epoch": 3.41, "learning_rate": 4.844410765724533e-06, "loss": 0.1003, "step": 6444 }, { "epoch": 3.41, "learning_rate": 4.841472648857517e-06, "loss": 0.0744, "step": 6445 }, { "epoch": 3.41, "learning_rate": 4.838535138642611e-06, "loss": 0.0692, "step": 6446 }, { "epoch": 3.41, "learning_rate": 4.835598235425276e-06, "loss": 0.053, "step": 6447 }, { "epoch": 3.41, "learning_rate": 4.832661939550889e-06, "loss": 0.0747, "step": 6448 }, { "epoch": 3.41, "learning_rate": 4.829726251364771e-06, "loss": 0.0983, "step": 6449 }, { "epoch": 3.41, "learning_rate": 4.826791171212158e-06, "loss": 0.1057, "step": 6450 }, { "epoch": 3.41, "learning_rate": 4.823856699438222e-06, "loss": 0.1, "step": 6451 }, { "epoch": 3.42, "learning_rate": 4.8209228363880666e-06, "loss": 0.0981, "step": 6452 }, { "epoch": 3.42, "learning_rate": 4.817989582406712e-06, "loss": 0.0735, "step": 6453 }, { "epoch": 3.42, "learning_rate": 4.815056937839122e-06, "loss": 0.0689, "step": 6454 }, { "epoch": 3.42, "learning_rate": 4.812124903030173e-06, "loss": 0.154, "step": 6455 }, { "epoch": 3.42, "learning_rate": 4.809193478324686e-06, "loss": 0.0939, "step": 6456 }, { "epoch": 3.42, "learning_rate": 4.8062626640673935e-06, "loss": 0.078, "step": 6457 }, { "epoch": 3.42, "learning_rate": 4.803332460602973e-06, "loss": 0.0951, "step": 6458 }, { "epoch": 3.42, "learning_rate": 4.800402868276017e-06, "loss": 0.0998, "step": 6459 }, { "epoch": 3.42, "learning_rate": 4.7974738874310475e-06, "loss": 0.073, "step": 6460 }, { "epoch": 3.42, "learning_rate": 4.794545518412521e-06, "loss": 0.1014, "step": 6461 }, { "epoch": 3.42, "learning_rate": 4.791617761564824e-06, "loss": 0.0963, "step": 6462 }, { "epoch": 3.42, "learning_rate": 4.788690617232257e-06, "loss": 0.1024, "step": 6463 }, { "epoch": 3.42, "learning_rate": 4.785764085759066e-06, "loss": 0.1042, "step": 6464 }, { "epoch": 3.42, "learning_rate": 4.782838167489408e-06, "loss": 0.1222, "step": 6465 }, { "epoch": 3.42, "learning_rate": 4.779912862767385e-06, "loss": 0.0817, "step": 6466 }, { "epoch": 3.42, "learning_rate": 4.776988171937014e-06, "loss": 0.0857, "step": 6467 }, { "epoch": 3.42, "learning_rate": 4.774064095342239e-06, "loss": 0.0896, "step": 6468 }, { "epoch": 3.42, "learning_rate": 4.7711406333269465e-06, "loss": 0.054, "step": 6469 }, { "epoch": 3.42, "learning_rate": 4.76821778623493e-06, "loss": 0.0864, "step": 6470 }, { "epoch": 3.43, "learning_rate": 4.765295554409932e-06, "loss": 0.0948, "step": 6471 }, { "epoch": 3.43, "learning_rate": 4.762373938195603e-06, "loss": 0.1112, "step": 6472 }, { "epoch": 3.43, "learning_rate": 4.759452937935536e-06, "loss": 0.0792, "step": 6473 }, { "epoch": 3.43, "learning_rate": 4.756532553973252e-06, "loss": 0.0822, "step": 6474 }, { "epoch": 3.43, "learning_rate": 4.753612786652178e-06, "loss": 0.1174, "step": 6475 }, { "epoch": 3.43, "learning_rate": 4.750693636315696e-06, "loss": 0.1024, "step": 6476 }, { "epoch": 3.43, "learning_rate": 4.7477751033070966e-06, "loss": 0.1161, "step": 6477 }, { "epoch": 3.43, "learning_rate": 4.744857187969606e-06, "loss": 0.0514, "step": 6478 }, { "epoch": 3.43, "learning_rate": 4.741939890646381e-06, "loss": 0.1281, "step": 6479 }, { "epoch": 3.43, "learning_rate": 4.739023211680495e-06, "loss": 0.0488, "step": 6480 }, { "epoch": 3.43, "learning_rate": 4.73610715141496e-06, "loss": 0.0909, "step": 6481 }, { "epoch": 3.43, "learning_rate": 4.733191710192708e-06, "loss": 0.1013, "step": 6482 }, { "epoch": 3.43, "learning_rate": 4.730276888356596e-06, "loss": 0.0854, "step": 6483 }, { "epoch": 3.43, "learning_rate": 4.727362686249419e-06, "loss": 0.0793, "step": 6484 }, { "epoch": 3.43, "learning_rate": 4.724449104213885e-06, "loss": 0.1065, "step": 6485 }, { "epoch": 3.43, "learning_rate": 4.721536142592646e-06, "loss": 0.086, "step": 6486 }, { "epoch": 3.43, "learning_rate": 4.7186238017282606e-06, "loss": 0.0908, "step": 6487 }, { "epoch": 3.43, "learning_rate": 4.71571208196323e-06, "loss": 0.1082, "step": 6488 }, { "epoch": 3.43, "learning_rate": 4.712800983639985e-06, "loss": 0.0916, "step": 6489 }, { "epoch": 3.44, "learning_rate": 4.709890507100868e-06, "loss": 0.0709, "step": 6490 }, { "epoch": 3.44, "learning_rate": 4.706980652688159e-06, "loss": 0.0995, "step": 6491 }, { "epoch": 3.44, "learning_rate": 4.704071420744056e-06, "loss": 0.0423, "step": 6492 }, { "epoch": 3.44, "learning_rate": 4.701162811610695e-06, "loss": 0.1017, "step": 6493 }, { "epoch": 3.44, "learning_rate": 4.698254825630138e-06, "loss": 0.0972, "step": 6494 }, { "epoch": 3.44, "learning_rate": 4.69534746314436e-06, "loss": 0.0797, "step": 6495 }, { "epoch": 3.44, "learning_rate": 4.692440724495283e-06, "loss": 0.084, "step": 6496 }, { "epoch": 3.44, "learning_rate": 4.689534610024733e-06, "loss": 0.0742, "step": 6497 }, { "epoch": 3.44, "learning_rate": 4.6866291200744854e-06, "loss": 0.0809, "step": 6498 }, { "epoch": 3.44, "learning_rate": 4.683724254986224e-06, "loss": 0.0932, "step": 6499 }, { "epoch": 3.44, "learning_rate": 4.680820015101565e-06, "loss": 0.153, "step": 6500 }, { "epoch": 3.44, "learning_rate": 4.6779164007620585e-06, "loss": 0.1134, "step": 6501 }, { "epoch": 3.44, "learning_rate": 4.675013412309167e-06, "loss": 0.1234, "step": 6502 }, { "epoch": 3.44, "learning_rate": 4.672111050084292e-06, "loss": 0.1223, "step": 6503 }, { "epoch": 3.44, "learning_rate": 4.669209314428761e-06, "loss": 0.1601, "step": 6504 }, { "epoch": 3.44, "learning_rate": 4.666308205683814e-06, "loss": 0.0664, "step": 6505 }, { "epoch": 3.44, "learning_rate": 4.663407724190638e-06, "loss": 0.0799, "step": 6506 }, { "epoch": 3.44, "learning_rate": 4.66050787029032e-06, "loss": 0.0765, "step": 6507 }, { "epoch": 3.44, "learning_rate": 4.6576086443238975e-06, "loss": 0.0715, "step": 6508 }, { "epoch": 3.45, "learning_rate": 4.654710046632326e-06, "loss": 0.0991, "step": 6509 }, { "epoch": 3.45, "learning_rate": 4.651812077556479e-06, "loss": 0.0677, "step": 6510 }, { "epoch": 3.45, "learning_rate": 4.6489147374371716e-06, "loss": 0.0747, "step": 6511 }, { "epoch": 3.45, "learning_rate": 4.646018026615126e-06, "loss": 0.0951, "step": 6512 }, { "epoch": 3.45, "learning_rate": 4.643121945431011e-06, "loss": 0.0756, "step": 6513 }, { "epoch": 3.45, "learning_rate": 4.640226494225406e-06, "loss": 0.0868, "step": 6514 }, { "epoch": 3.45, "learning_rate": 4.637331673338816e-06, "loss": 0.101, "step": 6515 }, { "epoch": 3.45, "learning_rate": 4.634437483111688e-06, "loss": 0.0686, "step": 6516 }, { "epoch": 3.45, "learning_rate": 4.631543923884374e-06, "loss": 0.0872, "step": 6517 }, { "epoch": 3.45, "learning_rate": 4.62865099599717e-06, "loss": 0.0983, "step": 6518 }, { "epoch": 3.45, "learning_rate": 4.625758699790282e-06, "loss": 0.0877, "step": 6519 }, { "epoch": 3.45, "learning_rate": 4.622867035603853e-06, "loss": 0.0955, "step": 6520 }, { "epoch": 3.45, "learning_rate": 4.619976003777952e-06, "loss": 0.0975, "step": 6521 }, { "epoch": 3.45, "learning_rate": 4.6170856046525644e-06, "loss": 0.0749, "step": 6522 }, { "epoch": 3.45, "learning_rate": 4.6141958385676046e-06, "loss": 0.0758, "step": 6523 }, { "epoch": 3.45, "learning_rate": 4.61130670586292e-06, "loss": 0.1339, "step": 6524 }, { "epoch": 3.45, "learning_rate": 4.6084182068782715e-06, "loss": 0.0877, "step": 6525 }, { "epoch": 3.45, "learning_rate": 4.605530341953359e-06, "loss": 0.0966, "step": 6526 }, { "epoch": 3.45, "learning_rate": 4.602643111427794e-06, "loss": 0.0987, "step": 6527 }, { "epoch": 3.46, "learning_rate": 4.599756515641125e-06, "loss": 0.1077, "step": 6528 }, { "epoch": 3.46, "learning_rate": 4.596870554932815e-06, "loss": 0.072, "step": 6529 }, { "epoch": 3.46, "learning_rate": 4.593985229642267e-06, "loss": 0.0928, "step": 6530 }, { "epoch": 3.46, "learning_rate": 4.5911005401087936e-06, "loss": 0.0912, "step": 6531 }, { "epoch": 3.46, "learning_rate": 4.58821648667164e-06, "loss": 0.0536, "step": 6532 }, { "epoch": 3.46, "learning_rate": 4.585333069669978e-06, "loss": 0.1152, "step": 6533 }, { "epoch": 3.46, "learning_rate": 4.5824502894429e-06, "loss": 0.0839, "step": 6534 }, { "epoch": 3.46, "learning_rate": 4.579568146329428e-06, "loss": 0.1148, "step": 6535 }, { "epoch": 3.46, "learning_rate": 4.576686640668509e-06, "loss": 0.1198, "step": 6536 }, { "epoch": 3.46, "learning_rate": 4.573805772799011e-06, "loss": 0.1069, "step": 6537 }, { "epoch": 3.46, "learning_rate": 4.570925543059731e-06, "loss": 0.1058, "step": 6538 }, { "epoch": 3.46, "learning_rate": 4.5680459517893895e-06, "loss": 0.0968, "step": 6539 }, { "epoch": 3.46, "learning_rate": 4.565166999326625e-06, "loss": 0.0896, "step": 6540 }, { "epoch": 3.46, "learning_rate": 4.562288686010017e-06, "loss": 0.0652, "step": 6541 }, { "epoch": 3.46, "learning_rate": 4.559411012178052e-06, "loss": 0.0545, "step": 6542 }, { "epoch": 3.46, "learning_rate": 4.556533978169159e-06, "loss": 0.1217, "step": 6543 }, { "epoch": 3.46, "learning_rate": 4.553657584321671e-06, "loss": 0.0786, "step": 6544 }, { "epoch": 3.46, "learning_rate": 4.5507818309738635e-06, "loss": 0.0957, "step": 6545 }, { "epoch": 3.46, "learning_rate": 4.547906718463938e-06, "loss": 0.1118, "step": 6546 }, { "epoch": 3.47, "learning_rate": 4.545032247129997e-06, "loss": 0.0972, "step": 6547 }, { "epoch": 3.47, "learning_rate": 4.542158417310095e-06, "loss": 0.132, "step": 6548 }, { "epoch": 3.47, "learning_rate": 4.5392852293421916e-06, "loss": 0.0827, "step": 6549 }, { "epoch": 3.47, "learning_rate": 4.536412683564183e-06, "loss": 0.1281, "step": 6550 }, { "epoch": 3.47, "learning_rate": 4.533540780313891e-06, "loss": 0.0619, "step": 6551 }, { "epoch": 3.47, "learning_rate": 4.530669519929047e-06, "loss": 0.077, "step": 6552 }, { "epoch": 3.47, "learning_rate": 4.527798902747323e-06, "loss": 0.1216, "step": 6553 }, { "epoch": 3.47, "learning_rate": 4.5249289291063084e-06, "loss": 0.092, "step": 6554 }, { "epoch": 3.47, "learning_rate": 4.522059599343512e-06, "loss": 0.0941, "step": 6555 }, { "epoch": 3.47, "learning_rate": 4.519190913796377e-06, "loss": 0.0975, "step": 6556 }, { "epoch": 3.47, "learning_rate": 4.516322872802263e-06, "loss": 0.0515, "step": 6557 }, { "epoch": 3.47, "learning_rate": 4.513455476698462e-06, "loss": 0.0762, "step": 6558 }, { "epoch": 3.47, "learning_rate": 4.510588725822177e-06, "loss": 0.0849, "step": 6559 }, { "epoch": 3.47, "learning_rate": 4.50772262051055e-06, "loss": 0.0391, "step": 6560 }, { "epoch": 3.47, "learning_rate": 4.504857161100637e-06, "loss": 0.1169, "step": 6561 }, { "epoch": 3.47, "learning_rate": 4.501992347929419e-06, "loss": 0.0648, "step": 6562 }, { "epoch": 3.47, "learning_rate": 4.499128181333814e-06, "loss": 0.0688, "step": 6563 }, { "epoch": 3.47, "learning_rate": 4.496264661650637e-06, "loss": 0.0853, "step": 6564 }, { "epoch": 3.47, "learning_rate": 4.493401789216653e-06, "loss": 0.094, "step": 6565 }, { "epoch": 3.48, "learning_rate": 4.490539564368541e-06, "loss": 0.0761, "step": 6566 }, { "epoch": 3.48, "learning_rate": 4.4876779874429e-06, "loss": 0.1619, "step": 6567 }, { "epoch": 3.48, "learning_rate": 4.4848170587762605e-06, "loss": 0.1097, "step": 6568 }, { "epoch": 3.48, "learning_rate": 4.481956778705068e-06, "loss": 0.1227, "step": 6569 }, { "epoch": 3.48, "learning_rate": 4.479097147565703e-06, "loss": 0.086, "step": 6570 }, { "epoch": 3.48, "learning_rate": 4.47623816569446e-06, "loss": 0.0868, "step": 6571 }, { "epoch": 3.48, "learning_rate": 4.473379833427555e-06, "loss": 0.0882, "step": 6572 }, { "epoch": 3.48, "learning_rate": 4.470522151101144e-06, "loss": 0.1117, "step": 6573 }, { "epoch": 3.48, "learning_rate": 4.467665119051285e-06, "loss": 0.0457, "step": 6574 }, { "epoch": 3.48, "learning_rate": 4.464808737613979e-06, "loss": 0.1065, "step": 6575 }, { "epoch": 3.48, "learning_rate": 4.461953007125134e-06, "loss": 0.0802, "step": 6576 }, { "epoch": 3.48, "learning_rate": 4.459097927920592e-06, "loss": 0.0723, "step": 6577 }, { "epoch": 3.48, "learning_rate": 4.456243500336119e-06, "loss": 0.0582, "step": 6578 }, { "epoch": 3.48, "learning_rate": 4.453389724707399e-06, "loss": 0.0981, "step": 6579 }, { "epoch": 3.48, "learning_rate": 4.450536601370036e-06, "loss": 0.1004, "step": 6580 }, { "epoch": 3.48, "learning_rate": 4.44768413065957e-06, "loss": 0.1352, "step": 6581 }, { "epoch": 3.48, "learning_rate": 4.444832312911449e-06, "loss": 0.0642, "step": 6582 }, { "epoch": 3.48, "learning_rate": 4.441981148461061e-06, "loss": 0.0683, "step": 6583 }, { "epoch": 3.48, "learning_rate": 4.439130637643698e-06, "loss": 0.1258, "step": 6584 }, { "epoch": 3.49, "learning_rate": 4.436280780794595e-06, "loss": 0.1416, "step": 6585 }, { "epoch": 3.49, "learning_rate": 4.433431578248891e-06, "loss": 0.1076, "step": 6586 }, { "epoch": 3.49, "learning_rate": 4.430583030341666e-06, "loss": 0.0858, "step": 6587 }, { "epoch": 3.49, "learning_rate": 4.427735137407911e-06, "loss": 0.0798, "step": 6588 }, { "epoch": 3.49, "learning_rate": 4.424887899782538e-06, "loss": 0.0959, "step": 6589 }, { "epoch": 3.49, "learning_rate": 4.422041317800396e-06, "loss": 0.0908, "step": 6590 }, { "epoch": 3.49, "learning_rate": 4.4191953917962405e-06, "loss": 0.0638, "step": 6591 }, { "epoch": 3.49, "learning_rate": 4.416350122104761e-06, "loss": 0.0695, "step": 6592 }, { "epoch": 3.49, "learning_rate": 4.41350550906057e-06, "loss": 0.0767, "step": 6593 }, { "epoch": 3.49, "learning_rate": 4.410661552998192e-06, "loss": 0.1369, "step": 6594 }, { "epoch": 3.49, "learning_rate": 4.407818254252093e-06, "loss": 0.0918, "step": 6595 }, { "epoch": 3.49, "learning_rate": 4.404975613156634e-06, "loss": 0.106, "step": 6596 }, { "epoch": 3.49, "learning_rate": 4.402133630046123e-06, "loss": 0.0871, "step": 6597 }, { "epoch": 3.49, "learning_rate": 4.399292305254786e-06, "loss": 0.068, "step": 6598 }, { "epoch": 3.49, "learning_rate": 4.396451639116759e-06, "loss": 0.1454, "step": 6599 }, { "epoch": 3.49, "learning_rate": 4.393611631966119e-06, "loss": 0.1439, "step": 6600 }, { "epoch": 3.49, "learning_rate": 4.390772284136847e-06, "loss": 0.0966, "step": 6601 }, { "epoch": 3.49, "learning_rate": 4.387933595962865e-06, "loss": 0.0486, "step": 6602 }, { "epoch": 3.5, "learning_rate": 4.385095567778002e-06, "loss": 0.1046, "step": 6603 }, { "epoch": 3.5, "learning_rate": 4.382258199916012e-06, "loss": 0.0729, "step": 6604 }, { "epoch": 3.5, "learning_rate": 4.3794214927105814e-06, "loss": 0.1169, "step": 6605 }, { "epoch": 3.5, "learning_rate": 4.376585446495306e-06, "loss": 0.0942, "step": 6606 }, { "epoch": 3.5, "learning_rate": 4.373750061603713e-06, "loss": 0.0895, "step": 6607 }, { "epoch": 3.5, "learning_rate": 4.3709153383692525e-06, "loss": 0.0909, "step": 6608 }, { "epoch": 3.5, "learning_rate": 4.368081277125287e-06, "loss": 0.1094, "step": 6609 }, { "epoch": 3.5, "learning_rate": 4.365247878205111e-06, "loss": 0.0699, "step": 6610 }, { "epoch": 3.5, "learning_rate": 4.3624151419419385e-06, "loss": 0.0648, "step": 6611 }, { "epoch": 3.5, "learning_rate": 4.359583068668896e-06, "loss": 0.0833, "step": 6612 }, { "epoch": 3.5, "learning_rate": 4.356751658719051e-06, "loss": 0.1016, "step": 6613 }, { "epoch": 3.5, "learning_rate": 4.353920912425374e-06, "loss": 0.0697, "step": 6614 }, { "epoch": 3.5, "learning_rate": 4.351090830120772e-06, "loss": 0.0729, "step": 6615 }, { "epoch": 3.5, "learning_rate": 4.348261412138062e-06, "loss": 0.0771, "step": 6616 }, { "epoch": 3.5, "learning_rate": 4.345432658809996e-06, "loss": 0.0461, "step": 6617 }, { "epoch": 3.5, "learning_rate": 4.342604570469233e-06, "loss": 0.0886, "step": 6618 }, { "epoch": 3.5, "learning_rate": 4.339777147448367e-06, "loss": 0.0999, "step": 6619 }, { "epoch": 3.5, "learning_rate": 4.336950390079908e-06, "loss": 0.0619, "step": 6620 }, { "epoch": 3.5, "learning_rate": 4.334124298696281e-06, "loss": 0.0737, "step": 6621 }, { "epoch": 3.51, "learning_rate": 4.331298873629847e-06, "loss": 0.0837, "step": 6622 }, { "epoch": 3.51, "learning_rate": 4.328474115212874e-06, "loss": 0.04, "step": 6623 }, { "epoch": 3.51, "learning_rate": 4.325650023777563e-06, "loss": 0.0778, "step": 6624 }, { "epoch": 3.51, "learning_rate": 4.322826599656037e-06, "loss": 0.1452, "step": 6625 }, { "epoch": 3.51, "learning_rate": 4.320003843180326e-06, "loss": 0.1275, "step": 6626 }, { "epoch": 3.51, "learning_rate": 4.317181754682399e-06, "loss": 0.1203, "step": 6627 }, { "epoch": 3.51, "learning_rate": 4.314360334494136e-06, "loss": 0.082, "step": 6628 }, { "epoch": 3.51, "learning_rate": 4.311539582947337e-06, "loss": 0.1046, "step": 6629 }, { "epoch": 3.51, "learning_rate": 4.308719500373736e-06, "loss": 0.1, "step": 6630 }, { "epoch": 3.51, "learning_rate": 4.305900087104971e-06, "loss": 0.0999, "step": 6631 }, { "epoch": 3.51, "learning_rate": 4.303081343472616e-06, "loss": 0.0952, "step": 6632 }, { "epoch": 3.51, "learning_rate": 4.300263269808156e-06, "loss": 0.14, "step": 6633 }, { "epoch": 3.51, "learning_rate": 4.297445866443004e-06, "loss": 0.0613, "step": 6634 }, { "epoch": 3.51, "learning_rate": 4.294629133708499e-06, "loss": 0.0873, "step": 6635 }, { "epoch": 3.51, "learning_rate": 4.29181307193588e-06, "loss": 0.1121, "step": 6636 }, { "epoch": 3.51, "learning_rate": 4.288997681456329e-06, "loss": 0.1168, "step": 6637 }, { "epoch": 3.51, "learning_rate": 4.286182962600938e-06, "loss": 0.1558, "step": 6638 }, { "epoch": 3.51, "learning_rate": 4.283368915700723e-06, "loss": 0.0875, "step": 6639 }, { "epoch": 3.51, "learning_rate": 4.280555541086627e-06, "loss": 0.1267, "step": 6640 }, { "epoch": 3.52, "learning_rate": 4.277742839089498e-06, "loss": 0.1869, "step": 6641 }, { "epoch": 3.52, "learning_rate": 4.274930810040126e-06, "loss": 0.0913, "step": 6642 }, { "epoch": 3.52, "learning_rate": 4.272119454269202e-06, "loss": 0.1173, "step": 6643 }, { "epoch": 3.52, "learning_rate": 4.269308772107349e-06, "loss": 0.0804, "step": 6644 }, { "epoch": 3.52, "learning_rate": 4.26649876388511e-06, "loss": 0.1158, "step": 6645 }, { "epoch": 3.52, "learning_rate": 4.263689429932943e-06, "loss": 0.0787, "step": 6646 }, { "epoch": 3.52, "learning_rate": 4.260880770581237e-06, "loss": 0.056, "step": 6647 }, { "epoch": 3.52, "learning_rate": 4.2580727861602875e-06, "loss": 0.0676, "step": 6648 }, { "epoch": 3.52, "learning_rate": 4.255265477000323e-06, "loss": 0.0888, "step": 6649 }, { "epoch": 3.52, "learning_rate": 4.252458843431493e-06, "loss": 0.122, "step": 6650 }, { "epoch": 3.52, "learning_rate": 4.2496528857838555e-06, "loss": 0.0485, "step": 6651 }, { "epoch": 3.52, "learning_rate": 4.2468476043874e-06, "loss": 0.1024, "step": 6652 }, { "epoch": 3.52, "learning_rate": 4.244042999572027e-06, "loss": 0.0849, "step": 6653 }, { "epoch": 3.52, "learning_rate": 4.241239071667566e-06, "loss": 0.0783, "step": 6654 }, { "epoch": 3.52, "learning_rate": 4.238435821003769e-06, "loss": 0.1146, "step": 6655 }, { "epoch": 3.52, "learning_rate": 4.235633247910296e-06, "loss": 0.1565, "step": 6656 }, { "epoch": 3.52, "learning_rate": 4.232831352716742e-06, "loss": 0.1326, "step": 6657 }, { "epoch": 3.52, "learning_rate": 4.2300301357526064e-06, "loss": 0.0622, "step": 6658 }, { "epoch": 3.52, "learning_rate": 4.2272295973473255e-06, "loss": 0.0902, "step": 6659 }, { "epoch": 3.53, "learning_rate": 4.224429737830244e-06, "loss": 0.1077, "step": 6660 }, { "epoch": 3.53, "learning_rate": 4.221630557530627e-06, "loss": 0.0755, "step": 6661 }, { "epoch": 3.53, "learning_rate": 4.218832056777671e-06, "loss": 0.1135, "step": 6662 }, { "epoch": 3.53, "learning_rate": 4.216034235900475e-06, "loss": 0.063, "step": 6663 }, { "epoch": 3.53, "learning_rate": 4.213237095228079e-06, "loss": 0.1216, "step": 6664 }, { "epoch": 3.53, "learning_rate": 4.21044063508942e-06, "loss": 0.1341, "step": 6665 }, { "epoch": 3.53, "learning_rate": 4.207644855813374e-06, "loss": 0.124, "step": 6666 }, { "epoch": 3.53, "learning_rate": 4.204849757728735e-06, "loss": 0.0839, "step": 6667 }, { "epoch": 3.53, "learning_rate": 4.202055341164197e-06, "loss": 0.1231, "step": 6668 }, { "epoch": 3.53, "learning_rate": 4.199261606448397e-06, "loss": 0.0658, "step": 6669 }, { "epoch": 3.53, "learning_rate": 4.196468553909886e-06, "loss": 0.1076, "step": 6670 }, { "epoch": 3.53, "learning_rate": 4.193676183877125e-06, "loss": 0.0763, "step": 6671 }, { "epoch": 3.53, "learning_rate": 4.190884496678509e-06, "loss": 0.1347, "step": 6672 }, { "epoch": 3.53, "learning_rate": 4.1880934926423376e-06, "loss": 0.0672, "step": 6673 }, { "epoch": 3.53, "learning_rate": 4.185303172096846e-06, "loss": 0.0554, "step": 6674 }, { "epoch": 3.53, "learning_rate": 4.182513535370177e-06, "loss": 0.0902, "step": 6675 }, { "epoch": 3.53, "learning_rate": 4.179724582790393e-06, "loss": 0.1158, "step": 6676 }, { "epoch": 3.53, "learning_rate": 4.176936314685488e-06, "loss": 0.1033, "step": 6677 }, { "epoch": 3.53, "learning_rate": 4.174148731383358e-06, "loss": 0.0632, "step": 6678 }, { "epoch": 3.54, "learning_rate": 4.171361833211837e-06, "loss": 0.07, "step": 6679 }, { "epoch": 3.54, "learning_rate": 4.168575620498663e-06, "loss": 0.0937, "step": 6680 }, { "epoch": 3.54, "learning_rate": 4.1657900935715e-06, "loss": 0.04, "step": 6681 }, { "epoch": 3.54, "learning_rate": 4.1630052527579376e-06, "loss": 0.1064, "step": 6682 }, { "epoch": 3.54, "learning_rate": 4.160221098385473e-06, "loss": 0.1018, "step": 6683 }, { "epoch": 3.54, "learning_rate": 4.157437630781523e-06, "loss": 0.0517, "step": 6684 }, { "epoch": 3.54, "learning_rate": 4.154654850273439e-06, "loss": 0.0921, "step": 6685 }, { "epoch": 3.54, "learning_rate": 4.15187275718847e-06, "loss": 0.1243, "step": 6686 }, { "epoch": 3.54, "learning_rate": 4.149091351853805e-06, "loss": 0.0834, "step": 6687 }, { "epoch": 3.54, "learning_rate": 4.146310634596535e-06, "loss": 0.0796, "step": 6688 }, { "epoch": 3.54, "learning_rate": 4.1435306057436816e-06, "loss": 0.0653, "step": 6689 }, { "epoch": 3.54, "learning_rate": 4.140751265622178e-06, "loss": 0.1232, "step": 6690 }, { "epoch": 3.54, "learning_rate": 4.137972614558885e-06, "loss": 0.0891, "step": 6691 }, { "epoch": 3.54, "learning_rate": 4.135194652880573e-06, "loss": 0.0521, "step": 6692 }, { "epoch": 3.54, "learning_rate": 4.132417380913931e-06, "loss": 0.0755, "step": 6693 }, { "epoch": 3.54, "learning_rate": 4.1296407989855815e-06, "loss": 0.138, "step": 6694 }, { "epoch": 3.54, "learning_rate": 4.1268649074220444e-06, "loss": 0.0773, "step": 6695 }, { "epoch": 3.54, "learning_rate": 4.124089706549777e-06, "loss": 0.0516, "step": 6696 }, { "epoch": 3.54, "learning_rate": 4.121315196695148e-06, "loss": 0.1378, "step": 6697 }, { "epoch": 3.55, "learning_rate": 4.118541378184439e-06, "loss": 0.0713, "step": 6698 }, { "epoch": 3.55, "learning_rate": 4.115768251343865e-06, "loss": 0.1057, "step": 6699 }, { "epoch": 3.55, "learning_rate": 4.112995816499544e-06, "loss": 0.1321, "step": 6700 }, { "epoch": 3.55, "learning_rate": 4.110224073977518e-06, "loss": 0.0897, "step": 6701 }, { "epoch": 3.55, "learning_rate": 4.107453024103755e-06, "loss": 0.0862, "step": 6702 }, { "epoch": 3.55, "learning_rate": 4.104682667204129e-06, "loss": 0.0869, "step": 6703 }, { "epoch": 3.55, "learning_rate": 4.101913003604446e-06, "loss": 0.0771, "step": 6704 }, { "epoch": 3.55, "learning_rate": 4.099144033630415e-06, "loss": 0.1297, "step": 6705 }, { "epoch": 3.55, "learning_rate": 4.09637575760768e-06, "loss": 0.1277, "step": 6706 }, { "epoch": 3.55, "learning_rate": 4.093608175861789e-06, "loss": 0.0921, "step": 6707 }, { "epoch": 3.55, "learning_rate": 4.09084128871822e-06, "loss": 0.099, "step": 6708 }, { "epoch": 3.55, "learning_rate": 4.088075096502362e-06, "loss": 0.0606, "step": 6709 }, { "epoch": 3.55, "learning_rate": 4.085309599539519e-06, "loss": 0.058, "step": 6710 }, { "epoch": 3.55, "learning_rate": 4.082544798154923e-06, "loss": 0.1144, "step": 6711 }, { "epoch": 3.55, "learning_rate": 4.079780692673723e-06, "loss": 0.0759, "step": 6712 }, { "epoch": 3.55, "learning_rate": 4.0770172834209764e-06, "loss": 0.0546, "step": 6713 }, { "epoch": 3.55, "learning_rate": 4.074254570721671e-06, "loss": 0.0959, "step": 6714 }, { "epoch": 3.55, "learning_rate": 4.0714925549007e-06, "loss": 0.1347, "step": 6715 }, { "epoch": 3.55, "learning_rate": 4.0687312362828894e-06, "loss": 0.1553, "step": 6716 }, { "epoch": 3.56, "learning_rate": 4.06597061519297e-06, "loss": 0.091, "step": 6717 }, { "epoch": 3.56, "learning_rate": 4.063210691955595e-06, "loss": 0.0876, "step": 6718 }, { "epoch": 3.56, "learning_rate": 4.0604514668953395e-06, "loss": 0.075, "step": 6719 }, { "epoch": 3.56, "learning_rate": 4.057692940336689e-06, "loss": 0.0961, "step": 6720 }, { "epoch": 3.56, "learning_rate": 4.054935112604059e-06, "loss": 0.0971, "step": 6721 }, { "epoch": 3.56, "learning_rate": 4.052177984021766e-06, "loss": 0.065, "step": 6722 }, { "epoch": 3.56, "learning_rate": 4.049421554914057e-06, "loss": 0.0677, "step": 6723 }, { "epoch": 3.56, "learning_rate": 4.046665825605101e-06, "loss": 0.1037, "step": 6724 }, { "epoch": 3.56, "learning_rate": 4.043910796418961e-06, "loss": 0.0794, "step": 6725 }, { "epoch": 3.56, "learning_rate": 4.0411564676796475e-06, "loss": 0.1475, "step": 6726 }, { "epoch": 3.56, "learning_rate": 4.038402839711064e-06, "loss": 0.0879, "step": 6727 }, { "epoch": 3.56, "learning_rate": 4.035649912837046e-06, "loss": 0.1276, "step": 6728 }, { "epoch": 3.56, "learning_rate": 4.032897687381348e-06, "loss": 0.1092, "step": 6729 }, { "epoch": 3.56, "learning_rate": 4.0301461636676284e-06, "loss": 0.0818, "step": 6730 }, { "epoch": 3.56, "learning_rate": 4.027395342019479e-06, "loss": 0.1204, "step": 6731 }, { "epoch": 3.56, "learning_rate": 4.024645222760397e-06, "loss": 0.0841, "step": 6732 }, { "epoch": 3.56, "learning_rate": 4.021895806213798e-06, "loss": 0.1156, "step": 6733 }, { "epoch": 3.56, "learning_rate": 4.019147092703027e-06, "loss": 0.0766, "step": 6734 }, { "epoch": 3.56, "learning_rate": 4.016399082551327e-06, "loss": 0.0668, "step": 6735 }, { "epoch": 3.57, "learning_rate": 4.013651776081879e-06, "loss": 0.0856, "step": 6736 }, { "epoch": 3.57, "learning_rate": 4.010905173617764e-06, "loss": 0.0681, "step": 6737 }, { "epoch": 3.57, "learning_rate": 4.008159275481991e-06, "loss": 0.1048, "step": 6738 }, { "epoch": 3.57, "learning_rate": 4.005414081997484e-06, "loss": 0.0995, "step": 6739 }, { "epoch": 3.57, "learning_rate": 4.002669593487082e-06, "loss": 0.1029, "step": 6740 }, { "epoch": 3.57, "learning_rate": 3.99992581027354e-06, "loss": 0.1528, "step": 6741 }, { "epoch": 3.57, "learning_rate": 3.997182732679528e-06, "loss": 0.1148, "step": 6742 }, { "epoch": 3.57, "learning_rate": 3.994440361027642e-06, "loss": 0.07, "step": 6743 }, { "epoch": 3.57, "learning_rate": 3.991698695640391e-06, "loss": 0.1028, "step": 6744 }, { "epoch": 3.57, "learning_rate": 3.988957736840197e-06, "loss": 0.086, "step": 6745 }, { "epoch": 3.57, "learning_rate": 3.986217484949404e-06, "loss": 0.0986, "step": 6746 }, { "epoch": 3.57, "learning_rate": 3.9834779402902656e-06, "loss": 0.1205, "step": 6747 }, { "epoch": 3.57, "learning_rate": 3.980739103184964e-06, "loss": 0.1337, "step": 6748 }, { "epoch": 3.57, "learning_rate": 3.978000973955588e-06, "loss": 0.0775, "step": 6749 }, { "epoch": 3.57, "learning_rate": 3.975263552924143e-06, "loss": 0.1237, "step": 6750 }, { "epoch": 3.57, "learning_rate": 3.972526840412563e-06, "loss": 0.1233, "step": 6751 }, { "epoch": 3.57, "learning_rate": 3.969790836742681e-06, "loss": 0.0714, "step": 6752 }, { "epoch": 3.57, "learning_rate": 3.96705554223626e-06, "loss": 0.0915, "step": 6753 }, { "epoch": 3.57, "learning_rate": 3.9643209572149795e-06, "loss": 0.1433, "step": 6754 }, { "epoch": 3.58, "learning_rate": 3.961587082000425e-06, "loss": 0.1009, "step": 6755 }, { "epoch": 3.58, "learning_rate": 3.958853916914116e-06, "loss": 0.0733, "step": 6756 }, { "epoch": 3.58, "learning_rate": 3.956121462277462e-06, "loss": 0.0512, "step": 6757 }, { "epoch": 3.58, "learning_rate": 3.953389718411813e-06, "loss": 0.15, "step": 6758 }, { "epoch": 3.58, "learning_rate": 3.95065868563843e-06, "loss": 0.129, "step": 6759 }, { "epoch": 3.58, "learning_rate": 3.94792836427848e-06, "loss": 0.0762, "step": 6760 }, { "epoch": 3.58, "learning_rate": 3.945198754653061e-06, "loss": 0.1312, "step": 6761 }, { "epoch": 3.58, "learning_rate": 3.942469857083173e-06, "loss": 0.0948, "step": 6762 }, { "epoch": 3.58, "learning_rate": 3.939741671889746e-06, "loss": 0.1039, "step": 6763 }, { "epoch": 3.58, "learning_rate": 3.937014199393618e-06, "loss": 0.0596, "step": 6764 }, { "epoch": 3.58, "learning_rate": 3.934287439915538e-06, "loss": 0.1079, "step": 6765 }, { "epoch": 3.58, "learning_rate": 3.931561393776188e-06, "loss": 0.0956, "step": 6766 }, { "epoch": 3.58, "learning_rate": 3.928836061296145e-06, "loss": 0.1123, "step": 6767 }, { "epoch": 3.58, "learning_rate": 3.926111442795925e-06, "loss": 0.0737, "step": 6768 }, { "epoch": 3.58, "learning_rate": 3.923387538595937e-06, "loss": 0.06, "step": 6769 }, { "epoch": 3.58, "learning_rate": 3.920664349016521e-06, "loss": 0.0597, "step": 6770 }, { "epoch": 3.58, "learning_rate": 3.917941874377934e-06, "loss": 0.0784, "step": 6771 }, { "epoch": 3.58, "learning_rate": 3.91522011500034e-06, "loss": 0.0977, "step": 6772 }, { "epoch": 3.59, "learning_rate": 3.912499071203819e-06, "loss": 0.1011, "step": 6773 }, { "epoch": 3.59, "learning_rate": 3.9097787433083775e-06, "loss": 0.0633, "step": 6774 }, { "epoch": 3.59, "learning_rate": 3.907059131633926e-06, "loss": 0.0766, "step": 6775 }, { "epoch": 3.59, "learning_rate": 3.904340236500299e-06, "loss": 0.0738, "step": 6776 }, { "epoch": 3.59, "learning_rate": 3.901622058227239e-06, "loss": 0.0903, "step": 6777 }, { "epoch": 3.59, "learning_rate": 3.898904597134415e-06, "loss": 0.1112, "step": 6778 }, { "epoch": 3.59, "learning_rate": 3.896187853541399e-06, "loss": 0.0923, "step": 6779 }, { "epoch": 3.59, "learning_rate": 3.893471827767691e-06, "loss": 0.0989, "step": 6780 }, { "epoch": 3.59, "learning_rate": 3.890756520132698e-06, "loss": 0.1445, "step": 6781 }, { "epoch": 3.59, "learning_rate": 3.88804193095574e-06, "loss": 0.0607, "step": 6782 }, { "epoch": 3.59, "learning_rate": 3.8853280605560666e-06, "loss": 0.1335, "step": 6783 }, { "epoch": 3.59, "learning_rate": 3.882614909252825e-06, "loss": 0.1729, "step": 6784 }, { "epoch": 3.59, "learning_rate": 3.879902477365091e-06, "loss": 0.1407, "step": 6785 }, { "epoch": 3.59, "learning_rate": 3.877190765211854e-06, "loss": 0.0741, "step": 6786 }, { "epoch": 3.59, "learning_rate": 3.874479773112011e-06, "loss": 0.2076, "step": 6787 }, { "epoch": 3.59, "learning_rate": 3.871769501384386e-06, "loss": 0.1101, "step": 6788 }, { "epoch": 3.59, "learning_rate": 3.869059950347708e-06, "loss": 0.0931, "step": 6789 }, { "epoch": 3.59, "learning_rate": 3.866351120320622e-06, "loss": 0.1123, "step": 6790 }, { "epoch": 3.59, "learning_rate": 3.863643011621697e-06, "loss": 0.0704, "step": 6791 }, { "epoch": 3.6, "learning_rate": 3.860935624569406e-06, "loss": 0.1065, "step": 6792 }, { "epoch": 3.6, "learning_rate": 3.858228959482148e-06, "loss": 0.0691, "step": 6793 }, { "epoch": 3.6, "learning_rate": 3.855523016678228e-06, "loss": 0.0819, "step": 6794 }, { "epoch": 3.6, "learning_rate": 3.852817796475869e-06, "loss": 0.0879, "step": 6795 }, { "epoch": 3.6, "learning_rate": 3.85011329919322e-06, "loss": 0.0782, "step": 6796 }, { "epoch": 3.6, "learning_rate": 3.847409525148319e-06, "loss": 0.1158, "step": 6797 }, { "epoch": 3.6, "learning_rate": 3.844706474659144e-06, "loss": 0.0739, "step": 6798 }, { "epoch": 3.6, "learning_rate": 3.842004148043576e-06, "loss": 0.0817, "step": 6799 }, { "epoch": 3.6, "learning_rate": 3.839302545619412e-06, "loss": 0.0718, "step": 6800 }, { "epoch": 3.6, "learning_rate": 3.836601667704372e-06, "loss": 0.1338, "step": 6801 }, { "epoch": 3.6, "learning_rate": 3.833901514616076e-06, "loss": 0.1249, "step": 6802 }, { "epoch": 3.6, "learning_rate": 3.831202086672075e-06, "loss": 0.116, "step": 6803 }, { "epoch": 3.6, "learning_rate": 3.828503384189821e-06, "loss": 0.0821, "step": 6804 }, { "epoch": 3.6, "learning_rate": 3.8258054074866845e-06, "loss": 0.0792, "step": 6805 }, { "epoch": 3.6, "learning_rate": 3.823108156879959e-06, "loss": 0.0719, "step": 6806 }, { "epoch": 3.6, "learning_rate": 3.8204116326868376e-06, "loss": 0.0844, "step": 6807 }, { "epoch": 3.6, "learning_rate": 3.817715835224446e-06, "loss": 0.081, "step": 6808 }, { "epoch": 3.6, "learning_rate": 3.815020764809806e-06, "loss": 0.0958, "step": 6809 }, { "epoch": 3.6, "learning_rate": 3.81232642175987e-06, "loss": 0.0751, "step": 6810 }, { "epoch": 3.61, "learning_rate": 3.809632806391491e-06, "loss": 0.1471, "step": 6811 }, { "epoch": 3.61, "learning_rate": 3.806939919021446e-06, "loss": 0.1936, "step": 6812 }, { "epoch": 3.61, "learning_rate": 3.8042477599664308e-06, "loss": 0.0638, "step": 6813 }, { "epoch": 3.61, "learning_rate": 3.8015563295430345e-06, "loss": 0.1425, "step": 6814 }, { "epoch": 3.61, "learning_rate": 3.7988656280677804e-06, "loss": 0.1336, "step": 6815 }, { "epoch": 3.61, "learning_rate": 3.796175655857105e-06, "loss": 0.1411, "step": 6816 }, { "epoch": 3.61, "learning_rate": 3.793486413227344e-06, "loss": 0.0678, "step": 6817 }, { "epoch": 3.61, "learning_rate": 3.790797900494767e-06, "loss": 0.1336, "step": 6818 }, { "epoch": 3.61, "learning_rate": 3.788110117975539e-06, "loss": 0.0686, "step": 6819 }, { "epoch": 3.61, "learning_rate": 3.785423065985755e-06, "loss": 0.1103, "step": 6820 }, { "epoch": 3.61, "learning_rate": 3.7827367448414155e-06, "loss": 0.0986, "step": 6821 }, { "epoch": 3.61, "learning_rate": 3.780051154858432e-06, "loss": 0.095, "step": 6822 }, { "epoch": 3.61, "learning_rate": 3.7773662963526416e-06, "loss": 0.0463, "step": 6823 }, { "epoch": 3.61, "learning_rate": 3.774682169639782e-06, "loss": 0.1113, "step": 6824 }, { "epoch": 3.61, "learning_rate": 3.771998775035518e-06, "loss": 0.0852, "step": 6825 }, { "epoch": 3.61, "learning_rate": 3.7693161128554144e-06, "loss": 0.1138, "step": 6826 }, { "epoch": 3.61, "learning_rate": 3.766634183414961e-06, "loss": 0.1119, "step": 6827 }, { "epoch": 3.61, "learning_rate": 3.763952987029562e-06, "loss": 0.0998, "step": 6828 }, { "epoch": 3.61, "learning_rate": 3.7612725240145255e-06, "loss": 0.1028, "step": 6829 }, { "epoch": 3.62, "learning_rate": 3.7585927946850762e-06, "loss": 0.0667, "step": 6830 }, { "epoch": 3.62, "learning_rate": 3.755913799356362e-06, "loss": 0.1131, "step": 6831 }, { "epoch": 3.62, "learning_rate": 3.7532355383434315e-06, "loss": 0.1005, "step": 6832 }, { "epoch": 3.62, "learning_rate": 3.7505580119612583e-06, "loss": 0.1062, "step": 6833 }, { "epoch": 3.62, "learning_rate": 3.7478812205247185e-06, "loss": 0.1613, "step": 6834 }, { "epoch": 3.62, "learning_rate": 3.7452051643486145e-06, "loss": 0.0913, "step": 6835 }, { "epoch": 3.62, "learning_rate": 3.7425298437476477e-06, "loss": 0.1047, "step": 6836 }, { "epoch": 3.62, "learning_rate": 3.739855259036448e-06, "loss": 0.0601, "step": 6837 }, { "epoch": 3.62, "learning_rate": 3.7371814105295477e-06, "loss": 0.0795, "step": 6838 }, { "epoch": 3.62, "learning_rate": 3.734508298541394e-06, "loss": 0.1358, "step": 6839 }, { "epoch": 3.62, "learning_rate": 3.7318359233863546e-06, "loss": 0.0917, "step": 6840 }, { "epoch": 3.62, "learning_rate": 3.7291642853786993e-06, "loss": 0.0982, "step": 6841 }, { "epoch": 3.62, "learning_rate": 3.7264933848326223e-06, "loss": 0.1188, "step": 6842 }, { "epoch": 3.62, "learning_rate": 3.7238232220622285e-06, "loss": 0.0711, "step": 6843 }, { "epoch": 3.62, "learning_rate": 3.7211537973815283e-06, "loss": 0.0965, "step": 6844 }, { "epoch": 3.62, "learning_rate": 3.7184851111044586e-06, "loss": 0.1237, "step": 6845 }, { "epoch": 3.62, "learning_rate": 3.7158171635448503e-06, "loss": 0.081, "step": 6846 }, { "epoch": 3.62, "learning_rate": 3.713149955016465e-06, "loss": 0.0845, "step": 6847 }, { "epoch": 3.62, "learning_rate": 3.7104834858329743e-06, "loss": 0.1193, "step": 6848 }, { "epoch": 3.63, "learning_rate": 3.707817756307953e-06, "loss": 0.0854, "step": 6849 }, { "epoch": 3.63, "learning_rate": 3.705152766754904e-06, "loss": 0.0782, "step": 6850 }, { "epoch": 3.63, "learning_rate": 3.702488517487226e-06, "loss": 0.1001, "step": 6851 }, { "epoch": 3.63, "learning_rate": 3.6998250088182476e-06, "loss": 0.0878, "step": 6852 }, { "epoch": 3.63, "learning_rate": 3.6971622410611974e-06, "loss": 0.0892, "step": 6853 }, { "epoch": 3.63, "learning_rate": 3.6945002145292195e-06, "loss": 0.0742, "step": 6854 }, { "epoch": 3.63, "learning_rate": 3.6918389295353808e-06, "loss": 0.0757, "step": 6855 }, { "epoch": 3.63, "learning_rate": 3.6891783863926433e-06, "loss": 0.0782, "step": 6856 }, { "epoch": 3.63, "learning_rate": 3.6865185854138986e-06, "loss": 0.107, "step": 6857 }, { "epoch": 3.63, "learning_rate": 3.683859526911945e-06, "loss": 0.0913, "step": 6858 }, { "epoch": 3.63, "learning_rate": 3.6812012111994867e-06, "loss": 0.0851, "step": 6859 }, { "epoch": 3.63, "learning_rate": 3.678543638589154e-06, "loss": 0.0851, "step": 6860 }, { "epoch": 3.63, "learning_rate": 3.675886809393476e-06, "loss": 0.0677, "step": 6861 }, { "epoch": 3.63, "learning_rate": 3.673230723924901e-06, "loss": 0.1292, "step": 6862 }, { "epoch": 3.63, "learning_rate": 3.6705753824957933e-06, "loss": 0.0671, "step": 6863 }, { "epoch": 3.63, "learning_rate": 3.6679207854184207e-06, "loss": 0.1097, "step": 6864 }, { "epoch": 3.63, "learning_rate": 3.665266933004974e-06, "loss": 0.1045, "step": 6865 }, { "epoch": 3.63, "learning_rate": 3.662613825567546e-06, "loss": 0.1256, "step": 6866 }, { "epoch": 3.63, "learning_rate": 3.6599614634181524e-06, "loss": 0.0906, "step": 6867 }, { "epoch": 3.64, "learning_rate": 3.6573098468687094e-06, "loss": 0.0791, "step": 6868 }, { "epoch": 3.64, "learning_rate": 3.654658976231058e-06, "loss": 0.1045, "step": 6869 }, { "epoch": 3.64, "learning_rate": 3.652008851816943e-06, "loss": 0.0728, "step": 6870 }, { "epoch": 3.64, "learning_rate": 3.64935947393802e-06, "loss": 0.0942, "step": 6871 }, { "epoch": 3.64, "learning_rate": 3.6467108429058673e-06, "loss": 0.0906, "step": 6872 }, { "epoch": 3.64, "learning_rate": 3.644062959031962e-06, "loss": 0.1248, "step": 6873 }, { "epoch": 3.64, "learning_rate": 3.641415822627704e-06, "loss": 0.0635, "step": 6874 }, { "epoch": 3.64, "learning_rate": 3.6387694340044034e-06, "loss": 0.1106, "step": 6875 }, { "epoch": 3.64, "learning_rate": 3.636123793473274e-06, "loss": 0.0719, "step": 6876 }, { "epoch": 3.64, "learning_rate": 3.6334789013454552e-06, "loss": 0.0904, "step": 6877 }, { "epoch": 3.64, "learning_rate": 3.630834757931988e-06, "loss": 0.0946, "step": 6878 }, { "epoch": 3.64, "learning_rate": 3.628191363543824e-06, "loss": 0.0693, "step": 6879 }, { "epoch": 3.64, "learning_rate": 3.625548718491837e-06, "loss": 0.1544, "step": 6880 }, { "epoch": 3.64, "learning_rate": 3.6229068230868026e-06, "loss": 0.1196, "step": 6881 }, { "epoch": 3.64, "learning_rate": 3.6202656776394173e-06, "loss": 0.0835, "step": 6882 }, { "epoch": 3.64, "learning_rate": 3.6176252824602785e-06, "loss": 0.1041, "step": 6883 }, { "epoch": 3.64, "learning_rate": 3.6149856378599057e-06, "loss": 0.1527, "step": 6884 }, { "epoch": 3.64, "learning_rate": 3.6123467441487312e-06, "loss": 0.1066, "step": 6885 }, { "epoch": 3.64, "learning_rate": 3.6097086016370807e-06, "loss": 0.1415, "step": 6886 }, { "epoch": 3.65, "learning_rate": 3.6070712106352134e-06, "loss": 0.1137, "step": 6887 }, { "epoch": 3.65, "learning_rate": 3.604434571453287e-06, "loss": 0.0797, "step": 6888 }, { "epoch": 3.65, "learning_rate": 3.6017986844013774e-06, "loss": 0.1157, "step": 6889 }, { "epoch": 3.65, "learning_rate": 3.5991635497894716e-06, "loss": 0.1372, "step": 6890 }, { "epoch": 3.65, "learning_rate": 3.5965291679274607e-06, "loss": 0.0853, "step": 6891 }, { "epoch": 3.65, "learning_rate": 3.5938955391251597e-06, "loss": 0.0847, "step": 6892 }, { "epoch": 3.65, "learning_rate": 3.5912626636922843e-06, "loss": 0.1599, "step": 6893 }, { "epoch": 3.65, "learning_rate": 3.5886305419384617e-06, "loss": 0.091, "step": 6894 }, { "epoch": 3.65, "learning_rate": 3.5859991741732413e-06, "loss": 0.136, "step": 6895 }, { "epoch": 3.65, "learning_rate": 3.5833685607060707e-06, "loss": 0.0738, "step": 6896 }, { "epoch": 3.65, "learning_rate": 3.58073870184632e-06, "loss": 0.0674, "step": 6897 }, { "epoch": 3.65, "learning_rate": 3.57810959790326e-06, "loss": 0.1065, "step": 6898 }, { "epoch": 3.65, "learning_rate": 3.575481249186079e-06, "loss": 0.1088, "step": 6899 }, { "epoch": 3.65, "learning_rate": 3.572853656003883e-06, "loss": 0.0648, "step": 6900 }, { "epoch": 3.65, "learning_rate": 3.570226818665674e-06, "loss": 0.1003, "step": 6901 }, { "epoch": 3.65, "learning_rate": 3.567600737480377e-06, "loss": 0.0799, "step": 6902 }, { "epoch": 3.65, "learning_rate": 3.564975412756816e-06, "loss": 0.1135, "step": 6903 }, { "epoch": 3.65, "learning_rate": 3.5623508448037415e-06, "loss": 0.0989, "step": 6904 }, { "epoch": 3.65, "learning_rate": 3.5597270339298072e-06, "loss": 0.1163, "step": 6905 }, { "epoch": 3.66, "learning_rate": 3.5571039804435736e-06, "loss": 0.1358, "step": 6906 }, { "epoch": 3.66, "learning_rate": 3.5544816846535237e-06, "loss": 0.1136, "step": 6907 }, { "epoch": 3.66, "learning_rate": 3.5518601468680346e-06, "loss": 0.1043, "step": 6908 }, { "epoch": 3.66, "learning_rate": 3.549239367395414e-06, "loss": 0.0561, "step": 6909 }, { "epoch": 3.66, "learning_rate": 3.5466193465438648e-06, "loss": 0.1393, "step": 6910 }, { "epoch": 3.66, "learning_rate": 3.544000084621503e-06, "loss": 0.0738, "step": 6911 }, { "epoch": 3.66, "learning_rate": 3.541381581936365e-06, "loss": 0.094, "step": 6912 }, { "epoch": 3.66, "learning_rate": 3.538763838796386e-06, "loss": 0.0698, "step": 6913 }, { "epoch": 3.66, "learning_rate": 3.5361468555094227e-06, "loss": 0.1476, "step": 6914 }, { "epoch": 3.66, "learning_rate": 3.533530632383232e-06, "loss": 0.089, "step": 6915 }, { "epoch": 3.66, "learning_rate": 3.5309151697254873e-06, "loss": 0.1224, "step": 6916 }, { "epoch": 3.66, "learning_rate": 3.5283004678437806e-06, "loss": 0.0973, "step": 6917 }, { "epoch": 3.66, "learning_rate": 3.525686527045592e-06, "loss": 0.0836, "step": 6918 }, { "epoch": 3.66, "learning_rate": 3.5230733476383306e-06, "loss": 0.0865, "step": 6919 }, { "epoch": 3.66, "learning_rate": 3.520460929929316e-06, "loss": 0.086, "step": 6920 }, { "epoch": 3.66, "learning_rate": 3.5178492742257663e-06, "loss": 0.1008, "step": 6921 }, { "epoch": 3.66, "learning_rate": 3.5152383808348235e-06, "loss": 0.0707, "step": 6922 }, { "epoch": 3.66, "learning_rate": 3.5126282500635256e-06, "loss": 0.1113, "step": 6923 }, { "epoch": 3.66, "learning_rate": 3.5100188822188364e-06, "loss": 0.0992, "step": 6924 }, { "epoch": 3.67, "learning_rate": 3.507410277607619e-06, "loss": 0.0613, "step": 6925 }, { "epoch": 3.67, "learning_rate": 3.5048024365366473e-06, "loss": 0.0535, "step": 6926 }, { "epoch": 3.67, "learning_rate": 3.5021953593126135e-06, "loss": 0.1197, "step": 6927 }, { "epoch": 3.67, "learning_rate": 3.4995890462421076e-06, "loss": 0.0857, "step": 6928 }, { "epoch": 3.67, "learning_rate": 3.496983497631645e-06, "loss": 0.0832, "step": 6929 }, { "epoch": 3.67, "learning_rate": 3.494378713787635e-06, "loss": 0.0426, "step": 6930 }, { "epoch": 3.67, "learning_rate": 3.4917746950164087e-06, "loss": 0.119, "step": 6931 }, { "epoch": 3.67, "learning_rate": 3.489171441624206e-06, "loss": 0.072, "step": 6932 }, { "epoch": 3.67, "learning_rate": 3.486568953917172e-06, "loss": 0.1094, "step": 6933 }, { "epoch": 3.67, "learning_rate": 3.4839672322013606e-06, "loss": 0.0809, "step": 6934 }, { "epoch": 3.67, "learning_rate": 3.481366276782745e-06, "loss": 0.1195, "step": 6935 }, { "epoch": 3.67, "learning_rate": 3.4787660879671947e-06, "loss": 0.1149, "step": 6936 }, { "epoch": 3.67, "learning_rate": 3.4761666660605055e-06, "loss": 0.1421, "step": 6937 }, { "epoch": 3.67, "learning_rate": 3.4735680113683655e-06, "loss": 0.111, "step": 6938 }, { "epoch": 3.67, "learning_rate": 3.4709701241963888e-06, "loss": 0.1215, "step": 6939 }, { "epoch": 3.67, "learning_rate": 3.468373004850084e-06, "loss": 0.1209, "step": 6940 }, { "epoch": 3.67, "learning_rate": 3.465776653634885e-06, "loss": 0.0815, "step": 6941 }, { "epoch": 3.67, "learning_rate": 3.463181070856123e-06, "loss": 0.0986, "step": 6942 }, { "epoch": 3.68, "learning_rate": 3.4605862568190407e-06, "loss": 0.0936, "step": 6943 }, { "epoch": 3.68, "learning_rate": 3.457992211828799e-06, "loss": 0.0979, "step": 6944 }, { "epoch": 3.68, "learning_rate": 3.455398936190455e-06, "loss": 0.0872, "step": 6945 }, { "epoch": 3.68, "learning_rate": 3.4528064302089868e-06, "loss": 0.0535, "step": 6946 }, { "epoch": 3.68, "learning_rate": 3.4502146941892824e-06, "loss": 0.0786, "step": 6947 }, { "epoch": 3.68, "learning_rate": 3.4476237284361258e-06, "loss": 0.1142, "step": 6948 }, { "epoch": 3.68, "learning_rate": 3.4450335332542264e-06, "loss": 0.0945, "step": 6949 }, { "epoch": 3.68, "learning_rate": 3.4424441089481917e-06, "loss": 0.1043, "step": 6950 }, { "epoch": 3.68, "learning_rate": 3.4398554558225416e-06, "loss": 0.0974, "step": 6951 }, { "epoch": 3.68, "learning_rate": 3.437267574181712e-06, "loss": 0.123, "step": 6952 }, { "epoch": 3.68, "learning_rate": 3.4346804643300347e-06, "loss": 0.0642, "step": 6953 }, { "epoch": 3.68, "learning_rate": 3.4320941265717665e-06, "loss": 0.1187, "step": 6954 }, { "epoch": 3.68, "learning_rate": 3.4295085612110577e-06, "loss": 0.1125, "step": 6955 }, { "epoch": 3.68, "learning_rate": 3.4269237685519843e-06, "loss": 0.0888, "step": 6956 }, { "epoch": 3.68, "learning_rate": 3.424339748898512e-06, "loss": 0.0997, "step": 6957 }, { "epoch": 3.68, "learning_rate": 3.4217565025545363e-06, "loss": 0.0742, "step": 6958 }, { "epoch": 3.68, "learning_rate": 3.419174029823846e-06, "loss": 0.0979, "step": 6959 }, { "epoch": 3.68, "learning_rate": 3.4165923310101435e-06, "loss": 0.0893, "step": 6960 }, { "epoch": 3.68, "learning_rate": 3.414011406417044e-06, "loss": 0.0856, "step": 6961 }, { "epoch": 3.69, "learning_rate": 3.41143125634807e-06, "loss": 0.1126, "step": 6962 }, { "epoch": 3.69, "learning_rate": 3.4088518811066473e-06, "loss": 0.0715, "step": 6963 }, { "epoch": 3.69, "learning_rate": 3.406273280996121e-06, "loss": 0.0928, "step": 6964 }, { "epoch": 3.69, "learning_rate": 3.4036954563197322e-06, "loss": 0.0708, "step": 6965 }, { "epoch": 3.69, "learning_rate": 3.4011184073806446e-06, "loss": 0.1052, "step": 6966 }, { "epoch": 3.69, "learning_rate": 3.3985421344819214e-06, "loss": 0.0514, "step": 6967 }, { "epoch": 3.69, "learning_rate": 3.395966637926533e-06, "loss": 0.0725, "step": 6968 }, { "epoch": 3.69, "learning_rate": 3.393391918017369e-06, "loss": 0.0809, "step": 6969 }, { "epoch": 3.69, "learning_rate": 3.3908179750572136e-06, "loss": 0.0748, "step": 6970 }, { "epoch": 3.69, "learning_rate": 3.3882448093487754e-06, "loss": 0.0823, "step": 6971 }, { "epoch": 3.69, "learning_rate": 3.3856724211946544e-06, "loss": 0.0992, "step": 6972 }, { "epoch": 3.69, "learning_rate": 3.383100810897374e-06, "loss": 0.0811, "step": 6973 }, { "epoch": 3.69, "learning_rate": 3.380529978759366e-06, "loss": 0.054, "step": 6974 }, { "epoch": 3.69, "learning_rate": 3.37795992508295e-06, "loss": 0.0915, "step": 6975 }, { "epoch": 3.69, "learning_rate": 3.3753906501703816e-06, "loss": 0.0897, "step": 6976 }, { "epoch": 3.69, "learning_rate": 3.3728221543238027e-06, "loss": 0.0991, "step": 6977 }, { "epoch": 3.69, "learning_rate": 3.3702544378452783e-06, "loss": 0.0946, "step": 6978 }, { "epoch": 3.69, "learning_rate": 3.3676875010367805e-06, "loss": 0.1603, "step": 6979 }, { "epoch": 3.69, "learning_rate": 3.365121344200176e-06, "loss": 0.0917, "step": 6980 }, { "epoch": 3.7, "learning_rate": 3.3625559676372587e-06, "loss": 0.1074, "step": 6981 }, { "epoch": 3.7, "learning_rate": 3.359991371649718e-06, "loss": 0.0851, "step": 6982 }, { "epoch": 3.7, "learning_rate": 3.3574275565391503e-06, "loss": 0.1424, "step": 6983 }, { "epoch": 3.7, "learning_rate": 3.3548645226070733e-06, "loss": 0.0869, "step": 6984 }, { "epoch": 3.7, "learning_rate": 3.3523022701548957e-06, "loss": 0.0881, "step": 6985 }, { "epoch": 3.7, "learning_rate": 3.3497407994839516e-06, "loss": 0.0749, "step": 6986 }, { "epoch": 3.7, "learning_rate": 3.347180110895466e-06, "loss": 0.0792, "step": 6987 }, { "epoch": 3.7, "learning_rate": 3.3446202046905853e-06, "loss": 0.1046, "step": 6988 }, { "epoch": 3.7, "learning_rate": 3.3420610811703624e-06, "loss": 0.0905, "step": 6989 }, { "epoch": 3.7, "learning_rate": 3.3395027406357505e-06, "loss": 0.0718, "step": 6990 }, { "epoch": 3.7, "learning_rate": 3.3369451833876165e-06, "loss": 0.1263, "step": 6991 }, { "epoch": 3.7, "learning_rate": 3.334388409726729e-06, "loss": 0.0751, "step": 6992 }, { "epoch": 3.7, "learning_rate": 3.331832419953773e-06, "loss": 0.0931, "step": 6993 }, { "epoch": 3.7, "learning_rate": 3.329277214369341e-06, "loss": 0.0944, "step": 6994 }, { "epoch": 3.7, "learning_rate": 3.326722793273923e-06, "loss": 0.0883, "step": 6995 }, { "epoch": 3.7, "learning_rate": 3.32416915696793e-06, "loss": 0.1337, "step": 6996 }, { "epoch": 3.7, "learning_rate": 3.3216163057516683e-06, "loss": 0.101, "step": 6997 }, { "epoch": 3.7, "learning_rate": 3.3190642399253637e-06, "loss": 0.0612, "step": 6998 }, { "epoch": 3.7, "learning_rate": 3.316512959789141e-06, "loss": 0.0809, "step": 6999 }, { "epoch": 3.71, "learning_rate": 3.3139624656430315e-06, "loss": 0.104, "step": 7000 }, { "epoch": 3.71, "learning_rate": 3.311412757786985e-06, "loss": 0.136, "step": 7001 }, { "epoch": 3.71, "learning_rate": 3.308863836520844e-06, "loss": 0.1261, "step": 7002 }, { "epoch": 3.71, "learning_rate": 3.3063157021443704e-06, "loss": 0.0795, "step": 7003 }, { "epoch": 3.71, "learning_rate": 3.303768354957233e-06, "loss": 0.1258, "step": 7004 }, { "epoch": 3.71, "learning_rate": 3.3012217952589975e-06, "loss": 0.1116, "step": 7005 }, { "epoch": 3.71, "learning_rate": 3.2986760233491544e-06, "loss": 0.075, "step": 7006 }, { "epoch": 3.71, "learning_rate": 3.296131039527076e-06, "loss": 0.0407, "step": 7007 }, { "epoch": 3.71, "learning_rate": 3.293586844092066e-06, "loss": 0.0956, "step": 7008 }, { "epoch": 3.71, "learning_rate": 3.2910434373433277e-06, "loss": 0.1664, "step": 7009 }, { "epoch": 3.71, "learning_rate": 3.288500819579965e-06, "loss": 0.0838, "step": 7010 }, { "epoch": 3.71, "learning_rate": 3.285958991101e-06, "loss": 0.0983, "step": 7011 }, { "epoch": 3.71, "learning_rate": 3.283417952205351e-06, "loss": 0.0807, "step": 7012 }, { "epoch": 3.71, "learning_rate": 3.2808777031918537e-06, "loss": 0.1083, "step": 7013 }, { "epoch": 3.71, "learning_rate": 3.278338244359244e-06, "loss": 0.0783, "step": 7014 }, { "epoch": 3.71, "learning_rate": 3.2757995760061624e-06, "loss": 0.0676, "step": 7015 }, { "epoch": 3.71, "learning_rate": 3.2732616984311684e-06, "loss": 0.1311, "step": 7016 }, { "epoch": 3.71, "learning_rate": 3.270724611932714e-06, "loss": 0.1033, "step": 7017 }, { "epoch": 3.71, "learning_rate": 3.268188316809172e-06, "loss": 0.0655, "step": 7018 }, { "epoch": 3.72, "learning_rate": 3.2656528133588084e-06, "loss": 0.0755, "step": 7019 }, { "epoch": 3.72, "learning_rate": 3.263118101879805e-06, "loss": 0.0756, "step": 7020 }, { "epoch": 3.72, "learning_rate": 3.2605841826702557e-06, "loss": 0.1183, "step": 7021 }, { "epoch": 3.72, "learning_rate": 3.2580510560281466e-06, "loss": 0.1198, "step": 7022 }, { "epoch": 3.72, "learning_rate": 3.255518722251375e-06, "loss": 0.0804, "step": 7023 }, { "epoch": 3.72, "learning_rate": 3.252987181637758e-06, "loss": 0.1054, "step": 7024 }, { "epoch": 3.72, "learning_rate": 3.2504564344849986e-06, "loss": 0.1878, "step": 7025 }, { "epoch": 3.72, "learning_rate": 3.2479264810907262e-06, "loss": 0.0895, "step": 7026 }, { "epoch": 3.72, "learning_rate": 3.2453973217524602e-06, "loss": 0.0692, "step": 7027 }, { "epoch": 3.72, "learning_rate": 3.2428689567676418e-06, "loss": 0.1087, "step": 7028 }, { "epoch": 3.72, "learning_rate": 3.2403413864336043e-06, "loss": 0.1383, "step": 7029 }, { "epoch": 3.72, "learning_rate": 3.2378146110476006e-06, "loss": 0.0469, "step": 7030 }, { "epoch": 3.72, "learning_rate": 3.235288630906782e-06, "loss": 0.0819, "step": 7031 }, { "epoch": 3.72, "learning_rate": 3.2327634463082035e-06, "loss": 0.1018, "step": 7032 }, { "epoch": 3.72, "learning_rate": 3.2302390575488385e-06, "loss": 0.1117, "step": 7033 }, { "epoch": 3.72, "learning_rate": 3.227715464925554e-06, "loss": 0.0916, "step": 7034 }, { "epoch": 3.72, "learning_rate": 3.2251926687351307e-06, "loss": 0.1168, "step": 7035 }, { "epoch": 3.72, "learning_rate": 3.222670669274258e-06, "loss": 0.156, "step": 7036 }, { "epoch": 3.72, "learning_rate": 3.2201494668395227e-06, "loss": 0.0735, "step": 7037 }, { "epoch": 3.73, "learning_rate": 3.2176290617274266e-06, "loss": 0.1042, "step": 7038 }, { "epoch": 3.73, "learning_rate": 3.215109454234372e-06, "loss": 0.0751, "step": 7039 }, { "epoch": 3.73, "learning_rate": 3.2125906446566656e-06, "loss": 0.1147, "step": 7040 }, { "epoch": 3.73, "learning_rate": 3.2100726332905307e-06, "loss": 0.0805, "step": 7041 }, { "epoch": 3.73, "learning_rate": 3.2075554204320834e-06, "loss": 0.1372, "step": 7042 }, { "epoch": 3.73, "learning_rate": 3.2050390063773583e-06, "loss": 0.1024, "step": 7043 }, { "epoch": 3.73, "learning_rate": 3.2025233914222854e-06, "loss": 0.0901, "step": 7044 }, { "epoch": 3.73, "learning_rate": 3.2000085758627074e-06, "loss": 0.0901, "step": 7045 }, { "epoch": 3.73, "learning_rate": 3.1974945599943787e-06, "loss": 0.0715, "step": 7046 }, { "epoch": 3.73, "learning_rate": 3.1949813441129374e-06, "loss": 0.0893, "step": 7047 }, { "epoch": 3.73, "learning_rate": 3.192468928513953e-06, "loss": 0.0678, "step": 7048 }, { "epoch": 3.73, "learning_rate": 3.1899573134928842e-06, "loss": 0.0614, "step": 7049 }, { "epoch": 3.73, "learning_rate": 3.1874464993451036e-06, "loss": 0.1045, "step": 7050 }, { "epoch": 3.73, "learning_rate": 3.1849364863658918e-06, "loss": 0.1583, "step": 7051 }, { "epoch": 3.73, "learning_rate": 3.182427274850424e-06, "loss": 0.0926, "step": 7052 }, { "epoch": 3.73, "learning_rate": 3.1799188650937928e-06, "loss": 0.073, "step": 7053 }, { "epoch": 3.73, "learning_rate": 3.177411257390991e-06, "loss": 0.0756, "step": 7054 }, { "epoch": 3.73, "learning_rate": 3.174904452036913e-06, "loss": 0.1381, "step": 7055 }, { "epoch": 3.73, "learning_rate": 3.1723984493263716e-06, "loss": 0.1345, "step": 7056 }, { "epoch": 3.74, "learning_rate": 3.169893249554069e-06, "loss": 0.1276, "step": 7057 }, { "epoch": 3.74, "learning_rate": 3.167388853014628e-06, "loss": 0.1014, "step": 7058 }, { "epoch": 3.74, "learning_rate": 3.1648852600025647e-06, "loss": 0.08, "step": 7059 }, { "epoch": 3.74, "learning_rate": 3.162382470812312e-06, "loss": 0.1252, "step": 7060 }, { "epoch": 3.74, "learning_rate": 3.159880485738196e-06, "loss": 0.1127, "step": 7061 }, { "epoch": 3.74, "learning_rate": 3.1573793050744574e-06, "loss": 0.1155, "step": 7062 }, { "epoch": 3.74, "learning_rate": 3.154878929115246e-06, "loss": 0.0848, "step": 7063 }, { "epoch": 3.74, "learning_rate": 3.152379358154599e-06, "loss": 0.1004, "step": 7064 }, { "epoch": 3.74, "learning_rate": 3.149880592486475e-06, "loss": 0.0605, "step": 7065 }, { "epoch": 3.74, "learning_rate": 3.1473826324047384e-06, "loss": 0.1349, "step": 7066 }, { "epoch": 3.74, "learning_rate": 3.144885478203146e-06, "loss": 0.059, "step": 7067 }, { "epoch": 3.74, "learning_rate": 3.142389130175375e-06, "loss": 0.0675, "step": 7068 }, { "epoch": 3.74, "learning_rate": 3.1398935886149928e-06, "loss": 0.1179, "step": 7069 }, { "epoch": 3.74, "learning_rate": 3.1373988538154875e-06, "loss": 0.1326, "step": 7070 }, { "epoch": 3.74, "learning_rate": 3.1349049260702403e-06, "loss": 0.0565, "step": 7071 }, { "epoch": 3.74, "learning_rate": 3.1324118056725385e-06, "loss": 0.1035, "step": 7072 }, { "epoch": 3.74, "learning_rate": 3.1299194929155828e-06, "loss": 0.0735, "step": 7073 }, { "epoch": 3.74, "learning_rate": 3.1274279880924695e-06, "loss": 0.0925, "step": 7074 }, { "epoch": 3.74, "learning_rate": 3.1249372914962095e-06, "loss": 0.1504, "step": 7075 }, { "epoch": 3.75, "learning_rate": 3.122447403419707e-06, "loss": 0.108, "step": 7076 }, { "epoch": 3.75, "learning_rate": 3.1199583241557797e-06, "loss": 0.1089, "step": 7077 }, { "epoch": 3.75, "learning_rate": 3.1174700539971525e-06, "loss": 0.0712, "step": 7078 }, { "epoch": 3.75, "learning_rate": 3.114982593236445e-06, "loss": 0.1182, "step": 7079 }, { "epoch": 3.75, "learning_rate": 3.112495942166186e-06, "loss": 0.1058, "step": 7080 }, { "epoch": 3.75, "learning_rate": 3.110010101078815e-06, "loss": 0.1037, "step": 7081 }, { "epoch": 3.75, "learning_rate": 3.107525070266666e-06, "loss": 0.096, "step": 7082 }, { "epoch": 3.75, "learning_rate": 3.10504085002199e-06, "loss": 0.1419, "step": 7083 }, { "epoch": 3.75, "learning_rate": 3.1025574406369276e-06, "loss": 0.1197, "step": 7084 }, { "epoch": 3.75, "learning_rate": 3.10007484240354e-06, "loss": 0.1013, "step": 7085 }, { "epoch": 3.75, "learning_rate": 3.097593055613778e-06, "loss": 0.0887, "step": 7086 }, { "epoch": 3.75, "learning_rate": 3.0951120805595114e-06, "loss": 0.0589, "step": 7087 }, { "epoch": 3.75, "learning_rate": 3.0926319175325037e-06, "loss": 0.0823, "step": 7088 }, { "epoch": 3.75, "learning_rate": 3.0901525668244237e-06, "loss": 0.0823, "step": 7089 }, { "epoch": 3.75, "learning_rate": 3.0876740287268537e-06, "loss": 0.1074, "step": 7090 }, { "epoch": 3.75, "learning_rate": 3.085196303531267e-06, "loss": 0.0925, "step": 7091 }, { "epoch": 3.75, "learning_rate": 3.0827193915290533e-06, "loss": 0.1378, "step": 7092 }, { "epoch": 3.75, "learning_rate": 3.0802432930115035e-06, "loss": 0.097, "step": 7093 }, { "epoch": 3.75, "learning_rate": 3.077768008269806e-06, "loss": 0.0745, "step": 7094 }, { "epoch": 3.76, "learning_rate": 3.0752935375950688e-06, "loss": 0.0989, "step": 7095 }, { "epoch": 3.76, "learning_rate": 3.0728198812782796e-06, "loss": 0.0793, "step": 7096 }, { "epoch": 3.76, "learning_rate": 3.0703470396103517e-06, "loss": 0.0823, "step": 7097 }, { "epoch": 3.76, "learning_rate": 3.0678750128821e-06, "loss": 0.0709, "step": 7098 }, { "epoch": 3.76, "learning_rate": 3.0654038013842324e-06, "loss": 0.1033, "step": 7099 }, { "epoch": 3.76, "learning_rate": 3.0629334054073744e-06, "loss": 0.0822, "step": 7100 }, { "epoch": 3.76, "learning_rate": 3.0604638252420415e-06, "loss": 0.0884, "step": 7101 }, { "epoch": 3.76, "learning_rate": 3.057995061178669e-06, "loss": 0.0911, "step": 7102 }, { "epoch": 3.76, "learning_rate": 3.055527113507584e-06, "loss": 0.0756, "step": 7103 }, { "epoch": 3.76, "learning_rate": 3.0530599825190164e-06, "loss": 0.1367, "step": 7104 }, { "epoch": 3.76, "learning_rate": 3.0505936685031147e-06, "loss": 0.1319, "step": 7105 }, { "epoch": 3.76, "learning_rate": 3.0481281717499146e-06, "loss": 0.1065, "step": 7106 }, { "epoch": 3.76, "learning_rate": 3.0456634925493657e-06, "loss": 0.0823, "step": 7107 }, { "epoch": 3.76, "learning_rate": 3.0431996311913214e-06, "loss": 0.1084, "step": 7108 }, { "epoch": 3.76, "learning_rate": 3.0407365879655305e-06, "loss": 0.0566, "step": 7109 }, { "epoch": 3.76, "learning_rate": 3.0382743631616576e-06, "loss": 0.097, "step": 7110 }, { "epoch": 3.76, "learning_rate": 3.035812957069262e-06, "loss": 0.1201, "step": 7111 }, { "epoch": 3.76, "learning_rate": 3.0333523699778055e-06, "loss": 0.153, "step": 7112 }, { "epoch": 3.76, "learning_rate": 3.0308926021766637e-06, "loss": 0.0809, "step": 7113 }, { "epoch": 3.77, "learning_rate": 3.028433653955105e-06, "loss": 0.0931, "step": 7114 }, { "epoch": 3.77, "learning_rate": 3.0259755256023115e-06, "loss": 0.13, "step": 7115 }, { "epoch": 3.77, "learning_rate": 3.0235182174073575e-06, "loss": 0.0954, "step": 7116 }, { "epoch": 3.77, "learning_rate": 3.0210617296592328e-06, "loss": 0.1136, "step": 7117 }, { "epoch": 3.77, "learning_rate": 3.0186060626468185e-06, "loss": 0.0939, "step": 7118 }, { "epoch": 3.77, "learning_rate": 3.0161512166589113e-06, "loss": 0.0779, "step": 7119 }, { "epoch": 3.77, "learning_rate": 3.013697191984203e-06, "loss": 0.0916, "step": 7120 }, { "epoch": 3.77, "learning_rate": 3.011243988911289e-06, "loss": 0.1161, "step": 7121 }, { "epoch": 3.77, "learning_rate": 3.008791607728675e-06, "loss": 0.098, "step": 7122 }, { "epoch": 3.77, "learning_rate": 3.0063400487247594e-06, "loss": 0.058, "step": 7123 }, { "epoch": 3.77, "learning_rate": 3.0038893121878545e-06, "loss": 0.0546, "step": 7124 }, { "epoch": 3.77, "learning_rate": 3.001439398406174e-06, "loss": 0.0676, "step": 7125 }, { "epoch": 3.77, "learning_rate": 2.998990307667825e-06, "loss": 0.0944, "step": 7126 }, { "epoch": 3.77, "learning_rate": 2.9965420402608316e-06, "loss": 0.1203, "step": 7127 }, { "epoch": 3.77, "learning_rate": 2.9940945964731117e-06, "loss": 0.0889, "step": 7128 }, { "epoch": 3.77, "learning_rate": 2.9916479765924854e-06, "loss": 0.0707, "step": 7129 }, { "epoch": 3.77, "learning_rate": 2.989202180906686e-06, "loss": 0.0413, "step": 7130 }, { "epoch": 3.77, "learning_rate": 2.986757209703338e-06, "loss": 0.0691, "step": 7131 }, { "epoch": 3.78, "learning_rate": 2.9843130632699803e-06, "loss": 0.1121, "step": 7132 }, { "epoch": 3.78, "learning_rate": 2.9818697418940424e-06, "loss": 0.1137, "step": 7133 }, { "epoch": 3.78, "learning_rate": 2.979427245862866e-06, "loss": 0.0714, "step": 7134 }, { "epoch": 3.78, "learning_rate": 2.9769855754637e-06, "loss": 0.0725, "step": 7135 }, { "epoch": 3.78, "learning_rate": 2.974544730983676e-06, "loss": 0.0539, "step": 7136 }, { "epoch": 3.78, "learning_rate": 2.972104712709852e-06, "loss": 0.1559, "step": 7137 }, { "epoch": 3.78, "learning_rate": 2.9696655209291727e-06, "loss": 0.1441, "step": 7138 }, { "epoch": 3.78, "learning_rate": 2.9672271559284926e-06, "loss": 0.1315, "step": 7139 }, { "epoch": 3.78, "learning_rate": 2.9647896179945733e-06, "loss": 0.0856, "step": 7140 }, { "epoch": 3.78, "learning_rate": 2.9623529074140654e-06, "loss": 0.0339, "step": 7141 }, { "epoch": 3.78, "learning_rate": 2.9599170244735377e-06, "loss": 0.1102, "step": 7142 }, { "epoch": 3.78, "learning_rate": 2.9574819694594516e-06, "loss": 0.0901, "step": 7143 }, { "epoch": 3.78, "learning_rate": 2.955047742658169e-06, "loss": 0.0785, "step": 7144 }, { "epoch": 3.78, "learning_rate": 2.9526143443559684e-06, "loss": 0.0768, "step": 7145 }, { "epoch": 3.78, "learning_rate": 2.950181774839014e-06, "loss": 0.0965, "step": 7146 }, { "epoch": 3.78, "learning_rate": 2.9477500343933864e-06, "loss": 0.0508, "step": 7147 }, { "epoch": 3.78, "learning_rate": 2.945319123305056e-06, "loss": 0.0793, "step": 7148 }, { "epoch": 3.78, "learning_rate": 2.9428890418599063e-06, "loss": 0.1284, "step": 7149 }, { "epoch": 3.78, "learning_rate": 2.940459790343724e-06, "loss": 0.0806, "step": 7150 }, { "epoch": 3.79, "learning_rate": 2.9380313690421867e-06, "loss": 0.071, "step": 7151 }, { "epoch": 3.79, "learning_rate": 2.935603778240884e-06, "loss": 0.0685, "step": 7152 }, { "epoch": 3.79, "learning_rate": 2.9331770182253006e-06, "loss": 0.0937, "step": 7153 }, { "epoch": 3.79, "learning_rate": 2.930751089280831e-06, "loss": 0.0654, "step": 7154 }, { "epoch": 3.79, "learning_rate": 2.928325991692772e-06, "loss": 0.0982, "step": 7155 }, { "epoch": 3.79, "learning_rate": 2.925901725746314e-06, "loss": 0.0793, "step": 7156 }, { "epoch": 3.79, "learning_rate": 2.9234782917265605e-06, "loss": 0.1385, "step": 7157 }, { "epoch": 3.79, "learning_rate": 2.921055689918506e-06, "loss": 0.0686, "step": 7158 }, { "epoch": 3.79, "learning_rate": 2.918633920607058e-06, "loss": 0.1236, "step": 7159 }, { "epoch": 3.79, "learning_rate": 2.9162129840770192e-06, "loss": 0.0612, "step": 7160 }, { "epoch": 3.79, "learning_rate": 2.913792880613091e-06, "loss": 0.1088, "step": 7161 }, { "epoch": 3.79, "learning_rate": 2.91137361049989e-06, "loss": 0.1078, "step": 7162 }, { "epoch": 3.79, "learning_rate": 2.9089551740219215e-06, "loss": 0.1391, "step": 7163 }, { "epoch": 3.79, "learning_rate": 2.906537571463601e-06, "loss": 0.0776, "step": 7164 }, { "epoch": 3.79, "learning_rate": 2.904120803109238e-06, "loss": 0.0902, "step": 7165 }, { "epoch": 3.79, "learning_rate": 2.901704869243054e-06, "loss": 0.0934, "step": 7166 }, { "epoch": 3.79, "learning_rate": 2.899289770149171e-06, "loss": 0.0647, "step": 7167 }, { "epoch": 3.79, "learning_rate": 2.896875506111596e-06, "loss": 0.1341, "step": 7168 }, { "epoch": 3.79, "learning_rate": 2.8944620774142596e-06, "loss": 0.0851, "step": 7169 }, { "epoch": 3.8, "learning_rate": 2.892049484340986e-06, "loss": 0.0847, "step": 7170 }, { "epoch": 3.8, "learning_rate": 2.889637727175496e-06, "loss": 0.0971, "step": 7171 }, { "epoch": 3.8, "learning_rate": 2.887226806201423e-06, "loss": 0.115, "step": 7172 }, { "epoch": 3.8, "learning_rate": 2.884816721702288e-06, "loss": 0.0703, "step": 7173 }, { "epoch": 3.8, "learning_rate": 2.882407473961528e-06, "loss": 0.0566, "step": 7174 }, { "epoch": 3.8, "learning_rate": 2.879999063262472e-06, "loss": 0.0565, "step": 7175 }, { "epoch": 3.8, "learning_rate": 2.8775914898883497e-06, "loss": 0.1831, "step": 7176 }, { "epoch": 3.8, "learning_rate": 2.875184754122303e-06, "loss": 0.0898, "step": 7177 }, { "epoch": 3.8, "learning_rate": 2.8727788562473622e-06, "loss": 0.1053, "step": 7178 }, { "epoch": 3.8, "learning_rate": 2.8703737965464717e-06, "loss": 0.0783, "step": 7179 }, { "epoch": 3.8, "learning_rate": 2.8679695753024638e-06, "loss": 0.0927, "step": 7180 }, { "epoch": 3.8, "learning_rate": 2.865566192798083e-06, "loss": 0.0814, "step": 7181 }, { "epoch": 3.8, "learning_rate": 2.863163649315973e-06, "loss": 0.1133, "step": 7182 }, { "epoch": 3.8, "learning_rate": 2.860761945138677e-06, "loss": 0.0736, "step": 7183 }, { "epoch": 3.8, "learning_rate": 2.858361080548634e-06, "loss": 0.1024, "step": 7184 }, { "epoch": 3.8, "learning_rate": 2.8559610558281968e-06, "loss": 0.0641, "step": 7185 }, { "epoch": 3.8, "learning_rate": 2.853561871259607e-06, "loss": 0.0856, "step": 7186 }, { "epoch": 3.8, "learning_rate": 2.8511635271250194e-06, "loss": 0.0971, "step": 7187 }, { "epoch": 3.8, "learning_rate": 2.8487660237064774e-06, "loss": 0.0674, "step": 7188 }, { "epoch": 3.81, "learning_rate": 2.846369361285937e-06, "loss": 0.0876, "step": 7189 }, { "epoch": 3.81, "learning_rate": 2.843973540145245e-06, "loss": 0.0964, "step": 7190 }, { "epoch": 3.81, "learning_rate": 2.8415785605661596e-06, "loss": 0.0715, "step": 7191 }, { "epoch": 3.81, "learning_rate": 2.839184422830331e-06, "loss": 0.0727, "step": 7192 }, { "epoch": 3.81, "learning_rate": 2.8367911272193127e-06, "loss": 0.0778, "step": 7193 }, { "epoch": 3.81, "learning_rate": 2.8343986740145656e-06, "loss": 0.075, "step": 7194 }, { "epoch": 3.81, "learning_rate": 2.8320070634974405e-06, "loss": 0.1062, "step": 7195 }, { "epoch": 3.81, "learning_rate": 2.829616295949199e-06, "loss": 0.0796, "step": 7196 }, { "epoch": 3.81, "learning_rate": 2.827226371651002e-06, "loss": 0.0826, "step": 7197 }, { "epoch": 3.81, "learning_rate": 2.8248372908839026e-06, "loss": 0.1201, "step": 7198 }, { "epoch": 3.81, "learning_rate": 2.822449053928867e-06, "loss": 0.124, "step": 7199 }, { "epoch": 3.81, "learning_rate": 2.820061661066755e-06, "loss": 0.0822, "step": 7200 }, { "epoch": 3.81, "learning_rate": 2.8176751125783233e-06, "loss": 0.1036, "step": 7201 }, { "epoch": 3.81, "learning_rate": 2.8152894087442397e-06, "loss": 0.1119, "step": 7202 }, { "epoch": 3.81, "learning_rate": 2.8129045498450624e-06, "loss": 0.1246, "step": 7203 }, { "epoch": 3.81, "learning_rate": 2.810520536161262e-06, "loss": 0.0797, "step": 7204 }, { "epoch": 3.81, "learning_rate": 2.808137367973196e-06, "loss": 0.1198, "step": 7205 }, { "epoch": 3.81, "learning_rate": 2.805755045561135e-06, "loss": 0.0631, "step": 7206 }, { "epoch": 3.81, "learning_rate": 2.8033735692052376e-06, "loss": 0.1191, "step": 7207 }, { "epoch": 3.82, "learning_rate": 2.8009929391855784e-06, "loss": 0.0631, "step": 7208 }, { "epoch": 3.82, "learning_rate": 2.7986131557821173e-06, "loss": 0.0798, "step": 7209 }, { "epoch": 3.82, "learning_rate": 2.7962342192747205e-06, "loss": 0.157, "step": 7210 }, { "epoch": 3.82, "learning_rate": 2.793856129943159e-06, "loss": 0.0906, "step": 7211 }, { "epoch": 3.82, "learning_rate": 2.7914788880670994e-06, "loss": 0.1482, "step": 7212 }, { "epoch": 3.82, "learning_rate": 2.789102493926108e-06, "loss": 0.1174, "step": 7213 }, { "epoch": 3.82, "learning_rate": 2.786726947799657e-06, "loss": 0.0964, "step": 7214 }, { "epoch": 3.82, "learning_rate": 2.7843522499671095e-06, "loss": 0.081, "step": 7215 }, { "epoch": 3.82, "learning_rate": 2.781978400707739e-06, "loss": 0.125, "step": 7216 }, { "epoch": 3.82, "learning_rate": 2.779605400300712e-06, "loss": 0.1138, "step": 7217 }, { "epoch": 3.82, "learning_rate": 2.7772332490250964e-06, "loss": 0.0864, "step": 7218 }, { "epoch": 3.82, "learning_rate": 2.774861947159865e-06, "loss": 0.0635, "step": 7219 }, { "epoch": 3.82, "learning_rate": 2.7724914949838834e-06, "loss": 0.0586, "step": 7220 }, { "epoch": 3.82, "learning_rate": 2.7701218927759243e-06, "loss": 0.0818, "step": 7221 }, { "epoch": 3.82, "learning_rate": 2.767753140814653e-06, "loss": 0.1437, "step": 7222 }, { "epoch": 3.82, "learning_rate": 2.765385239378643e-06, "loss": 0.0994, "step": 7223 }, { "epoch": 3.82, "learning_rate": 2.763018188746367e-06, "loss": 0.0811, "step": 7224 }, { "epoch": 3.82, "learning_rate": 2.7606519891961846e-06, "loss": 0.1085, "step": 7225 }, { "epoch": 3.82, "learning_rate": 2.75828664100637e-06, "loss": 0.0509, "step": 7226 }, { "epoch": 3.83, "learning_rate": 2.7559221444550964e-06, "loss": 0.1115, "step": 7227 }, { "epoch": 3.83, "learning_rate": 2.753558499820427e-06, "loss": 0.0792, "step": 7228 }, { "epoch": 3.83, "learning_rate": 2.7511957073803354e-06, "loss": 0.1478, "step": 7229 }, { "epoch": 3.83, "learning_rate": 2.7488337674126842e-06, "loss": 0.1112, "step": 7230 }, { "epoch": 3.83, "learning_rate": 2.746472680195249e-06, "loss": 0.1131, "step": 7231 }, { "epoch": 3.83, "learning_rate": 2.7441124460056943e-06, "loss": 0.0796, "step": 7232 }, { "epoch": 3.83, "learning_rate": 2.7417530651215853e-06, "loss": 0.11, "step": 7233 }, { "epoch": 3.83, "learning_rate": 2.7393945378203933e-06, "loss": 0.0994, "step": 7234 }, { "epoch": 3.83, "learning_rate": 2.737036864379483e-06, "loss": 0.0489, "step": 7235 }, { "epoch": 3.83, "learning_rate": 2.7346800450761247e-06, "loss": 0.1084, "step": 7236 }, { "epoch": 3.83, "learning_rate": 2.7323240801874795e-06, "loss": 0.1027, "step": 7237 }, { "epoch": 3.83, "learning_rate": 2.7299689699906153e-06, "loss": 0.0571, "step": 7238 }, { "epoch": 3.83, "learning_rate": 2.7276147147624996e-06, "loss": 0.1501, "step": 7239 }, { "epoch": 3.83, "learning_rate": 2.7252613147799967e-06, "loss": 0.1171, "step": 7240 }, { "epoch": 3.83, "learning_rate": 2.7229087703198685e-06, "loss": 0.1382, "step": 7241 }, { "epoch": 3.83, "learning_rate": 2.7205570816587755e-06, "loss": 0.0809, "step": 7242 }, { "epoch": 3.83, "learning_rate": 2.718206249073285e-06, "loss": 0.0999, "step": 7243 }, { "epoch": 3.83, "learning_rate": 2.715856272839862e-06, "loss": 0.0799, "step": 7244 }, { "epoch": 3.83, "learning_rate": 2.71350715323486e-06, "loss": 0.1245, "step": 7245 }, { "epoch": 3.84, "learning_rate": 2.7111588905345465e-06, "loss": 0.0663, "step": 7246 }, { "epoch": 3.84, "learning_rate": 2.708811485015077e-06, "loss": 0.068, "step": 7247 }, { "epoch": 3.84, "learning_rate": 2.7064649369525154e-06, "loss": 0.1012, "step": 7248 }, { "epoch": 3.84, "learning_rate": 2.704119246622816e-06, "loss": 0.1394, "step": 7249 }, { "epoch": 3.84, "learning_rate": 2.7017744143018354e-06, "loss": 0.0701, "step": 7250 }, { "epoch": 3.84, "learning_rate": 2.6994304402653337e-06, "loss": 0.1037, "step": 7251 }, { "epoch": 3.84, "learning_rate": 2.6970873247889626e-06, "loss": 0.1048, "step": 7252 }, { "epoch": 3.84, "learning_rate": 2.694745068148278e-06, "loss": 0.1286, "step": 7253 }, { "epoch": 3.84, "learning_rate": 2.6924036706187383e-06, "loss": 0.1058, "step": 7254 }, { "epoch": 3.84, "learning_rate": 2.6900631324756886e-06, "loss": 0.0848, "step": 7255 }, { "epoch": 3.84, "learning_rate": 2.6877234539943895e-06, "loss": 0.1276, "step": 7256 }, { "epoch": 3.84, "learning_rate": 2.6853846354499803e-06, "loss": 0.0952, "step": 7257 }, { "epoch": 3.84, "learning_rate": 2.6830466771175154e-06, "loss": 0.0613, "step": 7258 }, { "epoch": 3.84, "learning_rate": 2.6807095792719473e-06, "loss": 0.1008, "step": 7259 }, { "epoch": 3.84, "learning_rate": 2.678373342188115e-06, "loss": 0.0776, "step": 7260 }, { "epoch": 3.84, "learning_rate": 2.676037966140771e-06, "loss": 0.0779, "step": 7261 }, { "epoch": 3.84, "learning_rate": 2.6737034514045546e-06, "loss": 0.0763, "step": 7262 }, { "epoch": 3.84, "learning_rate": 2.671369798254013e-06, "loss": 0.1281, "step": 7263 }, { "epoch": 3.84, "learning_rate": 2.6690370069635863e-06, "loss": 0.0422, "step": 7264 }, { "epoch": 3.85, "learning_rate": 2.6667050778076132e-06, "loss": 0.1195, "step": 7265 }, { "epoch": 3.85, "learning_rate": 2.6643740110603356e-06, "loss": 0.066, "step": 7266 }, { "epoch": 3.85, "learning_rate": 2.6620438069958875e-06, "loss": 0.0794, "step": 7267 }, { "epoch": 3.85, "learning_rate": 2.659714465888311e-06, "loss": 0.0863, "step": 7268 }, { "epoch": 3.85, "learning_rate": 2.657385988011535e-06, "loss": 0.111, "step": 7269 }, { "epoch": 3.85, "learning_rate": 2.6550583736393933e-06, "loss": 0.0727, "step": 7270 }, { "epoch": 3.85, "learning_rate": 2.6527316230456234e-06, "loss": 0.0918, "step": 7271 }, { "epoch": 3.85, "learning_rate": 2.6504057365038517e-06, "loss": 0.0592, "step": 7272 }, { "epoch": 3.85, "learning_rate": 2.648080714287602e-06, "loss": 0.099, "step": 7273 }, { "epoch": 3.85, "learning_rate": 2.6457565566703094e-06, "loss": 0.0859, "step": 7274 }, { "epoch": 3.85, "learning_rate": 2.6434332639252904e-06, "loss": 0.1659, "step": 7275 }, { "epoch": 3.85, "learning_rate": 2.6411108363257764e-06, "loss": 0.109, "step": 7276 }, { "epoch": 3.85, "learning_rate": 2.638789274144884e-06, "loss": 0.079, "step": 7277 }, { "epoch": 3.85, "learning_rate": 2.6364685776556365e-06, "loss": 0.0956, "step": 7278 }, { "epoch": 3.85, "learning_rate": 2.634148747130947e-06, "loss": 0.0794, "step": 7279 }, { "epoch": 3.85, "learning_rate": 2.631829782843639e-06, "loss": 0.0736, "step": 7280 }, { "epoch": 3.85, "learning_rate": 2.6295116850664217e-06, "loss": 0.0781, "step": 7281 }, { "epoch": 3.85, "learning_rate": 2.6271944540719063e-06, "loss": 0.1001, "step": 7282 }, { "epoch": 3.85, "learning_rate": 2.624878090132609e-06, "loss": 0.0682, "step": 7283 }, { "epoch": 3.86, "learning_rate": 2.622562593520931e-06, "loss": 0.1204, "step": 7284 }, { "epoch": 3.86, "learning_rate": 2.620247964509184e-06, "loss": 0.098, "step": 7285 }, { "epoch": 3.86, "learning_rate": 2.6179342033695732e-06, "loss": 0.0872, "step": 7286 }, { "epoch": 3.86, "learning_rate": 2.615621310374198e-06, "loss": 0.1076, "step": 7287 }, { "epoch": 3.86, "learning_rate": 2.6133092857950625e-06, "loss": 0.1201, "step": 7288 }, { "epoch": 3.86, "learning_rate": 2.610998129904062e-06, "loss": 0.0984, "step": 7289 }, { "epoch": 3.86, "learning_rate": 2.6086878429729912e-06, "loss": 0.1315, "step": 7290 }, { "epoch": 3.86, "learning_rate": 2.606378425273549e-06, "loss": 0.1483, "step": 7291 }, { "epoch": 3.86, "learning_rate": 2.60406987707732e-06, "loss": 0.0882, "step": 7292 }, { "epoch": 3.86, "learning_rate": 2.601762198655802e-06, "loss": 0.1152, "step": 7293 }, { "epoch": 3.86, "learning_rate": 2.5994553902803743e-06, "loss": 0.0727, "step": 7294 }, { "epoch": 3.86, "learning_rate": 2.597149452222325e-06, "loss": 0.0579, "step": 7295 }, { "epoch": 3.86, "learning_rate": 2.594844384752844e-06, "loss": 0.1273, "step": 7296 }, { "epoch": 3.86, "learning_rate": 2.5925401881429968e-06, "loss": 0.1036, "step": 7297 }, { "epoch": 3.86, "learning_rate": 2.5902368626637717e-06, "loss": 0.1009, "step": 7298 }, { "epoch": 3.86, "learning_rate": 2.5879344085860393e-06, "loss": 0.0849, "step": 7299 }, { "epoch": 3.86, "learning_rate": 2.585632826180572e-06, "loss": 0.1437, "step": 7300 }, { "epoch": 3.86, "learning_rate": 2.583332115718046e-06, "loss": 0.0656, "step": 7301 }, { "epoch": 3.87, "learning_rate": 2.5810322774690212e-06, "loss": 0.118, "step": 7302 }, { "epoch": 3.87, "learning_rate": 2.57873331170397e-06, "loss": 0.1043, "step": 7303 }, { "epoch": 3.87, "learning_rate": 2.576435218693251e-06, "loss": 0.1046, "step": 7304 }, { "epoch": 3.87, "learning_rate": 2.5741379987071214e-06, "loss": 0.0448, "step": 7305 }, { "epoch": 3.87, "learning_rate": 2.5718416520157442e-06, "loss": 0.0956, "step": 7306 }, { "epoch": 3.87, "learning_rate": 2.5695461788891675e-06, "loss": 0.0884, "step": 7307 }, { "epoch": 3.87, "learning_rate": 2.5672515795973507e-06, "loss": 0.0919, "step": 7308 }, { "epoch": 3.87, "learning_rate": 2.564957854410135e-06, "loss": 0.0412, "step": 7309 }, { "epoch": 3.87, "learning_rate": 2.5626650035972733e-06, "loss": 0.0778, "step": 7310 }, { "epoch": 3.87, "learning_rate": 2.560373027428403e-06, "loss": 0.0627, "step": 7311 }, { "epoch": 3.87, "learning_rate": 2.5580819261730683e-06, "loss": 0.1028, "step": 7312 }, { "epoch": 3.87, "learning_rate": 2.5557917001007115e-06, "loss": 0.0904, "step": 7313 }, { "epoch": 3.87, "learning_rate": 2.5535023494806566e-06, "loss": 0.1408, "step": 7314 }, { "epoch": 3.87, "learning_rate": 2.551213874582141e-06, "loss": 0.1503, "step": 7315 }, { "epoch": 3.87, "learning_rate": 2.5489262756742962e-06, "loss": 0.0728, "step": 7316 }, { "epoch": 3.87, "learning_rate": 2.5466395530261412e-06, "loss": 0.1071, "step": 7317 }, { "epoch": 3.87, "learning_rate": 2.544353706906606e-06, "loss": 0.0795, "step": 7318 }, { "epoch": 3.87, "learning_rate": 2.542068737584504e-06, "loss": 0.0531, "step": 7319 }, { "epoch": 3.87, "learning_rate": 2.5397846453285568e-06, "loss": 0.0847, "step": 7320 }, { "epoch": 3.88, "learning_rate": 2.5375014304073764e-06, "loss": 0.0925, "step": 7321 }, { "epoch": 3.88, "learning_rate": 2.5352190930894683e-06, "loss": 0.0947, "step": 7322 }, { "epoch": 3.88, "learning_rate": 2.5329376336432465e-06, "loss": 0.1595, "step": 7323 }, { "epoch": 3.88, "learning_rate": 2.5306570523370075e-06, "loss": 0.0745, "step": 7324 }, { "epoch": 3.88, "learning_rate": 2.5283773494389597e-06, "loss": 0.0784, "step": 7325 }, { "epoch": 3.88, "learning_rate": 2.526098525217193e-06, "loss": 0.0724, "step": 7326 }, { "epoch": 3.88, "learning_rate": 2.5238205799397044e-06, "loss": 0.1336, "step": 7327 }, { "epoch": 3.88, "learning_rate": 2.521543513874387e-06, "loss": 0.0852, "step": 7328 }, { "epoch": 3.88, "learning_rate": 2.5192673272890258e-06, "loss": 0.2087, "step": 7329 }, { "epoch": 3.88, "learning_rate": 2.516992020451301e-06, "loss": 0.0801, "step": 7330 }, { "epoch": 3.88, "learning_rate": 2.514717593628799e-06, "loss": 0.0542, "step": 7331 }, { "epoch": 3.88, "learning_rate": 2.51244404708899e-06, "loss": 0.1124, "step": 7332 }, { "epoch": 3.88, "learning_rate": 2.5101713810992536e-06, "loss": 0.0793, "step": 7333 }, { "epoch": 3.88, "learning_rate": 2.5078995959268527e-06, "loss": 0.0797, "step": 7334 }, { "epoch": 3.88, "learning_rate": 2.5056286918389605e-06, "loss": 0.0866, "step": 7335 }, { "epoch": 3.88, "learning_rate": 2.5033586691026324e-06, "loss": 0.0696, "step": 7336 }, { "epoch": 3.88, "learning_rate": 2.5010895279848337e-06, "loss": 0.0764, "step": 7337 }, { "epoch": 3.88, "learning_rate": 2.498821268752416e-06, "loss": 0.0704, "step": 7338 }, { "epoch": 3.88, "learning_rate": 2.4965538916721275e-06, "loss": 0.158, "step": 7339 }, { "epoch": 3.89, "learning_rate": 2.494287397010623e-06, "loss": 0.0873, "step": 7340 }, { "epoch": 3.89, "learning_rate": 2.4920217850344397e-06, "loss": 0.1201, "step": 7341 }, { "epoch": 3.89, "learning_rate": 2.4897570560100203e-06, "loss": 0.0962, "step": 7342 }, { "epoch": 3.89, "learning_rate": 2.487493210203705e-06, "loss": 0.1053, "step": 7343 }, { "epoch": 3.89, "learning_rate": 2.485230247881719e-06, "loss": 0.0976, "step": 7344 }, { "epoch": 3.89, "learning_rate": 2.4829681693102002e-06, "loss": 0.0606, "step": 7345 }, { "epoch": 3.89, "learning_rate": 2.4807069747551615e-06, "loss": 0.0882, "step": 7346 }, { "epoch": 3.89, "learning_rate": 2.478446664482529e-06, "loss": 0.0729, "step": 7347 }, { "epoch": 3.89, "learning_rate": 2.476187238758122e-06, "loss": 0.1312, "step": 7348 }, { "epoch": 3.89, "learning_rate": 2.4739286978476474e-06, "loss": 0.1008, "step": 7349 }, { "epoch": 3.89, "learning_rate": 2.47167104201672e-06, "loss": 0.0688, "step": 7350 }, { "epoch": 3.89, "learning_rate": 2.469414271530838e-06, "loss": 0.0816, "step": 7351 }, { "epoch": 3.89, "learning_rate": 2.4671583866554073e-06, "loss": 0.1287, "step": 7352 }, { "epoch": 3.89, "learning_rate": 2.4649033876557217e-06, "loss": 0.1308, "step": 7353 }, { "epoch": 3.89, "learning_rate": 2.462649274796969e-06, "loss": 0.1234, "step": 7354 }, { "epoch": 3.89, "learning_rate": 2.4603960483442433e-06, "loss": 0.075, "step": 7355 }, { "epoch": 3.89, "learning_rate": 2.458143708562524e-06, "loss": 0.0875, "step": 7356 }, { "epoch": 3.89, "learning_rate": 2.45589225571669e-06, "loss": 0.1352, "step": 7357 }, { "epoch": 3.89, "learning_rate": 2.453641690071522e-06, "loss": 0.0767, "step": 7358 }, { "epoch": 3.9, "learning_rate": 2.4513920118916833e-06, "loss": 0.0554, "step": 7359 }, { "epoch": 3.9, "learning_rate": 2.4491432214417467e-06, "loss": 0.1039, "step": 7360 }, { "epoch": 3.9, "learning_rate": 2.44689531898617e-06, "loss": 0.081, "step": 7361 }, { "epoch": 3.9, "learning_rate": 2.444648304789308e-06, "loss": 0.0846, "step": 7362 }, { "epoch": 3.9, "learning_rate": 2.4424021791154206e-06, "loss": 0.0886, "step": 7363 }, { "epoch": 3.9, "learning_rate": 2.4401569422286496e-06, "loss": 0.1092, "step": 7364 }, { "epoch": 3.9, "learning_rate": 2.4379125943930438e-06, "loss": 0.1287, "step": 7365 }, { "epoch": 3.9, "learning_rate": 2.435669135872538e-06, "loss": 0.0847, "step": 7366 }, { "epoch": 3.9, "learning_rate": 2.433426566930973e-06, "loss": 0.0998, "step": 7367 }, { "epoch": 3.9, "learning_rate": 2.4311848878320723e-06, "loss": 0.0864, "step": 7368 }, { "epoch": 3.9, "learning_rate": 2.4289440988394675e-06, "loss": 0.0873, "step": 7369 }, { "epoch": 3.9, "learning_rate": 2.4267042002166752e-06, "loss": 0.1577, "step": 7370 }, { "epoch": 3.9, "learning_rate": 2.4244651922271113e-06, "loss": 0.0696, "step": 7371 }, { "epoch": 3.9, "learning_rate": 2.4222270751340916e-06, "loss": 0.0941, "step": 7372 }, { "epoch": 3.9, "learning_rate": 2.4199898492008155e-06, "loss": 0.0522, "step": 7373 }, { "epoch": 3.9, "learning_rate": 2.417753514690391e-06, "loss": 0.0542, "step": 7374 }, { "epoch": 3.9, "learning_rate": 2.4155180718658146e-06, "loss": 0.0822, "step": 7375 }, { "epoch": 3.9, "learning_rate": 2.413283520989974e-06, "loss": 0.0765, "step": 7376 }, { "epoch": 3.9, "learning_rate": 2.4110498623256627e-06, "loss": 0.1711, "step": 7377 }, { "epoch": 3.91, "learning_rate": 2.4088170961355595e-06, "loss": 0.0762, "step": 7378 }, { "epoch": 3.91, "learning_rate": 2.4065852226822384e-06, "loss": 0.089, "step": 7379 }, { "epoch": 3.91, "learning_rate": 2.4043542422281797e-06, "loss": 0.1055, "step": 7380 }, { "epoch": 3.91, "learning_rate": 2.402124155035741e-06, "loss": 0.0704, "step": 7381 }, { "epoch": 3.91, "learning_rate": 2.399894961367194e-06, "loss": 0.0626, "step": 7382 }, { "epoch": 3.91, "learning_rate": 2.3976666614846888e-06, "loss": 0.1235, "step": 7383 }, { "epoch": 3.91, "learning_rate": 2.3954392556502805e-06, "loss": 0.0581, "step": 7384 }, { "epoch": 3.91, "learning_rate": 2.393212744125921e-06, "loss": 0.1056, "step": 7385 }, { "epoch": 3.91, "learning_rate": 2.390987127173442e-06, "loss": 0.0671, "step": 7386 }, { "epoch": 3.91, "learning_rate": 2.3887624050545877e-06, "loss": 0.0941, "step": 7387 }, { "epoch": 3.91, "learning_rate": 2.3865385780309823e-06, "loss": 0.0932, "step": 7388 }, { "epoch": 3.91, "learning_rate": 2.384315646364158e-06, "loss": 0.0879, "step": 7389 }, { "epoch": 3.91, "learning_rate": 2.3820936103155366e-06, "loss": 0.0484, "step": 7390 }, { "epoch": 3.91, "learning_rate": 2.379872470146427e-06, "loss": 0.063, "step": 7391 }, { "epoch": 3.91, "learning_rate": 2.377652226118047e-06, "loss": 0.1041, "step": 7392 }, { "epoch": 3.91, "learning_rate": 2.375432878491497e-06, "loss": 0.101, "step": 7393 }, { "epoch": 3.91, "learning_rate": 2.3732144275277734e-06, "loss": 0.0808, "step": 7394 }, { "epoch": 3.91, "learning_rate": 2.3709968734877777e-06, "loss": 0.0633, "step": 7395 }, { "epoch": 3.91, "learning_rate": 2.3687802166322892e-06, "loss": 0.1062, "step": 7396 }, { "epoch": 3.92, "learning_rate": 2.3665644572219993e-06, "loss": 0.0812, "step": 7397 }, { "epoch": 3.92, "learning_rate": 2.364349595517478e-06, "loss": 0.0972, "step": 7398 }, { "epoch": 3.92, "learning_rate": 2.362135631779201e-06, "loss": 0.0658, "step": 7399 }, { "epoch": 3.92, "learning_rate": 2.359922566267536e-06, "loss": 0.0783, "step": 7400 }, { "epoch": 3.92, "learning_rate": 2.357710399242742e-06, "loss": 0.096, "step": 7401 }, { "epoch": 3.92, "learning_rate": 2.3554991309649723e-06, "loss": 0.0755, "step": 7402 }, { "epoch": 3.92, "learning_rate": 2.3532887616942746e-06, "loss": 0.0956, "step": 7403 }, { "epoch": 3.92, "learning_rate": 2.3510792916905946e-06, "loss": 0.0678, "step": 7404 }, { "epoch": 3.92, "learning_rate": 2.348870721213773e-06, "loss": 0.0739, "step": 7405 }, { "epoch": 3.92, "learning_rate": 2.3466630505235353e-06, "loss": 0.0756, "step": 7406 }, { "epoch": 3.92, "learning_rate": 2.3444562798795133e-06, "loss": 0.1206, "step": 7407 }, { "epoch": 3.92, "learning_rate": 2.3422504095412236e-06, "loss": 0.0942, "step": 7408 }, { "epoch": 3.92, "learning_rate": 2.340045439768083e-06, "loss": 0.0653, "step": 7409 }, { "epoch": 3.92, "learning_rate": 2.337841370819399e-06, "loss": 0.1169, "step": 7410 }, { "epoch": 3.92, "learning_rate": 2.335638202954371e-06, "loss": 0.1289, "step": 7411 }, { "epoch": 3.92, "learning_rate": 2.3334359364321015e-06, "loss": 0.064, "step": 7412 }, { "epoch": 3.92, "learning_rate": 2.331234571511575e-06, "loss": 0.119, "step": 7413 }, { "epoch": 3.92, "learning_rate": 2.32903410845168e-06, "loss": 0.1535, "step": 7414 }, { "epoch": 3.92, "learning_rate": 2.326834547511192e-06, "loss": 0.1014, "step": 7415 }, { "epoch": 3.93, "learning_rate": 2.3246358889487852e-06, "loss": 0.0983, "step": 7416 }, { "epoch": 3.93, "learning_rate": 2.3224381330230305e-06, "loss": 0.0749, "step": 7417 }, { "epoch": 3.93, "learning_rate": 2.320241279992378e-06, "loss": 0.0978, "step": 7418 }, { "epoch": 3.93, "learning_rate": 2.3180453301151863e-06, "loss": 0.0721, "step": 7419 }, { "epoch": 3.93, "learning_rate": 2.3158502836497055e-06, "loss": 0.0799, "step": 7420 }, { "epoch": 3.93, "learning_rate": 2.3136561408540723e-06, "loss": 0.0709, "step": 7421 }, { "epoch": 3.93, "learning_rate": 2.3114629019863277e-06, "loss": 0.0918, "step": 7422 }, { "epoch": 3.93, "learning_rate": 2.3092705673043945e-06, "loss": 0.1011, "step": 7423 }, { "epoch": 3.93, "learning_rate": 2.3070791370661007e-06, "loss": 0.1013, "step": 7424 }, { "epoch": 3.93, "learning_rate": 2.3048886115291593e-06, "loss": 0.1273, "step": 7425 }, { "epoch": 3.93, "learning_rate": 2.3026989909511777e-06, "loss": 0.0923, "step": 7426 }, { "epoch": 3.93, "learning_rate": 2.3005102755896657e-06, "loss": 0.0689, "step": 7427 }, { "epoch": 3.93, "learning_rate": 2.298322465702013e-06, "loss": 0.0792, "step": 7428 }, { "epoch": 3.93, "learning_rate": 2.296135561545516e-06, "loss": 0.0753, "step": 7429 }, { "epoch": 3.93, "learning_rate": 2.2939495633773533e-06, "loss": 0.0972, "step": 7430 }, { "epoch": 3.93, "learning_rate": 2.291764471454605e-06, "loss": 0.1422, "step": 7431 }, { "epoch": 3.93, "learning_rate": 2.2895802860342455e-06, "loss": 0.0836, "step": 7432 }, { "epoch": 3.93, "learning_rate": 2.2873970073731334e-06, "loss": 0.0657, "step": 7433 }, { "epoch": 3.93, "learning_rate": 2.285214635728027e-06, "loss": 0.1126, "step": 7434 }, { "epoch": 3.94, "learning_rate": 2.28303317135558e-06, "loss": 0.0929, "step": 7435 }, { "epoch": 3.94, "learning_rate": 2.280852614512332e-06, "loss": 0.0768, "step": 7436 }, { "epoch": 3.94, "learning_rate": 2.278672965454726e-06, "loss": 0.0907, "step": 7437 }, { "epoch": 3.94, "learning_rate": 2.2764942244390866e-06, "loss": 0.1351, "step": 7438 }, { "epoch": 3.94, "learning_rate": 2.274316391721644e-06, "loss": 0.0643, "step": 7439 }, { "epoch": 3.94, "learning_rate": 2.2721394675585097e-06, "loss": 0.0764, "step": 7440 }, { "epoch": 3.94, "learning_rate": 2.269963452205698e-06, "loss": 0.0857, "step": 7441 }, { "epoch": 3.94, "learning_rate": 2.2677883459191117e-06, "loss": 0.1113, "step": 7442 }, { "epoch": 3.94, "learning_rate": 2.2656141489545426e-06, "loss": 0.0929, "step": 7443 }, { "epoch": 3.94, "learning_rate": 2.2634408615676862e-06, "loss": 0.0793, "step": 7444 }, { "epoch": 3.94, "learning_rate": 2.2612684840141208e-06, "loss": 0.0985, "step": 7445 }, { "epoch": 3.94, "learning_rate": 2.259097016549322e-06, "loss": 0.0678, "step": 7446 }, { "epoch": 3.94, "learning_rate": 2.2569264594286634e-06, "loss": 0.0564, "step": 7447 }, { "epoch": 3.94, "learning_rate": 2.2547568129074006e-06, "loss": 0.1121, "step": 7448 }, { "epoch": 3.94, "learning_rate": 2.2525880772406917e-06, "loss": 0.0797, "step": 7449 }, { "epoch": 3.94, "learning_rate": 2.250420252683583e-06, "loss": 0.0596, "step": 7450 }, { "epoch": 3.94, "learning_rate": 2.248253339491011e-06, "loss": 0.1158, "step": 7451 }, { "epoch": 3.94, "learning_rate": 2.246087337917815e-06, "loss": 0.105, "step": 7452 }, { "epoch": 3.94, "learning_rate": 2.243922248218714e-06, "loss": 0.0897, "step": 7453 }, { "epoch": 3.95, "learning_rate": 2.2417580706483334e-06, "loss": 0.0746, "step": 7454 }, { "epoch": 3.95, "learning_rate": 2.239594805461177e-06, "loss": 0.0927, "step": 7455 }, { "epoch": 3.95, "learning_rate": 2.237432452911655e-06, "loss": 0.1276, "step": 7456 }, { "epoch": 3.95, "learning_rate": 2.235271013254059e-06, "loss": 0.0667, "step": 7457 }, { "epoch": 3.95, "learning_rate": 2.2331104867425836e-06, "loss": 0.1323, "step": 7458 }, { "epoch": 3.95, "learning_rate": 2.2309508736313067e-06, "loss": 0.0966, "step": 7459 }, { "epoch": 3.95, "learning_rate": 2.228792174174201e-06, "loss": 0.1012, "step": 7460 }, { "epoch": 3.95, "learning_rate": 2.2266343886251364e-06, "loss": 0.0859, "step": 7461 }, { "epoch": 3.95, "learning_rate": 2.2244775172378754e-06, "loss": 0.0568, "step": 7462 }, { "epoch": 3.95, "learning_rate": 2.222321560266062e-06, "loss": 0.096, "step": 7463 }, { "epoch": 3.95, "learning_rate": 2.220166517963249e-06, "loss": 0.0624, "step": 7464 }, { "epoch": 3.95, "learning_rate": 2.2180123905828667e-06, "loss": 0.0999, "step": 7465 }, { "epoch": 3.95, "learning_rate": 2.215859178378249e-06, "loss": 0.1025, "step": 7466 }, { "epoch": 3.95, "learning_rate": 2.2137068816026165e-06, "loss": 0.1091, "step": 7467 }, { "epoch": 3.95, "learning_rate": 2.2115555005090796e-06, "loss": 0.0813, "step": 7468 }, { "epoch": 3.95, "learning_rate": 2.209405035350649e-06, "loss": 0.1015, "step": 7469 }, { "epoch": 3.95, "learning_rate": 2.2072554863802197e-06, "loss": 0.1098, "step": 7470 }, { "epoch": 3.95, "learning_rate": 2.205106853850587e-06, "loss": 0.1166, "step": 7471 }, { "epoch": 3.96, "learning_rate": 2.2029591380144276e-06, "loss": 0.1025, "step": 7472 }, { "epoch": 3.96, "learning_rate": 2.2008123391243217e-06, "loss": 0.1278, "step": 7473 }, { "epoch": 3.96, "learning_rate": 2.198666457432741e-06, "loss": 0.0702, "step": 7474 }, { "epoch": 3.96, "learning_rate": 2.196521493192033e-06, "loss": 0.0632, "step": 7475 }, { "epoch": 3.96, "learning_rate": 2.1943774466544564e-06, "loss": 0.1126, "step": 7476 }, { "epoch": 3.96, "learning_rate": 2.192234318072157e-06, "loss": 0.0925, "step": 7477 }, { "epoch": 3.96, "learning_rate": 2.190092107697166e-06, "loss": 0.1444, "step": 7478 }, { "epoch": 3.96, "learning_rate": 2.1879508157814155e-06, "loss": 0.1044, "step": 7479 }, { "epoch": 3.96, "learning_rate": 2.18581044257672e-06, "loss": 0.0762, "step": 7480 }, { "epoch": 3.96, "learning_rate": 2.183670988334797e-06, "loss": 0.0636, "step": 7481 }, { "epoch": 3.96, "learning_rate": 2.181532453307249e-06, "loss": 0.0765, "step": 7482 }, { "epoch": 3.96, "learning_rate": 2.1793948377455652e-06, "loss": 0.0773, "step": 7483 }, { "epoch": 3.96, "learning_rate": 2.177258141901142e-06, "loss": 0.0692, "step": 7484 }, { "epoch": 3.96, "learning_rate": 2.1751223660252507e-06, "loss": 0.1486, "step": 7485 }, { "epoch": 3.96, "learning_rate": 2.17298751036907e-06, "loss": 0.1145, "step": 7486 }, { "epoch": 3.96, "learning_rate": 2.170853575183656e-06, "loss": 0.0597, "step": 7487 }, { "epoch": 3.96, "learning_rate": 2.168720560719966e-06, "loss": 0.0919, "step": 7488 }, { "epoch": 3.96, "learning_rate": 2.1665884672288483e-06, "loss": 0.0825, "step": 7489 }, { "epoch": 3.96, "learning_rate": 2.164457294961041e-06, "loss": 0.134, "step": 7490 }, { "epoch": 3.97, "learning_rate": 2.1623270441671716e-06, "loss": 0.1395, "step": 7491 }, { "epoch": 3.97, "learning_rate": 2.160197715097758e-06, "loss": 0.1127, "step": 7492 }, { "epoch": 3.97, "learning_rate": 2.158069308003219e-06, "loss": 0.1202, "step": 7493 }, { "epoch": 3.97, "learning_rate": 2.1559418231338593e-06, "loss": 0.0575, "step": 7494 }, { "epoch": 3.97, "learning_rate": 2.1538152607398688e-06, "loss": 0.0865, "step": 7495 }, { "epoch": 3.97, "learning_rate": 2.1516896210713435e-06, "loss": 0.0679, "step": 7496 }, { "epoch": 3.97, "learning_rate": 2.149564904378254e-06, "loss": 0.0799, "step": 7497 }, { "epoch": 3.97, "learning_rate": 2.1474411109104785e-06, "loss": 0.0904, "step": 7498 }, { "epoch": 3.97, "learning_rate": 2.145318240917775e-06, "loss": 0.064, "step": 7499 }, { "epoch": 3.97, "learning_rate": 2.143196294649794e-06, "loss": 0.1032, "step": 7500 }, { "epoch": 3.97, "learning_rate": 2.141075272356087e-06, "loss": 0.1101, "step": 7501 }, { "epoch": 3.97, "learning_rate": 2.1389551742860824e-06, "loss": 0.1786, "step": 7502 }, { "epoch": 3.97, "learning_rate": 2.136836000689112e-06, "loss": 0.087, "step": 7503 }, { "epoch": 3.97, "learning_rate": 2.134717751814397e-06, "loss": 0.0634, "step": 7504 }, { "epoch": 3.97, "learning_rate": 2.132600427911041e-06, "loss": 0.0954, "step": 7505 }, { "epoch": 3.97, "learning_rate": 2.130484029228055e-06, "loss": 0.0765, "step": 7506 }, { "epoch": 3.97, "learning_rate": 2.1283685560143188e-06, "loss": 0.0988, "step": 7507 }, { "epoch": 3.97, "learning_rate": 2.12625400851862e-06, "loss": 0.1152, "step": 7508 }, { "epoch": 3.97, "learning_rate": 2.1241403869896393e-06, "loss": 0.0576, "step": 7509 }, { "epoch": 3.98, "learning_rate": 2.122027691675935e-06, "loss": 0.0979, "step": 7510 }, { "epoch": 3.98, "learning_rate": 2.119915922825969e-06, "loss": 0.0709, "step": 7511 }, { "epoch": 3.98, "learning_rate": 2.1178050806880856e-06, "loss": 0.0759, "step": 7512 }, { "epoch": 3.98, "learning_rate": 2.1156951655105263e-06, "loss": 0.0752, "step": 7513 }, { "epoch": 3.98, "learning_rate": 2.11358617754142e-06, "loss": 0.1085, "step": 7514 }, { "epoch": 3.98, "learning_rate": 2.1114781170287845e-06, "loss": 0.1016, "step": 7515 }, { "epoch": 3.98, "learning_rate": 2.109370984220538e-06, "loss": 0.113, "step": 7516 }, { "epoch": 3.98, "learning_rate": 2.1072647793644753e-06, "loss": 0.0911, "step": 7517 }, { "epoch": 3.98, "learning_rate": 2.105159502708296e-06, "loss": 0.0943, "step": 7518 }, { "epoch": 3.98, "learning_rate": 2.1030551544995813e-06, "loss": 0.0645, "step": 7519 }, { "epoch": 3.98, "learning_rate": 2.1009517349858065e-06, "loss": 0.0736, "step": 7520 }, { "epoch": 3.98, "learning_rate": 2.0988492444143406e-06, "loss": 0.089, "step": 7521 }, { "epoch": 3.98, "learning_rate": 2.0967476830324384e-06, "loss": 0.083, "step": 7522 }, { "epoch": 3.98, "learning_rate": 2.094647051087244e-06, "loss": 0.108, "step": 7523 }, { "epoch": 3.98, "learning_rate": 2.0925473488258017e-06, "loss": 0.0703, "step": 7524 }, { "epoch": 3.98, "learning_rate": 2.0904485764950344e-06, "loss": 0.0967, "step": 7525 }, { "epoch": 3.98, "learning_rate": 2.0883507343417664e-06, "loss": 0.0526, "step": 7526 }, { "epoch": 3.98, "learning_rate": 2.086253822612703e-06, "loss": 0.0947, "step": 7527 }, { "epoch": 3.98, "learning_rate": 2.0841578415544493e-06, "loss": 0.1078, "step": 7528 }, { "epoch": 3.99, "learning_rate": 2.0820627914134916e-06, "loss": 0.0573, "step": 7529 }, { "epoch": 3.99, "learning_rate": 2.079968672436218e-06, "loss": 0.0814, "step": 7530 }, { "epoch": 3.99, "learning_rate": 2.0778754848688966e-06, "loss": 0.0863, "step": 7531 }, { "epoch": 3.99, "learning_rate": 2.0757832289576875e-06, "loss": 0.0888, "step": 7532 }, { "epoch": 3.99, "learning_rate": 2.0736919049486494e-06, "loss": 0.0878, "step": 7533 }, { "epoch": 3.99, "learning_rate": 2.0716015130877197e-06, "loss": 0.1084, "step": 7534 }, { "epoch": 3.99, "learning_rate": 2.069512053620737e-06, "loss": 0.0662, "step": 7535 }, { "epoch": 3.99, "learning_rate": 2.0674235267934262e-06, "loss": 0.0583, "step": 7536 }, { "epoch": 3.99, "learning_rate": 2.0653359328513967e-06, "loss": 0.0892, "step": 7537 }, { "epoch": 3.99, "learning_rate": 2.06324927204016e-06, "loss": 0.0611, "step": 7538 }, { "epoch": 3.99, "learning_rate": 2.061163544605107e-06, "loss": 0.0674, "step": 7539 }, { "epoch": 3.99, "learning_rate": 2.05907875079152e-06, "loss": 0.124, "step": 7540 }, { "epoch": 3.99, "learning_rate": 2.0569948908445815e-06, "loss": 0.0799, "step": 7541 }, { "epoch": 3.99, "learning_rate": 2.054911965009351e-06, "loss": 0.0973, "step": 7542 }, { "epoch": 3.99, "learning_rate": 2.05282997353079e-06, "loss": 0.124, "step": 7543 }, { "epoch": 3.99, "learning_rate": 2.0507489166537384e-06, "loss": 0.0921, "step": 7544 }, { "epoch": 3.99, "learning_rate": 2.0486687946229356e-06, "loss": 0.0496, "step": 7545 }, { "epoch": 3.99, "learning_rate": 2.0465896076830128e-06, "loss": 0.1147, "step": 7546 }, { "epoch": 3.99, "learning_rate": 2.044511356078475e-06, "loss": 0.0655, "step": 7547 }, { "epoch": 4.0, "learning_rate": 2.042434040053737e-06, "loss": 0.0727, "step": 7548 }, { "epoch": 4.0, "learning_rate": 2.0403576598530895e-06, "loss": 0.0857, "step": 7549 }, { "epoch": 4.0, "learning_rate": 2.0382822157207204e-06, "loss": 0.1157, "step": 7550 }, { "epoch": 4.0, "learning_rate": 2.0362077079007105e-06, "loss": 0.1568, "step": 7551 }, { "epoch": 4.0, "learning_rate": 2.034134136637017e-06, "loss": 0.0813, "step": 7552 }, { "epoch": 4.0, "learning_rate": 2.0320615021735036e-06, "loss": 0.0637, "step": 7553 }, { "epoch": 4.0, "learning_rate": 2.0299898047539126e-06, "loss": 0.0671, "step": 7554 }, { "epoch": 4.0, "learning_rate": 2.0279190446218756e-06, "loss": 0.096, "step": 7555 }, { "epoch": 4.0, "learning_rate": 2.025849222020925e-06, "loss": 0.0516, "step": 7556 }, { "epoch": 4.0, "learning_rate": 2.0237803371944675e-06, "loss": 0.0658, "step": 7557 }, { "epoch": 4.0, "learning_rate": 2.021712390385816e-06, "loss": 0.0992, "step": 7558 }, { "epoch": 4.0, "learning_rate": 2.019645381838158e-06, "loss": 0.0402, "step": 7559 }, { "epoch": 4.0, "learning_rate": 2.017579311794583e-06, "loss": 0.0622, "step": 7560 }, { "epoch": 4.0, "learning_rate": 2.0155141804980594e-06, "loss": 0.0636, "step": 7561 }, { "epoch": 4.0, "learning_rate": 2.013449988191455e-06, "loss": 0.0912, "step": 7562 }, { "epoch": 4.0, "learning_rate": 2.0113867351175197e-06, "loss": 0.0764, "step": 7563 }, { "epoch": 4.0, "learning_rate": 2.0093244215188944e-06, "loss": 0.0486, "step": 7564 }, { "epoch": 4.0, "learning_rate": 2.0072630476381137e-06, "loss": 0.0981, "step": 7565 }, { "epoch": 4.0, "learning_rate": 2.0052026137175995e-06, "loss": 0.0871, "step": 7566 }, { "epoch": 4.01, "learning_rate": 2.0031431199996587e-06, "loss": 0.0704, "step": 7567 }, { "epoch": 4.01, "learning_rate": 2.0010845667264967e-06, "loss": 0.0483, "step": 7568 }, { "epoch": 4.01, "learning_rate": 1.9990269541401983e-06, "loss": 0.0524, "step": 7569 }, { "epoch": 4.01, "learning_rate": 1.996970282482746e-06, "loss": 0.0453, "step": 7570 }, { "epoch": 4.01, "learning_rate": 1.9949145519960067e-06, "loss": 0.0702, "step": 7571 }, { "epoch": 4.01, "learning_rate": 1.992859762921736e-06, "loss": 0.0769, "step": 7572 }, { "epoch": 4.01, "learning_rate": 1.9908059155015846e-06, "loss": 0.0572, "step": 7573 }, { "epoch": 4.01, "learning_rate": 1.9887530099770834e-06, "loss": 0.0875, "step": 7574 }, { "epoch": 4.01, "learning_rate": 1.9867010465896643e-06, "loss": 0.0599, "step": 7575 }, { "epoch": 4.01, "learning_rate": 1.9846500255806365e-06, "loss": 0.0755, "step": 7576 }, { "epoch": 4.01, "learning_rate": 1.9825999471912048e-06, "loss": 0.0936, "step": 7577 }, { "epoch": 4.01, "learning_rate": 1.9805508116624663e-06, "loss": 0.0762, "step": 7578 }, { "epoch": 4.01, "learning_rate": 1.9785026192354007e-06, "loss": 0.0739, "step": 7579 }, { "epoch": 4.01, "learning_rate": 1.976455370150875e-06, "loss": 0.0657, "step": 7580 }, { "epoch": 4.01, "learning_rate": 1.974409064649655e-06, "loss": 0.0488, "step": 7581 }, { "epoch": 4.01, "learning_rate": 1.972363702972385e-06, "loss": 0.103, "step": 7582 }, { "epoch": 4.01, "learning_rate": 1.97031928535961e-06, "loss": 0.0512, "step": 7583 }, { "epoch": 4.01, "learning_rate": 1.968275812051749e-06, "loss": 0.0677, "step": 7584 }, { "epoch": 4.01, "learning_rate": 1.966233283289125e-06, "loss": 0.1045, "step": 7585 }, { "epoch": 4.02, "learning_rate": 1.964191699311938e-06, "loss": 0.0659, "step": 7586 }, { "epoch": 4.02, "learning_rate": 1.9621510603602867e-06, "loss": 0.0938, "step": 7587 }, { "epoch": 4.02, "learning_rate": 1.9601113666741523e-06, "loss": 0.0979, "step": 7588 }, { "epoch": 4.02, "learning_rate": 1.9580726184934017e-06, "loss": 0.0555, "step": 7589 }, { "epoch": 4.02, "learning_rate": 1.9560348160578025e-06, "loss": 0.0534, "step": 7590 }, { "epoch": 4.02, "learning_rate": 1.953997959606998e-06, "loss": 0.079, "step": 7591 }, { "epoch": 4.02, "learning_rate": 1.951962049380529e-06, "loss": 0.0899, "step": 7592 }, { "epoch": 4.02, "learning_rate": 1.949927085617824e-06, "loss": 0.0657, "step": 7593 }, { "epoch": 4.02, "learning_rate": 1.947893068558194e-06, "loss": 0.0738, "step": 7594 }, { "epoch": 4.02, "learning_rate": 1.945859998440852e-06, "loss": 0.0565, "step": 7595 }, { "epoch": 4.02, "learning_rate": 1.9438278755048777e-06, "loss": 0.0805, "step": 7596 }, { "epoch": 4.02, "learning_rate": 1.94179669998926e-06, "loss": 0.067, "step": 7597 }, { "epoch": 4.02, "learning_rate": 1.9397664721328702e-06, "loss": 0.101, "step": 7598 }, { "epoch": 4.02, "learning_rate": 1.9377371921744616e-06, "loss": 0.1155, "step": 7599 }, { "epoch": 4.02, "learning_rate": 1.935708860352686e-06, "loss": 0.108, "step": 7600 }, { "epoch": 4.02, "learning_rate": 1.933681476906074e-06, "loss": 0.0552, "step": 7601 }, { "epoch": 4.02, "learning_rate": 1.931655042073056e-06, "loss": 0.0835, "step": 7602 }, { "epoch": 4.02, "learning_rate": 1.9296295560919397e-06, "loss": 0.0992, "step": 7603 }, { "epoch": 4.02, "learning_rate": 1.9276050192009256e-06, "loss": 0.1232, "step": 7604 }, { "epoch": 4.03, "learning_rate": 1.9255814316381063e-06, "loss": 0.0727, "step": 7605 }, { "epoch": 4.03, "learning_rate": 1.9235587936414547e-06, "loss": 0.0766, "step": 7606 }, { "epoch": 4.03, "learning_rate": 1.9215371054488385e-06, "loss": 0.0756, "step": 7607 }, { "epoch": 4.03, "learning_rate": 1.9195163672980156e-06, "loss": 0.0723, "step": 7608 }, { "epoch": 4.03, "learning_rate": 1.9174965794266232e-06, "loss": 0.0839, "step": 7609 }, { "epoch": 4.03, "learning_rate": 1.915477742072197e-06, "loss": 0.0987, "step": 7610 }, { "epoch": 4.03, "learning_rate": 1.913459855472153e-06, "loss": 0.0841, "step": 7611 }, { "epoch": 4.03, "learning_rate": 1.9114429198637953e-06, "loss": 0.0805, "step": 7612 }, { "epoch": 4.03, "learning_rate": 1.9094269354843253e-06, "loss": 0.0596, "step": 7613 }, { "epoch": 4.03, "learning_rate": 1.9074119025708205e-06, "loss": 0.0873, "step": 7614 }, { "epoch": 4.03, "learning_rate": 1.9053978213602586e-06, "loss": 0.0796, "step": 7615 }, { "epoch": 4.03, "learning_rate": 1.9033846920894926e-06, "loss": 0.0736, "step": 7616 }, { "epoch": 4.03, "learning_rate": 1.9013725149952766e-06, "loss": 0.0761, "step": 7617 }, { "epoch": 4.03, "learning_rate": 1.89936129031424e-06, "loss": 0.0835, "step": 7618 }, { "epoch": 4.03, "learning_rate": 1.8973510182829124e-06, "loss": 0.1197, "step": 7619 }, { "epoch": 4.03, "learning_rate": 1.8953416991377026e-06, "loss": 0.0916, "step": 7620 }, { "epoch": 4.03, "learning_rate": 1.8933333331149073e-06, "loss": 0.0787, "step": 7621 }, { "epoch": 4.03, "learning_rate": 1.8913259204507196e-06, "loss": 0.0453, "step": 7622 }, { "epoch": 4.03, "learning_rate": 1.8893194613812093e-06, "loss": 0.0902, "step": 7623 }, { "epoch": 4.04, "learning_rate": 1.8873139561423425e-06, "loss": 0.0703, "step": 7624 }, { "epoch": 4.04, "learning_rate": 1.8853094049699728e-06, "loss": 0.0528, "step": 7625 }, { "epoch": 4.04, "learning_rate": 1.8833058080998323e-06, "loss": 0.0631, "step": 7626 }, { "epoch": 4.04, "learning_rate": 1.881303165767555e-06, "loss": 0.0577, "step": 7627 }, { "epoch": 4.04, "learning_rate": 1.8793014782086528e-06, "loss": 0.0954, "step": 7628 }, { "epoch": 4.04, "learning_rate": 1.8773007456585224e-06, "loss": 0.1003, "step": 7629 }, { "epoch": 4.04, "learning_rate": 1.8753009683524614e-06, "loss": 0.0874, "step": 7630 }, { "epoch": 4.04, "learning_rate": 1.8733021465256408e-06, "loss": 0.0527, "step": 7631 }, { "epoch": 4.04, "learning_rate": 1.8713042804131321e-06, "loss": 0.0697, "step": 7632 }, { "epoch": 4.04, "learning_rate": 1.869307370249881e-06, "loss": 0.0919, "step": 7633 }, { "epoch": 4.04, "learning_rate": 1.8673114162707319e-06, "loss": 0.0769, "step": 7634 }, { "epoch": 4.04, "learning_rate": 1.8653164187104156e-06, "loss": 0.0658, "step": 7635 }, { "epoch": 4.04, "learning_rate": 1.8633223778035393e-06, "loss": 0.0472, "step": 7636 }, { "epoch": 4.04, "learning_rate": 1.861329293784613e-06, "loss": 0.0928, "step": 7637 }, { "epoch": 4.04, "learning_rate": 1.8593371668880212e-06, "loss": 0.0902, "step": 7638 }, { "epoch": 4.04, "learning_rate": 1.8573459973480446e-06, "loss": 0.0674, "step": 7639 }, { "epoch": 4.04, "learning_rate": 1.8553557853988514e-06, "loss": 0.0842, "step": 7640 }, { "epoch": 4.04, "learning_rate": 1.8533665312744875e-06, "loss": 0.0716, "step": 7641 }, { "epoch": 4.04, "learning_rate": 1.8513782352088994e-06, "loss": 0.0472, "step": 7642 }, { "epoch": 4.05, "learning_rate": 1.8493908974359108e-06, "loss": 0.0704, "step": 7643 }, { "epoch": 4.05, "learning_rate": 1.8474045181892341e-06, "loss": 0.0729, "step": 7644 }, { "epoch": 4.05, "learning_rate": 1.8454190977024767e-06, "loss": 0.0682, "step": 7645 }, { "epoch": 4.05, "learning_rate": 1.8434346362091215e-06, "loss": 0.09, "step": 7646 }, { "epoch": 4.05, "learning_rate": 1.84145113394255e-06, "loss": 0.066, "step": 7647 }, { "epoch": 4.05, "learning_rate": 1.839468591136021e-06, "loss": 0.0543, "step": 7648 }, { "epoch": 4.05, "learning_rate": 1.837487008022687e-06, "loss": 0.0517, "step": 7649 }, { "epoch": 4.05, "learning_rate": 1.8355063848355892e-06, "loss": 0.0918, "step": 7650 }, { "epoch": 4.05, "learning_rate": 1.833526721807649e-06, "loss": 0.0755, "step": 7651 }, { "epoch": 4.05, "learning_rate": 1.8315480191716794e-06, "loss": 0.1119, "step": 7652 }, { "epoch": 4.05, "learning_rate": 1.8295702771603763e-06, "loss": 0.0806, "step": 7653 }, { "epoch": 4.05, "learning_rate": 1.827593496006328e-06, "loss": 0.0668, "step": 7654 }, { "epoch": 4.05, "learning_rate": 1.8256176759420097e-06, "loss": 0.0365, "step": 7655 }, { "epoch": 4.05, "learning_rate": 1.8236428171997778e-06, "loss": 0.08, "step": 7656 }, { "epoch": 4.05, "learning_rate": 1.8216689200118832e-06, "loss": 0.0786, "step": 7657 }, { "epoch": 4.05, "learning_rate": 1.8196959846104545e-06, "loss": 0.0829, "step": 7658 }, { "epoch": 4.05, "learning_rate": 1.8177240112275174e-06, "loss": 0.0686, "step": 7659 }, { "epoch": 4.05, "learning_rate": 1.8157530000949787e-06, "loss": 0.0905, "step": 7660 }, { "epoch": 4.06, "learning_rate": 1.8137829514446281e-06, "loss": 0.0488, "step": 7661 }, { "epoch": 4.06, "learning_rate": 1.811813865508154e-06, "loss": 0.0587, "step": 7662 }, { "epoch": 4.06, "learning_rate": 1.8098457425171178e-06, "loss": 0.0338, "step": 7663 }, { "epoch": 4.06, "learning_rate": 1.8078785827029799e-06, "loss": 0.0412, "step": 7664 }, { "epoch": 4.06, "learning_rate": 1.8059123862970773e-06, "loss": 0.0973, "step": 7665 }, { "epoch": 4.06, "learning_rate": 1.8039471535306397e-06, "loss": 0.054, "step": 7666 }, { "epoch": 4.06, "learning_rate": 1.8019828846347875e-06, "loss": 0.0767, "step": 7667 }, { "epoch": 4.06, "learning_rate": 1.8000195798405128e-06, "loss": 0.0682, "step": 7668 }, { "epoch": 4.06, "learning_rate": 1.7980572393787088e-06, "loss": 0.0543, "step": 7669 }, { "epoch": 4.06, "learning_rate": 1.7960958634801518e-06, "loss": 0.0567, "step": 7670 }, { "epoch": 4.06, "learning_rate": 1.7941354523754984e-06, "loss": 0.0448, "step": 7671 }, { "epoch": 4.06, "learning_rate": 1.792176006295302e-06, "loss": 0.0705, "step": 7672 }, { "epoch": 4.06, "learning_rate": 1.7902175254699917e-06, "loss": 0.0888, "step": 7673 }, { "epoch": 4.06, "learning_rate": 1.788260010129893e-06, "loss": 0.0798, "step": 7674 }, { "epoch": 4.06, "learning_rate": 1.7863034605052122e-06, "loss": 0.0815, "step": 7675 }, { "epoch": 4.06, "learning_rate": 1.784347876826038e-06, "loss": 0.0858, "step": 7676 }, { "epoch": 4.06, "learning_rate": 1.7823932593223582e-06, "loss": 0.0501, "step": 7677 }, { "epoch": 4.06, "learning_rate": 1.7804396082240326e-06, "loss": 0.1138, "step": 7678 }, { "epoch": 4.06, "learning_rate": 1.7784869237608203e-06, "loss": 0.0766, "step": 7679 }, { "epoch": 4.07, "learning_rate": 1.7765352061623543e-06, "loss": 0.0457, "step": 7680 }, { "epoch": 4.07, "learning_rate": 1.7745844556581638e-06, "loss": 0.1053, "step": 7681 }, { "epoch": 4.07, "learning_rate": 1.7726346724776622e-06, "loss": 0.0772, "step": 7682 }, { "epoch": 4.07, "learning_rate": 1.7706858568501451e-06, "loss": 0.0725, "step": 7683 }, { "epoch": 4.07, "learning_rate": 1.7687380090047944e-06, "loss": 0.0878, "step": 7684 }, { "epoch": 4.07, "learning_rate": 1.7667911291706853e-06, "loss": 0.0613, "step": 7685 }, { "epoch": 4.07, "learning_rate": 1.7648452175767693e-06, "loss": 0.0667, "step": 7686 }, { "epoch": 4.07, "learning_rate": 1.7629002744518953e-06, "loss": 0.0554, "step": 7687 }, { "epoch": 4.07, "learning_rate": 1.7609563000247866e-06, "loss": 0.0919, "step": 7688 }, { "epoch": 4.07, "learning_rate": 1.7590132945240624e-06, "loss": 0.2203, "step": 7689 }, { "epoch": 4.07, "learning_rate": 1.7570712581782179e-06, "loss": 0.0369, "step": 7690 }, { "epoch": 4.07, "learning_rate": 1.755130191215645e-06, "loss": 0.0672, "step": 7691 }, { "epoch": 4.07, "learning_rate": 1.7531900938646196e-06, "loss": 0.0643, "step": 7692 }, { "epoch": 4.07, "learning_rate": 1.7512509663532918e-06, "loss": 0.1051, "step": 7693 }, { "epoch": 4.07, "learning_rate": 1.7493128089097121e-06, "loss": 0.0614, "step": 7694 }, { "epoch": 4.07, "learning_rate": 1.74737562176181e-06, "loss": 0.0918, "step": 7695 }, { "epoch": 4.07, "learning_rate": 1.7454394051374012e-06, "loss": 0.0791, "step": 7696 }, { "epoch": 4.07, "learning_rate": 1.7435041592641922e-06, "loss": 0.0656, "step": 7697 }, { "epoch": 4.07, "learning_rate": 1.7415698843697659e-06, "loss": 0.069, "step": 7698 }, { "epoch": 4.08, "learning_rate": 1.7396365806816019e-06, "loss": 0.0572, "step": 7699 }, { "epoch": 4.08, "learning_rate": 1.7377042484270578e-06, "loss": 0.0662, "step": 7700 }, { "epoch": 4.08, "learning_rate": 1.7357728878333768e-06, "loss": 0.0639, "step": 7701 }, { "epoch": 4.08, "learning_rate": 1.7338424991276959e-06, "loss": 0.059, "step": 7702 }, { "epoch": 4.08, "learning_rate": 1.7319130825370256e-06, "loss": 0.0562, "step": 7703 }, { "epoch": 4.08, "learning_rate": 1.7299846382882747e-06, "loss": 0.0629, "step": 7704 }, { "epoch": 4.08, "learning_rate": 1.728057166608228e-06, "loss": 0.0407, "step": 7705 }, { "epoch": 4.08, "learning_rate": 1.7261306677235623e-06, "loss": 0.1142, "step": 7706 }, { "epoch": 4.08, "learning_rate": 1.7242051418608341e-06, "loss": 0.0902, "step": 7707 }, { "epoch": 4.08, "learning_rate": 1.7222805892464934e-06, "loss": 0.0587, "step": 7708 }, { "epoch": 4.08, "learning_rate": 1.7203570101068678e-06, "loss": 0.1112, "step": 7709 }, { "epoch": 4.08, "learning_rate": 1.7184344046681722e-06, "loss": 0.1116, "step": 7710 }, { "epoch": 4.08, "learning_rate": 1.7165127731565102e-06, "loss": 0.0538, "step": 7711 }, { "epoch": 4.08, "learning_rate": 1.7145921157978717e-06, "loss": 0.0696, "step": 7712 }, { "epoch": 4.08, "learning_rate": 1.7126724328181244e-06, "loss": 0.0588, "step": 7713 }, { "epoch": 4.08, "learning_rate": 1.7107537244430317e-06, "loss": 0.077, "step": 7714 }, { "epoch": 4.08, "learning_rate": 1.7088359908982322e-06, "loss": 0.0732, "step": 7715 }, { "epoch": 4.08, "learning_rate": 1.7069192324092599e-06, "loss": 0.044, "step": 7716 }, { "epoch": 4.08, "learning_rate": 1.7050034492015266e-06, "loss": 0.1053, "step": 7717 }, { "epoch": 4.09, "learning_rate": 1.7030886415003288e-06, "loss": 0.108, "step": 7718 }, { "epoch": 4.09, "learning_rate": 1.7011748095308577e-06, "loss": 0.091, "step": 7719 }, { "epoch": 4.09, "learning_rate": 1.6992619535181766e-06, "loss": 0.0613, "step": 7720 }, { "epoch": 4.09, "learning_rate": 1.6973500736872473e-06, "loss": 0.0591, "step": 7721 }, { "epoch": 4.09, "learning_rate": 1.6954391702629037e-06, "loss": 0.0584, "step": 7722 }, { "epoch": 4.09, "learning_rate": 1.6935292434698746e-06, "loss": 0.0616, "step": 7723 }, { "epoch": 4.09, "learning_rate": 1.6916202935327774e-06, "loss": 0.1179, "step": 7724 }, { "epoch": 4.09, "learning_rate": 1.6897123206760957e-06, "loss": 0.0618, "step": 7725 }, { "epoch": 4.09, "learning_rate": 1.6878053251242177e-06, "loss": 0.0866, "step": 7726 }, { "epoch": 4.09, "learning_rate": 1.6858993071014108e-06, "loss": 0.0756, "step": 7727 }, { "epoch": 4.09, "learning_rate": 1.683994266831821e-06, "loss": 0.0615, "step": 7728 }, { "epoch": 4.09, "learning_rate": 1.6820902045394893e-06, "loss": 0.0668, "step": 7729 }, { "epoch": 4.09, "learning_rate": 1.6801871204483334e-06, "loss": 0.0534, "step": 7730 }, { "epoch": 4.09, "learning_rate": 1.6782850147821629e-06, "loss": 0.0742, "step": 7731 }, { "epoch": 4.09, "learning_rate": 1.6763838877646654e-06, "loss": 0.0638, "step": 7732 }, { "epoch": 4.09, "learning_rate": 1.674483739619418e-06, "loss": 0.0551, "step": 7733 }, { "epoch": 4.09, "learning_rate": 1.672584570569883e-06, "loss": 0.0813, "step": 7734 }, { "epoch": 4.09, "learning_rate": 1.6706863808394025e-06, "loss": 0.0701, "step": 7735 }, { "epoch": 4.09, "learning_rate": 1.6687891706512128e-06, "loss": 0.0791, "step": 7736 }, { "epoch": 4.1, "learning_rate": 1.6668929402284229e-06, "loss": 0.0603, "step": 7737 }, { "epoch": 4.1, "learning_rate": 1.6649976897940357e-06, "loss": 0.0864, "step": 7738 }, { "epoch": 4.1, "learning_rate": 1.6631034195709383e-06, "loss": 0.0629, "step": 7739 }, { "epoch": 4.1, "learning_rate": 1.6612101297818994e-06, "loss": 0.058, "step": 7740 }, { "epoch": 4.1, "learning_rate": 1.6593178206495707e-06, "loss": 0.1111, "step": 7741 }, { "epoch": 4.1, "learning_rate": 1.6574264923964912e-06, "loss": 0.0486, "step": 7742 }, { "epoch": 4.1, "learning_rate": 1.6555361452450858e-06, "loss": 0.0417, "step": 7743 }, { "epoch": 4.1, "learning_rate": 1.653646779417666e-06, "loss": 0.0681, "step": 7744 }, { "epoch": 4.1, "learning_rate": 1.651758395136418e-06, "loss": 0.0586, "step": 7745 }, { "epoch": 4.1, "learning_rate": 1.6498709926234246e-06, "loss": 0.0622, "step": 7746 }, { "epoch": 4.1, "learning_rate": 1.647984572100645e-06, "loss": 0.0482, "step": 7747 }, { "epoch": 4.1, "learning_rate": 1.6460991337899278e-06, "loss": 0.0824, "step": 7748 }, { "epoch": 4.1, "learning_rate": 1.644214677913003e-06, "loss": 0.0591, "step": 7749 }, { "epoch": 4.1, "learning_rate": 1.642331204691483e-06, "loss": 0.0557, "step": 7750 }, { "epoch": 4.1, "learning_rate": 1.6404487143468717e-06, "loss": 0.1019, "step": 7751 }, { "epoch": 4.1, "learning_rate": 1.6385672071005509e-06, "loss": 0.099, "step": 7752 }, { "epoch": 4.1, "learning_rate": 1.6366866831737894e-06, "loss": 0.0425, "step": 7753 }, { "epoch": 4.1, "learning_rate": 1.6348071427877443e-06, "loss": 0.0674, "step": 7754 }, { "epoch": 4.1, "learning_rate": 1.6329285861634469e-06, "loss": 0.054, "step": 7755 }, { "epoch": 4.11, "learning_rate": 1.631051013521826e-06, "loss": 0.057, "step": 7756 }, { "epoch": 4.11, "learning_rate": 1.6291744250836793e-06, "loss": 0.1055, "step": 7757 }, { "epoch": 4.11, "learning_rate": 1.6272988210696993e-06, "loss": 0.0659, "step": 7758 }, { "epoch": 4.11, "learning_rate": 1.6254242017004661e-06, "loss": 0.0525, "step": 7759 }, { "epoch": 4.11, "learning_rate": 1.6235505671964303e-06, "loss": 0.0951, "step": 7760 }, { "epoch": 4.11, "learning_rate": 1.6216779177779419e-06, "loss": 0.0779, "step": 7761 }, { "epoch": 4.11, "learning_rate": 1.6198062536652214e-06, "loss": 0.0564, "step": 7762 }, { "epoch": 4.11, "learning_rate": 1.6179355750783865e-06, "loss": 0.0889, "step": 7763 }, { "epoch": 4.11, "learning_rate": 1.6160658822374276e-06, "loss": 0.0671, "step": 7764 }, { "epoch": 4.11, "learning_rate": 1.6141971753622232e-06, "loss": 0.0284, "step": 7765 }, { "epoch": 4.11, "learning_rate": 1.612329454672541e-06, "loss": 0.0733, "step": 7766 }, { "epoch": 4.11, "learning_rate": 1.610462720388024e-06, "loss": 0.0323, "step": 7767 }, { "epoch": 4.11, "learning_rate": 1.608596972728207e-06, "loss": 0.1163, "step": 7768 }, { "epoch": 4.11, "learning_rate": 1.6067322119125017e-06, "loss": 0.0953, "step": 7769 }, { "epoch": 4.11, "learning_rate": 1.6048684381602097e-06, "loss": 0.0644, "step": 7770 }, { "epoch": 4.11, "learning_rate": 1.6030056516905157e-06, "loss": 0.0683, "step": 7771 }, { "epoch": 4.11, "learning_rate": 1.6011438527224855e-06, "loss": 0.0689, "step": 7772 }, { "epoch": 4.11, "learning_rate": 1.599283041475066e-06, "loss": 0.0557, "step": 7773 }, { "epoch": 4.11, "learning_rate": 1.5974232181670979e-06, "loss": 0.0612, "step": 7774 }, { "epoch": 4.12, "learning_rate": 1.595564383017295e-06, "loss": 0.0425, "step": 7775 }, { "epoch": 4.12, "learning_rate": 1.5937065362442638e-06, "loss": 0.0593, "step": 7776 }, { "epoch": 4.12, "learning_rate": 1.5918496780664872e-06, "loss": 0.0685, "step": 7777 }, { "epoch": 4.12, "learning_rate": 1.5899938087023381e-06, "loss": 0.061, "step": 7778 }, { "epoch": 4.12, "learning_rate": 1.5881389283700654e-06, "loss": 0.1126, "step": 7779 }, { "epoch": 4.12, "learning_rate": 1.5862850372878125e-06, "loss": 0.0617, "step": 7780 }, { "epoch": 4.12, "learning_rate": 1.584432135673596e-06, "loss": 0.0496, "step": 7781 }, { "epoch": 4.12, "learning_rate": 1.5825802237453202e-06, "loss": 0.0846, "step": 7782 }, { "epoch": 4.12, "learning_rate": 1.580729301720776e-06, "loss": 0.0897, "step": 7783 }, { "epoch": 4.12, "learning_rate": 1.578879369817633e-06, "loss": 0.0675, "step": 7784 }, { "epoch": 4.12, "learning_rate": 1.5770304282534453e-06, "loss": 0.048, "step": 7785 }, { "epoch": 4.12, "learning_rate": 1.575182477245658e-06, "loss": 0.0779, "step": 7786 }, { "epoch": 4.12, "learning_rate": 1.573335517011586e-06, "loss": 0.0644, "step": 7787 }, { "epoch": 4.12, "learning_rate": 1.571489547768441e-06, "loss": 0.0406, "step": 7788 }, { "epoch": 4.12, "learning_rate": 1.5696445697333096e-06, "loss": 0.0911, "step": 7789 }, { "epoch": 4.12, "learning_rate": 1.5678005831231623e-06, "loss": 0.0764, "step": 7790 }, { "epoch": 4.12, "learning_rate": 1.5659575881548605e-06, "loss": 0.0571, "step": 7791 }, { "epoch": 4.12, "learning_rate": 1.5641155850451385e-06, "loss": 0.0697, "step": 7792 }, { "epoch": 4.12, "learning_rate": 1.5622745740106239e-06, "loss": 0.0646, "step": 7793 }, { "epoch": 4.13, "learning_rate": 1.5604345552678168e-06, "loss": 0.0685, "step": 7794 }, { "epoch": 4.13, "learning_rate": 1.5585955290331122e-06, "loss": 0.1097, "step": 7795 }, { "epoch": 4.13, "learning_rate": 1.5567574955227849e-06, "loss": 0.1031, "step": 7796 }, { "epoch": 4.13, "learning_rate": 1.5549204549529828e-06, "loss": 0.051, "step": 7797 }, { "epoch": 4.13, "learning_rate": 1.553084407539751e-06, "loss": 0.0464, "step": 7798 }, { "epoch": 4.13, "learning_rate": 1.5512493534990092e-06, "loss": 0.0719, "step": 7799 }, { "epoch": 4.13, "learning_rate": 1.5494152930465644e-06, "loss": 0.0523, "step": 7800 }, { "epoch": 4.13, "learning_rate": 1.5475822263981077e-06, "loss": 0.0714, "step": 7801 }, { "epoch": 4.13, "learning_rate": 1.5457501537692065e-06, "loss": 0.0848, "step": 7802 }, { "epoch": 4.13, "learning_rate": 1.5439190753753207e-06, "loss": 0.0798, "step": 7803 }, { "epoch": 4.13, "learning_rate": 1.542088991431786e-06, "loss": 0.1027, "step": 7804 }, { "epoch": 4.13, "learning_rate": 1.54025990215382e-06, "loss": 0.1334, "step": 7805 }, { "epoch": 4.13, "learning_rate": 1.538431807756534e-06, "loss": 0.066, "step": 7806 }, { "epoch": 4.13, "learning_rate": 1.5366047084549085e-06, "loss": 0.0918, "step": 7807 }, { "epoch": 4.13, "learning_rate": 1.5347786044638191e-06, "loss": 0.0585, "step": 7808 }, { "epoch": 4.13, "learning_rate": 1.5329534959980142e-06, "loss": 0.0836, "step": 7809 }, { "epoch": 4.13, "learning_rate": 1.5311293832721341e-06, "loss": 0.0537, "step": 7810 }, { "epoch": 4.13, "learning_rate": 1.529306266500694e-06, "loss": 0.0802, "step": 7811 }, { "epoch": 4.13, "learning_rate": 1.5274841458980993e-06, "loss": 0.0693, "step": 7812 }, { "epoch": 4.14, "learning_rate": 1.5256630216786327e-06, "loss": 0.0938, "step": 7813 }, { "epoch": 4.14, "learning_rate": 1.523842894056461e-06, "loss": 0.0742, "step": 7814 }, { "epoch": 4.14, "learning_rate": 1.5220237632456335e-06, "loss": 0.0771, "step": 7815 }, { "epoch": 4.14, "learning_rate": 1.5202056294600887e-06, "loss": 0.0677, "step": 7816 }, { "epoch": 4.14, "learning_rate": 1.5183884929136361e-06, "loss": 0.0882, "step": 7817 }, { "epoch": 4.14, "learning_rate": 1.5165723538199783e-06, "loss": 0.0705, "step": 7818 }, { "epoch": 4.14, "learning_rate": 1.5147572123926945e-06, "loss": 0.0852, "step": 7819 }, { "epoch": 4.14, "learning_rate": 1.5129430688452507e-06, "loss": 0.0447, "step": 7820 }, { "epoch": 4.14, "learning_rate": 1.5111299233909926e-06, "loss": 0.0573, "step": 7821 }, { "epoch": 4.14, "learning_rate": 1.5093177762431455e-06, "loss": 0.0732, "step": 7822 }, { "epoch": 4.14, "learning_rate": 1.5075066276148275e-06, "loss": 0.0812, "step": 7823 }, { "epoch": 4.14, "learning_rate": 1.5056964777190275e-06, "loss": 0.0484, "step": 7824 }, { "epoch": 4.14, "learning_rate": 1.5038873267686272e-06, "loss": 0.065, "step": 7825 }, { "epoch": 4.14, "learning_rate": 1.5020791749763808e-06, "loss": 0.0974, "step": 7826 }, { "epoch": 4.14, "learning_rate": 1.5002720225549338e-06, "loss": 0.0718, "step": 7827 }, { "epoch": 4.14, "learning_rate": 1.4984658697168108e-06, "loss": 0.0587, "step": 7828 }, { "epoch": 4.14, "learning_rate": 1.4966607166744184e-06, "loss": 0.0718, "step": 7829 }, { "epoch": 4.14, "learning_rate": 1.4948565636400437e-06, "loss": 0.0837, "step": 7830 }, { "epoch": 4.15, "learning_rate": 1.493053410825861e-06, "loss": 0.0475, "step": 7831 }, { "epoch": 4.15, "learning_rate": 1.4912512584439209e-06, "loss": 0.0837, "step": 7832 }, { "epoch": 4.15, "learning_rate": 1.489450106706164e-06, "loss": 0.085, "step": 7833 }, { "epoch": 4.15, "learning_rate": 1.487649955824405e-06, "loss": 0.0679, "step": 7834 }, { "epoch": 4.15, "learning_rate": 1.4858508060103483e-06, "loss": 0.0818, "step": 7835 }, { "epoch": 4.15, "learning_rate": 1.4840526574755743e-06, "loss": 0.0841, "step": 7836 }, { "epoch": 4.15, "learning_rate": 1.4822555104315516e-06, "loss": 0.0616, "step": 7837 }, { "epoch": 4.15, "learning_rate": 1.480459365089627e-06, "loss": 0.0473, "step": 7838 }, { "epoch": 4.15, "learning_rate": 1.4786642216610258e-06, "loss": 0.0811, "step": 7839 }, { "epoch": 4.15, "learning_rate": 1.476870080356868e-06, "loss": 0.0656, "step": 7840 }, { "epoch": 4.15, "learning_rate": 1.4750769413881405e-06, "loss": 0.0493, "step": 7841 }, { "epoch": 4.15, "learning_rate": 1.4732848049657223e-06, "loss": 0.1021, "step": 7842 }, { "epoch": 4.15, "learning_rate": 1.4714936713003757e-06, "loss": 0.0585, "step": 7843 }, { "epoch": 4.15, "learning_rate": 1.469703540602736e-06, "loss": 0.0608, "step": 7844 }, { "epoch": 4.15, "learning_rate": 1.4679144130833322e-06, "loss": 0.1032, "step": 7845 }, { "epoch": 4.15, "learning_rate": 1.4661262889525596e-06, "loss": 0.0709, "step": 7846 }, { "epoch": 4.15, "learning_rate": 1.4643391684207097e-06, "loss": 0.0603, "step": 7847 }, { "epoch": 4.15, "learning_rate": 1.4625530516979535e-06, "loss": 0.0775, "step": 7848 }, { "epoch": 4.15, "learning_rate": 1.4607679389943375e-06, "loss": 0.0674, "step": 7849 }, { "epoch": 4.16, "learning_rate": 1.4589838305197978e-06, "loss": 0.0347, "step": 7850 }, { "epoch": 4.16, "learning_rate": 1.4572007264841449e-06, "loss": 0.0518, "step": 7851 }, { "epoch": 4.16, "learning_rate": 1.4554186270970782e-06, "loss": 0.0983, "step": 7852 }, { "epoch": 4.16, "learning_rate": 1.4536375325681763e-06, "loss": 0.1292, "step": 7853 }, { "epoch": 4.16, "learning_rate": 1.4518574431068934e-06, "loss": 0.0499, "step": 7854 }, { "epoch": 4.16, "learning_rate": 1.4500783589225787e-06, "loss": 0.1161, "step": 7855 }, { "epoch": 4.16, "learning_rate": 1.4483002802244495e-06, "loss": 0.0559, "step": 7856 }, { "epoch": 4.16, "learning_rate": 1.446523207221614e-06, "loss": 0.1208, "step": 7857 }, { "epoch": 4.16, "learning_rate": 1.4447471401230616e-06, "loss": 0.0518, "step": 7858 }, { "epoch": 4.16, "learning_rate": 1.4429720791376556e-06, "loss": 0.0817, "step": 7859 }, { "epoch": 4.16, "learning_rate": 1.441198024474152e-06, "loss": 0.0606, "step": 7860 }, { "epoch": 4.16, "learning_rate": 1.4394249763411795e-06, "loss": 0.0631, "step": 7861 }, { "epoch": 4.16, "learning_rate": 1.43765293494725e-06, "loss": 0.0475, "step": 7862 }, { "epoch": 4.16, "learning_rate": 1.435881900500764e-06, "loss": 0.0731, "step": 7863 }, { "epoch": 4.16, "learning_rate": 1.4341118732099923e-06, "loss": 0.0844, "step": 7864 }, { "epoch": 4.16, "learning_rate": 1.4323428532830985e-06, "loss": 0.0828, "step": 7865 }, { "epoch": 4.16, "learning_rate": 1.4305748409281185e-06, "loss": 0.0812, "step": 7866 }, { "epoch": 4.16, "learning_rate": 1.4288078363529777e-06, "loss": 0.0433, "step": 7867 }, { "epoch": 4.16, "learning_rate": 1.427041839765475e-06, "loss": 0.0812, "step": 7868 }, { "epoch": 4.17, "learning_rate": 1.4252768513732996e-06, "loss": 0.0553, "step": 7869 }, { "epoch": 4.17, "learning_rate": 1.423512871384013e-06, "loss": 0.0958, "step": 7870 }, { "epoch": 4.17, "learning_rate": 1.4217499000050628e-06, "loss": 0.059, "step": 7871 }, { "epoch": 4.17, "learning_rate": 1.4199879374437798e-06, "loss": 0.0516, "step": 7872 }, { "epoch": 4.17, "learning_rate": 1.4182269839073715e-06, "loss": 0.0899, "step": 7873 }, { "epoch": 4.17, "learning_rate": 1.41646703960293e-06, "loss": 0.0605, "step": 7874 }, { "epoch": 4.17, "learning_rate": 1.414708104737431e-06, "loss": 0.0589, "step": 7875 }, { "epoch": 4.17, "learning_rate": 1.4129501795177236e-06, "loss": 0.0689, "step": 7876 }, { "epoch": 4.17, "learning_rate": 1.4111932641505477e-06, "loss": 0.0435, "step": 7877 }, { "epoch": 4.17, "learning_rate": 1.409437358842517e-06, "loss": 0.1108, "step": 7878 }, { "epoch": 4.17, "learning_rate": 1.4076824638001273e-06, "loss": 0.0452, "step": 7879 }, { "epoch": 4.17, "learning_rate": 1.4059285792297617e-06, "loss": 0.0791, "step": 7880 }, { "epoch": 4.17, "learning_rate": 1.404175705337676e-06, "loss": 0.0672, "step": 7881 }, { "epoch": 4.17, "learning_rate": 1.4024238423300153e-06, "loss": 0.0665, "step": 7882 }, { "epoch": 4.17, "learning_rate": 1.400672990412797e-06, "loss": 0.0708, "step": 7883 }, { "epoch": 4.17, "learning_rate": 1.3989231497919276e-06, "loss": 0.0909, "step": 7884 }, { "epoch": 4.17, "learning_rate": 1.397174320673197e-06, "loss": 0.071, "step": 7885 }, { "epoch": 4.17, "learning_rate": 1.3954265032622583e-06, "loss": 0.1048, "step": 7886 }, { "epoch": 4.17, "learning_rate": 1.3936796977646683e-06, "loss": 0.0962, "step": 7887 }, { "epoch": 4.18, "learning_rate": 1.3919339043858482e-06, "loss": 0.1075, "step": 7888 }, { "epoch": 4.18, "learning_rate": 1.3901891233311093e-06, "loss": 0.0446, "step": 7889 }, { "epoch": 4.18, "learning_rate": 1.388445354805642e-06, "loss": 0.0487, "step": 7890 }, { "epoch": 4.18, "learning_rate": 1.3867025990145133e-06, "loss": 0.1066, "step": 7891 }, { "epoch": 4.18, "learning_rate": 1.3849608561626792e-06, "loss": 0.0636, "step": 7892 }, { "epoch": 4.18, "learning_rate": 1.3832201264549682e-06, "loss": 0.0937, "step": 7893 }, { "epoch": 4.18, "learning_rate": 1.3814804100960933e-06, "loss": 0.0762, "step": 7894 }, { "epoch": 4.18, "learning_rate": 1.3797417072906493e-06, "loss": 0.0664, "step": 7895 }, { "epoch": 4.18, "learning_rate": 1.3780040182431098e-06, "loss": 0.0511, "step": 7896 }, { "epoch": 4.18, "learning_rate": 1.3762673431578322e-06, "loss": 0.0588, "step": 7897 }, { "epoch": 4.18, "learning_rate": 1.3745316822390497e-06, "loss": 0.0676, "step": 7898 }, { "epoch": 4.18, "learning_rate": 1.3727970356908805e-06, "loss": 0.0822, "step": 7899 }, { "epoch": 4.18, "learning_rate": 1.3710634037173243e-06, "loss": 0.0552, "step": 7900 }, { "epoch": 4.18, "learning_rate": 1.3693307865222583e-06, "loss": 0.0734, "step": 7901 }, { "epoch": 4.18, "learning_rate": 1.3675991843094394e-06, "loss": 0.1019, "step": 7902 }, { "epoch": 4.18, "learning_rate": 1.365868597282507e-06, "loss": 0.0809, "step": 7903 }, { "epoch": 4.18, "learning_rate": 1.3641390256449828e-06, "loss": 0.0767, "step": 7904 }, { "epoch": 4.18, "learning_rate": 1.3624104696002682e-06, "loss": 0.0615, "step": 7905 }, { "epoch": 4.18, "learning_rate": 1.360682929351642e-06, "loss": 0.0794, "step": 7906 }, { "epoch": 4.19, "learning_rate": 1.3589564051022707e-06, "loss": 0.0992, "step": 7907 }, { "epoch": 4.19, "learning_rate": 1.3572308970551906e-06, "loss": 0.0661, "step": 7908 }, { "epoch": 4.19, "learning_rate": 1.3555064054133292e-06, "loss": 0.1004, "step": 7909 }, { "epoch": 4.19, "learning_rate": 1.3537829303794892e-06, "loss": 0.1317, "step": 7910 }, { "epoch": 4.19, "learning_rate": 1.3520604721563501e-06, "loss": 0.0826, "step": 7911 }, { "epoch": 4.19, "learning_rate": 1.3503390309464826e-06, "loss": 0.0508, "step": 7912 }, { "epoch": 4.19, "learning_rate": 1.3486186069523255e-06, "loss": 0.0806, "step": 7913 }, { "epoch": 4.19, "learning_rate": 1.3468992003762082e-06, "loss": 0.0721, "step": 7914 }, { "epoch": 4.19, "learning_rate": 1.3451808114203325e-06, "loss": 0.1146, "step": 7915 }, { "epoch": 4.19, "learning_rate": 1.3434634402867853e-06, "loss": 0.0774, "step": 7916 }, { "epoch": 4.19, "learning_rate": 1.3417470871775373e-06, "loss": 0.0415, "step": 7917 }, { "epoch": 4.19, "learning_rate": 1.3400317522944251e-06, "loss": 0.0625, "step": 7918 }, { "epoch": 4.19, "learning_rate": 1.3383174358391816e-06, "loss": 0.0964, "step": 7919 }, { "epoch": 4.19, "learning_rate": 1.3366041380134132e-06, "loss": 0.0565, "step": 7920 }, { "epoch": 4.19, "learning_rate": 1.3348918590186044e-06, "loss": 0.0848, "step": 7921 }, { "epoch": 4.19, "learning_rate": 1.3331805990561263e-06, "loss": 0.0787, "step": 7922 }, { "epoch": 4.19, "learning_rate": 1.3314703583272215e-06, "loss": 0.0585, "step": 7923 }, { "epoch": 4.19, "learning_rate": 1.329761137033021e-06, "loss": 0.0545, "step": 7924 }, { "epoch": 4.19, "learning_rate": 1.3280529353745307e-06, "loss": 0.0747, "step": 7925 }, { "epoch": 4.2, "learning_rate": 1.3263457535526369e-06, "loss": 0.0884, "step": 7926 }, { "epoch": 4.2, "learning_rate": 1.324639591768111e-06, "loss": 0.0571, "step": 7927 }, { "epoch": 4.2, "learning_rate": 1.3229344502215958e-06, "loss": 0.0779, "step": 7928 }, { "epoch": 4.2, "learning_rate": 1.321230329113623e-06, "loss": 0.1048, "step": 7929 }, { "epoch": 4.2, "learning_rate": 1.3195272286445981e-06, "loss": 0.0511, "step": 7930 }, { "epoch": 4.2, "learning_rate": 1.3178251490148086e-06, "loss": 0.0796, "step": 7931 }, { "epoch": 4.2, "learning_rate": 1.3161240904244255e-06, "loss": 0.0707, "step": 7932 }, { "epoch": 4.2, "learning_rate": 1.3144240530734943e-06, "loss": 0.0803, "step": 7933 }, { "epoch": 4.2, "learning_rate": 1.3127250371619394e-06, "loss": 0.0492, "step": 7934 }, { "epoch": 4.2, "learning_rate": 1.3110270428895733e-06, "loss": 0.0809, "step": 7935 }, { "epoch": 4.2, "learning_rate": 1.3093300704560785e-06, "loss": 0.0982, "step": 7936 }, { "epoch": 4.2, "learning_rate": 1.3076341200610265e-06, "loss": 0.0778, "step": 7937 }, { "epoch": 4.2, "learning_rate": 1.30593919190386e-06, "loss": 0.0458, "step": 7938 }, { "epoch": 4.2, "learning_rate": 1.3042452861839094e-06, "loss": 0.0695, "step": 7939 }, { "epoch": 4.2, "learning_rate": 1.3025524031003766e-06, "loss": 0.0788, "step": 7940 }, { "epoch": 4.2, "learning_rate": 1.3008605428523514e-06, "loss": 0.0602, "step": 7941 }, { "epoch": 4.2, "learning_rate": 1.2991697056388019e-06, "loss": 0.0872, "step": 7942 }, { "epoch": 4.2, "learning_rate": 1.2974798916585652e-06, "loss": 0.0714, "step": 7943 }, { "epoch": 4.2, "learning_rate": 1.2957911011103751e-06, "loss": 0.0542, "step": 7944 }, { "epoch": 4.21, "learning_rate": 1.29410333419283e-06, "loss": 0.1332, "step": 7945 }, { "epoch": 4.21, "learning_rate": 1.2924165911044161e-06, "loss": 0.0991, "step": 7946 }, { "epoch": 4.21, "learning_rate": 1.2907308720435019e-06, "loss": 0.0931, "step": 7947 }, { "epoch": 4.21, "learning_rate": 1.2890461772083252e-06, "loss": 0.0801, "step": 7948 }, { "epoch": 4.21, "learning_rate": 1.287362506797013e-06, "loss": 0.0416, "step": 7949 }, { "epoch": 4.21, "learning_rate": 1.285679861007566e-06, "loss": 0.095, "step": 7950 }, { "epoch": 4.21, "learning_rate": 1.2839982400378658e-06, "loss": 0.0923, "step": 7951 }, { "epoch": 4.21, "learning_rate": 1.2823176440856766e-06, "loss": 0.0485, "step": 7952 }, { "epoch": 4.21, "learning_rate": 1.2806380733486356e-06, "loss": 0.0707, "step": 7953 }, { "epoch": 4.21, "learning_rate": 1.2789595280242684e-06, "loss": 0.0509, "step": 7954 }, { "epoch": 4.21, "learning_rate": 1.2772820083099702e-06, "loss": 0.0713, "step": 7955 }, { "epoch": 4.21, "learning_rate": 1.2756055144030244e-06, "loss": 0.0569, "step": 7956 }, { "epoch": 4.21, "learning_rate": 1.2739300465005843e-06, "loss": 0.1053, "step": 7957 }, { "epoch": 4.21, "learning_rate": 1.2722556047996948e-06, "loss": 0.0543, "step": 7958 }, { "epoch": 4.21, "learning_rate": 1.2705821894972702e-06, "loss": 0.0596, "step": 7959 }, { "epoch": 4.21, "learning_rate": 1.2689098007901035e-06, "loss": 0.0411, "step": 7960 }, { "epoch": 4.21, "learning_rate": 1.267238438874875e-06, "loss": 0.0478, "step": 7961 }, { "epoch": 4.21, "learning_rate": 1.2655681039481405e-06, "loss": 0.0966, "step": 7962 }, { "epoch": 4.21, "learning_rate": 1.2638987962063298e-06, "loss": 0.0653, "step": 7963 }, { "epoch": 4.22, "learning_rate": 1.2622305158457626e-06, "loss": 0.0926, "step": 7964 }, { "epoch": 4.22, "learning_rate": 1.2605632630626253e-06, "loss": 0.0713, "step": 7965 }, { "epoch": 4.22, "learning_rate": 1.2588970380529952e-06, "loss": 0.0738, "step": 7966 }, { "epoch": 4.22, "learning_rate": 1.2572318410128215e-06, "loss": 0.0589, "step": 7967 }, { "epoch": 4.22, "learning_rate": 1.2555676721379318e-06, "loss": 0.0663, "step": 7968 }, { "epoch": 4.22, "learning_rate": 1.2539045316240405e-06, "loss": 0.0841, "step": 7969 }, { "epoch": 4.22, "learning_rate": 1.2522424196667294e-06, "loss": 0.0671, "step": 7970 }, { "epoch": 4.22, "learning_rate": 1.2505813364614726e-06, "loss": 0.0575, "step": 7971 }, { "epoch": 4.22, "learning_rate": 1.2489212822036112e-06, "loss": 0.0627, "step": 7972 }, { "epoch": 4.22, "learning_rate": 1.247262257088372e-06, "loss": 0.0655, "step": 7973 }, { "epoch": 4.22, "learning_rate": 1.2456042613108644e-06, "loss": 0.0567, "step": 7974 }, { "epoch": 4.22, "learning_rate": 1.2439472950660637e-06, "loss": 0.1097, "step": 7975 }, { "epoch": 4.22, "learning_rate": 1.242291358548835e-06, "loss": 0.0741, "step": 7976 }, { "epoch": 4.22, "learning_rate": 1.2406364519539226e-06, "loss": 0.0602, "step": 7977 }, { "epoch": 4.22, "learning_rate": 1.2389825754759422e-06, "loss": 0.0638, "step": 7978 }, { "epoch": 4.22, "learning_rate": 1.2373297293093967e-06, "loss": 0.0687, "step": 7979 }, { "epoch": 4.22, "learning_rate": 1.2356779136486586e-06, "loss": 0.0667, "step": 7980 }, { "epoch": 4.22, "learning_rate": 1.2340271286879901e-06, "loss": 0.0693, "step": 7981 }, { "epoch": 4.22, "learning_rate": 1.232377374621524e-06, "loss": 0.0638, "step": 7982 }, { "epoch": 4.23, "learning_rate": 1.2307286516432715e-06, "loss": 0.0904, "step": 7983 }, { "epoch": 4.23, "learning_rate": 1.2290809599471299e-06, "loss": 0.0756, "step": 7984 }, { "epoch": 4.23, "learning_rate": 1.2274342997268672e-06, "loss": 0.0784, "step": 7985 }, { "epoch": 4.23, "learning_rate": 1.225788671176138e-06, "loss": 0.0919, "step": 7986 }, { "epoch": 4.23, "learning_rate": 1.2241440744884647e-06, "loss": 0.0529, "step": 7987 }, { "epoch": 4.23, "learning_rate": 1.2225005098572596e-06, "loss": 0.0597, "step": 7988 }, { "epoch": 4.23, "learning_rate": 1.2208579774758101e-06, "loss": 0.0912, "step": 7989 }, { "epoch": 4.23, "learning_rate": 1.2192164775372784e-06, "loss": 0.0581, "step": 7990 }, { "epoch": 4.23, "learning_rate": 1.2175760102347078e-06, "loss": 0.068, "step": 7991 }, { "epoch": 4.23, "learning_rate": 1.2159365757610187e-06, "loss": 0.0881, "step": 7992 }, { "epoch": 4.23, "learning_rate": 1.2142981743090143e-06, "loss": 0.0709, "step": 7993 }, { "epoch": 4.23, "learning_rate": 1.2126608060713729e-06, "loss": 0.0649, "step": 7994 }, { "epoch": 4.23, "learning_rate": 1.2110244712406517e-06, "loss": 0.098, "step": 7995 }, { "epoch": 4.23, "learning_rate": 1.2093891700092874e-06, "loss": 0.066, "step": 7996 }, { "epoch": 4.23, "learning_rate": 1.2077549025695911e-06, "loss": 0.0612, "step": 7997 }, { "epoch": 4.23, "learning_rate": 1.206121669113761e-06, "loss": 0.0571, "step": 7998 }, { "epoch": 4.23, "learning_rate": 1.2044894698338649e-06, "loss": 0.0828, "step": 7999 }, { "epoch": 4.23, "learning_rate": 1.2028583049218513e-06, "loss": 0.0627, "step": 8000 }, { "epoch": 4.24, "learning_rate": 1.201228174569552e-06, "loss": 0.0632, "step": 8001 }, { "epoch": 4.24, "learning_rate": 1.1995990789686685e-06, "loss": 0.063, "step": 8002 }, { "epoch": 4.24, "learning_rate": 1.1979710183107873e-06, "loss": 0.047, "step": 8003 }, { "epoch": 4.24, "learning_rate": 1.1963439927873743e-06, "loss": 0.0607, "step": 8004 }, { "epoch": 4.24, "learning_rate": 1.1947180025897664e-06, "loss": 0.0532, "step": 8005 }, { "epoch": 4.24, "learning_rate": 1.1930930479091884e-06, "loss": 0.0799, "step": 8006 }, { "epoch": 4.24, "learning_rate": 1.1914691289367307e-06, "loss": 0.0733, "step": 8007 }, { "epoch": 4.24, "learning_rate": 1.189846245863372e-06, "loss": 0.1018, "step": 8008 }, { "epoch": 4.24, "learning_rate": 1.188224398879969e-06, "loss": 0.0661, "step": 8009 }, { "epoch": 4.24, "learning_rate": 1.1866035881772508e-06, "loss": 0.0757, "step": 8010 }, { "epoch": 4.24, "learning_rate": 1.184983813945829e-06, "loss": 0.0478, "step": 8011 }, { "epoch": 4.24, "learning_rate": 1.1833650763761905e-06, "loss": 0.0797, "step": 8012 }, { "epoch": 4.24, "learning_rate": 1.181747375658705e-06, "loss": 0.1093, "step": 8013 }, { "epoch": 4.24, "learning_rate": 1.1801307119836137e-06, "loss": 0.1189, "step": 8014 }, { "epoch": 4.24, "learning_rate": 1.1785150855410387e-06, "loss": 0.0535, "step": 8015 }, { "epoch": 4.24, "learning_rate": 1.1769004965209852e-06, "loss": 0.0572, "step": 8016 }, { "epoch": 4.24, "learning_rate": 1.1752869451133254e-06, "loss": 0.0621, "step": 8017 }, { "epoch": 4.24, "learning_rate": 1.1736744315078219e-06, "loss": 0.0512, "step": 8018 }, { "epoch": 4.24, "learning_rate": 1.1720629558941044e-06, "loss": 0.1028, "step": 8019 }, { "epoch": 4.25, "learning_rate": 1.1704525184616867e-06, "loss": 0.078, "step": 8020 }, { "epoch": 4.25, "learning_rate": 1.1688431193999628e-06, "loss": 0.0994, "step": 8021 }, { "epoch": 4.25, "learning_rate": 1.167234758898198e-06, "loss": 0.0718, "step": 8022 }, { "epoch": 4.25, "learning_rate": 1.1656274371455355e-06, "loss": 0.0797, "step": 8023 }, { "epoch": 4.25, "learning_rate": 1.1640211543310053e-06, "loss": 0.1033, "step": 8024 }, { "epoch": 4.25, "learning_rate": 1.162415910643503e-06, "loss": 0.0736, "step": 8025 }, { "epoch": 4.25, "learning_rate": 1.1608117062718138e-06, "loss": 0.1178, "step": 8026 }, { "epoch": 4.25, "learning_rate": 1.1592085414045895e-06, "loss": 0.0673, "step": 8027 }, { "epoch": 4.25, "learning_rate": 1.1576064162303713e-06, "loss": 0.04, "step": 8028 }, { "epoch": 4.25, "learning_rate": 1.1560053309375652e-06, "loss": 0.0621, "step": 8029 }, { "epoch": 4.25, "learning_rate": 1.1544052857144684e-06, "loss": 0.0348, "step": 8030 }, { "epoch": 4.25, "learning_rate": 1.1528062807492457e-06, "loss": 0.078, "step": 8031 }, { "epoch": 4.25, "learning_rate": 1.1512083162299415e-06, "loss": 0.0908, "step": 8032 }, { "epoch": 4.25, "learning_rate": 1.1496113923444817e-06, "loss": 0.0423, "step": 8033 }, { "epoch": 4.25, "learning_rate": 1.1480155092806666e-06, "loss": 0.0793, "step": 8034 }, { "epoch": 4.25, "learning_rate": 1.1464206672261735e-06, "loss": 0.065, "step": 8035 }, { "epoch": 4.25, "learning_rate": 1.1448268663685636e-06, "loss": 0.0484, "step": 8036 }, { "epoch": 4.25, "learning_rate": 1.1432341068952647e-06, "loss": 0.0877, "step": 8037 }, { "epoch": 4.25, "learning_rate": 1.1416423889935923e-06, "loss": 0.0841, "step": 8038 }, { "epoch": 4.26, "learning_rate": 1.140051712850735e-06, "loss": 0.0707, "step": 8039 }, { "epoch": 4.26, "learning_rate": 1.138462078653756e-06, "loss": 0.0755, "step": 8040 }, { "epoch": 4.26, "learning_rate": 1.1368734865896025e-06, "loss": 0.0532, "step": 8041 }, { "epoch": 4.26, "learning_rate": 1.135285936845093e-06, "loss": 0.0749, "step": 8042 }, { "epoch": 4.26, "learning_rate": 1.1336994296069292e-06, "loss": 0.0513, "step": 8043 }, { "epoch": 4.26, "learning_rate": 1.1321139650616842e-06, "loss": 0.0627, "step": 8044 }, { "epoch": 4.26, "learning_rate": 1.130529543395813e-06, "loss": 0.0835, "step": 8045 }, { "epoch": 4.26, "learning_rate": 1.1289461647956512e-06, "loss": 0.0772, "step": 8046 }, { "epoch": 4.26, "learning_rate": 1.1273638294473966e-06, "loss": 0.0947, "step": 8047 }, { "epoch": 4.26, "learning_rate": 1.1257825375371434e-06, "loss": 0.077, "step": 8048 }, { "epoch": 4.26, "learning_rate": 1.1242022892508498e-06, "loss": 0.0835, "step": 8049 }, { "epoch": 4.26, "learning_rate": 1.1226230847743558e-06, "loss": 0.0473, "step": 8050 }, { "epoch": 4.26, "learning_rate": 1.1210449242933841e-06, "loss": 0.0871, "step": 8051 }, { "epoch": 4.26, "learning_rate": 1.1194678079935216e-06, "loss": 0.0792, "step": 8052 }, { "epoch": 4.26, "learning_rate": 1.1178917360602471e-06, "loss": 0.0507, "step": 8053 }, { "epoch": 4.26, "learning_rate": 1.116316708678905e-06, "loss": 0.0912, "step": 8054 }, { "epoch": 4.26, "learning_rate": 1.114742726034721e-06, "loss": 0.0678, "step": 8055 }, { "epoch": 4.26, "learning_rate": 1.1131697883128024e-06, "loss": 0.1217, "step": 8056 }, { "epoch": 4.26, "learning_rate": 1.1115978956981243e-06, "loss": 0.0941, "step": 8057 }, { "epoch": 4.27, "learning_rate": 1.1100270483755492e-06, "loss": 0.0471, "step": 8058 }, { "epoch": 4.27, "learning_rate": 1.1084572465298072e-06, "loss": 0.0532, "step": 8059 }, { "epoch": 4.27, "learning_rate": 1.1068884903455135e-06, "loss": 0.0822, "step": 8060 }, { "epoch": 4.27, "learning_rate": 1.105320780007152e-06, "loss": 0.1184, "step": 8061 }, { "epoch": 4.27, "learning_rate": 1.1037541156990938e-06, "loss": 0.062, "step": 8062 }, { "epoch": 4.27, "learning_rate": 1.1021884976055797e-06, "loss": 0.0489, "step": 8063 }, { "epoch": 4.27, "learning_rate": 1.1006239259107253e-06, "loss": 0.0879, "step": 8064 }, { "epoch": 4.27, "learning_rate": 1.0990604007985307e-06, "loss": 0.0789, "step": 8065 }, { "epoch": 4.27, "learning_rate": 1.0974979224528693e-06, "loss": 0.0652, "step": 8066 }, { "epoch": 4.27, "learning_rate": 1.09593649105749e-06, "loss": 0.0597, "step": 8067 }, { "epoch": 4.27, "learning_rate": 1.0943761067960223e-06, "loss": 0.0928, "step": 8068 }, { "epoch": 4.27, "learning_rate": 1.0928167698519666e-06, "loss": 0.0849, "step": 8069 }, { "epoch": 4.27, "learning_rate": 1.0912584804087078e-06, "loss": 0.0662, "step": 8070 }, { "epoch": 4.27, "learning_rate": 1.089701238649501e-06, "loss": 0.0872, "step": 8071 }, { "epoch": 4.27, "learning_rate": 1.0881450447574793e-06, "loss": 0.0706, "step": 8072 }, { "epoch": 4.27, "learning_rate": 1.0865898989156575e-06, "loss": 0.0639, "step": 8073 }, { "epoch": 4.27, "learning_rate": 1.0850358013069196e-06, "loss": 0.0735, "step": 8074 }, { "epoch": 4.27, "learning_rate": 1.0834827521140346e-06, "loss": 0.084, "step": 8075 }, { "epoch": 4.27, "learning_rate": 1.0819307515196397e-06, "loss": 0.1015, "step": 8076 }, { "epoch": 4.28, "learning_rate": 1.0803797997062548e-06, "loss": 0.068, "step": 8077 }, { "epoch": 4.28, "learning_rate": 1.0788298968562772e-06, "loss": 0.0681, "step": 8078 }, { "epoch": 4.28, "learning_rate": 1.0772810431519753e-06, "loss": 0.065, "step": 8079 }, { "epoch": 4.28, "learning_rate": 1.075733238775496e-06, "loss": 0.0886, "step": 8080 }, { "epoch": 4.28, "learning_rate": 1.0741864839088678e-06, "loss": 0.0672, "step": 8081 }, { "epoch": 4.28, "learning_rate": 1.0726407787339865e-06, "loss": 0.0582, "step": 8082 }, { "epoch": 4.28, "learning_rate": 1.0710961234326357e-06, "loss": 0.0734, "step": 8083 }, { "epoch": 4.28, "learning_rate": 1.0695525181864652e-06, "loss": 0.0458, "step": 8084 }, { "epoch": 4.28, "learning_rate": 1.068009963177008e-06, "loss": 0.0571, "step": 8085 }, { "epoch": 4.28, "learning_rate": 1.06646845858567e-06, "loss": 0.0662, "step": 8086 }, { "epoch": 4.28, "learning_rate": 1.0649280045937372e-06, "loss": 0.0623, "step": 8087 }, { "epoch": 4.28, "learning_rate": 1.063388601382368e-06, "loss": 0.0669, "step": 8088 }, { "epoch": 4.28, "learning_rate": 1.0618502491325977e-06, "loss": 0.1175, "step": 8089 }, { "epoch": 4.28, "learning_rate": 1.0603129480253426e-06, "loss": 0.0692, "step": 8090 }, { "epoch": 4.28, "learning_rate": 1.0587766982413882e-06, "loss": 0.0885, "step": 8091 }, { "epoch": 4.28, "learning_rate": 1.0572414999614022e-06, "loss": 0.0604, "step": 8092 }, { "epoch": 4.28, "learning_rate": 1.055707353365929e-06, "loss": 0.0592, "step": 8093 }, { "epoch": 4.28, "learning_rate": 1.0541742586353831e-06, "loss": 0.0969, "step": 8094 }, { "epoch": 4.28, "learning_rate": 1.0526422159500637e-06, "loss": 0.0462, "step": 8095 }, { "epoch": 4.29, "learning_rate": 1.0511112254901368e-06, "loss": 0.0522, "step": 8096 }, { "epoch": 4.29, "learning_rate": 1.0495812874356503e-06, "loss": 0.0732, "step": 8097 }, { "epoch": 4.29, "learning_rate": 1.0480524019665317e-06, "loss": 0.0763, "step": 8098 }, { "epoch": 4.29, "learning_rate": 1.046524569262576e-06, "loss": 0.0658, "step": 8099 }, { "epoch": 4.29, "learning_rate": 1.0449977895034635e-06, "loss": 0.0719, "step": 8100 }, { "epoch": 4.29, "learning_rate": 1.0434720628687423e-06, "loss": 0.0643, "step": 8101 }, { "epoch": 4.29, "learning_rate": 1.041947389537844e-06, "loss": 0.054, "step": 8102 }, { "epoch": 4.29, "learning_rate": 1.040423769690071e-06, "loss": 0.0624, "step": 8103 }, { "epoch": 4.29, "learning_rate": 1.038901203504602e-06, "loss": 0.0947, "step": 8104 }, { "epoch": 4.29, "learning_rate": 1.0373796911604983e-06, "loss": 0.1094, "step": 8105 }, { "epoch": 4.29, "learning_rate": 1.0358592328366867e-06, "loss": 0.0633, "step": 8106 }, { "epoch": 4.29, "learning_rate": 1.03433982871198e-06, "loss": 0.0923, "step": 8107 }, { "epoch": 4.29, "learning_rate": 1.0328214789650626e-06, "loss": 0.1173, "step": 8108 }, { "epoch": 4.29, "learning_rate": 1.031304183774493e-06, "loss": 0.0976, "step": 8109 }, { "epoch": 4.29, "learning_rate": 1.0297879433187108e-06, "loss": 0.0663, "step": 8110 }, { "epoch": 4.29, "learning_rate": 1.0282727577760265e-06, "loss": 0.0814, "step": 8111 }, { "epoch": 4.29, "learning_rate": 1.0267586273246277e-06, "loss": 0.0542, "step": 8112 }, { "epoch": 4.29, "learning_rate": 1.025245552142582e-06, "loss": 0.112, "step": 8113 }, { "epoch": 4.29, "learning_rate": 1.0237335324078257e-06, "loss": 0.0922, "step": 8114 }, { "epoch": 4.3, "learning_rate": 1.0222225682981802e-06, "loss": 0.1069, "step": 8115 }, { "epoch": 4.3, "learning_rate": 1.0207126599913319e-06, "loss": 0.0769, "step": 8116 }, { "epoch": 4.3, "learning_rate": 1.0192038076648536e-06, "loss": 0.0617, "step": 8117 }, { "epoch": 4.3, "learning_rate": 1.0176960114961864e-06, "loss": 0.0801, "step": 8118 }, { "epoch": 4.3, "learning_rate": 1.0161892716626509e-06, "loss": 0.0823, "step": 8119 }, { "epoch": 4.3, "learning_rate": 1.014683588341443e-06, "loss": 0.0638, "step": 8120 }, { "epoch": 4.3, "learning_rate": 1.01317896170963e-06, "loss": 0.0813, "step": 8121 }, { "epoch": 4.3, "learning_rate": 1.0116753919441635e-06, "loss": 0.0476, "step": 8122 }, { "epoch": 4.3, "learning_rate": 1.0101728792218623e-06, "loss": 0.0788, "step": 8123 }, { "epoch": 4.3, "learning_rate": 1.0086714237194272e-06, "loss": 0.0912, "step": 8124 }, { "epoch": 4.3, "learning_rate": 1.0071710256134315e-06, "loss": 0.0632, "step": 8125 }, { "epoch": 4.3, "learning_rate": 1.0056716850803238e-06, "loss": 0.064, "step": 8126 }, { "epoch": 4.3, "learning_rate": 1.0041734022964312e-06, "loss": 0.0699, "step": 8127 }, { "epoch": 4.3, "learning_rate": 1.0026761774379533e-06, "loss": 0.085, "step": 8128 }, { "epoch": 4.3, "learning_rate": 1.0011800106809645e-06, "loss": 0.1111, "step": 8129 }, { "epoch": 4.3, "learning_rate": 9.9968490220142e-07, "loss": 0.0679, "step": 8130 }, { "epoch": 4.3, "learning_rate": 9.981908521751426e-07, "loss": 0.1218, "step": 8131 }, { "epoch": 4.3, "learning_rate": 9.966978607778422e-07, "loss": 0.0914, "step": 8132 }, { "epoch": 4.3, "learning_rate": 9.952059281850912e-07, "loss": 0.0734, "step": 8133 }, { "epoch": 4.31, "learning_rate": 9.937150545723451e-07, "loss": 0.0704, "step": 8134 }, { "epoch": 4.31, "learning_rate": 9.92225240114938e-07, "loss": 0.0906, "step": 8135 }, { "epoch": 4.31, "learning_rate": 9.907364849880674e-07, "loss": 0.0716, "step": 8136 }, { "epoch": 4.31, "learning_rate": 9.892487893668178e-07, "loss": 0.0566, "step": 8137 }, { "epoch": 4.31, "learning_rate": 9.877621534261439e-07, "loss": 0.1087, "step": 8138 }, { "epoch": 4.31, "learning_rate": 9.862765773408755e-07, "loss": 0.0551, "step": 8139 }, { "epoch": 4.31, "learning_rate": 9.84792061285723e-07, "loss": 0.1039, "step": 8140 }, { "epoch": 4.31, "learning_rate": 9.833086054352637e-07, "loss": 0.108, "step": 8141 }, { "epoch": 4.31, "learning_rate": 9.818262099639586e-07, "loss": 0.0818, "step": 8142 }, { "epoch": 4.31, "learning_rate": 9.803448750461375e-07, "loss": 0.0835, "step": 8143 }, { "epoch": 4.31, "learning_rate": 9.788646008560066e-07, "loss": 0.0762, "step": 8144 }, { "epoch": 4.31, "learning_rate": 9.77385387567653e-07, "loss": 0.0734, "step": 8145 }, { "epoch": 4.31, "learning_rate": 9.759072353550315e-07, "loss": 0.0408, "step": 8146 }, { "epoch": 4.31, "learning_rate": 9.744301443919779e-07, "loss": 0.0587, "step": 8147 }, { "epoch": 4.31, "learning_rate": 9.729541148521982e-07, "loss": 0.086, "step": 8148 }, { "epoch": 4.31, "learning_rate": 9.714791469092766e-07, "loss": 0.0537, "step": 8149 }, { "epoch": 4.31, "learning_rate": 9.700052407366755e-07, "loss": 0.0643, "step": 8150 }, { "epoch": 4.31, "learning_rate": 9.685323965077254e-07, "loss": 0.0692, "step": 8151 }, { "epoch": 4.31, "learning_rate": 9.670606143956373e-07, "loss": 0.063, "step": 8152 }, { "epoch": 4.32, "learning_rate": 9.65589894573492e-07, "loss": 0.0761, "step": 8153 }, { "epoch": 4.32, "learning_rate": 9.64120237214251e-07, "loss": 0.074, "step": 8154 }, { "epoch": 4.32, "learning_rate": 9.626516424907506e-07, "loss": 0.1139, "step": 8155 }, { "epoch": 4.32, "learning_rate": 9.61184110575697e-07, "loss": 0.0585, "step": 8156 }, { "epoch": 4.32, "learning_rate": 9.597176416416777e-07, "loss": 0.0574, "step": 8157 }, { "epoch": 4.32, "learning_rate": 9.58252235861149e-07, "loss": 0.0842, "step": 8158 }, { "epoch": 4.32, "learning_rate": 9.567878934064478e-07, "loss": 0.0772, "step": 8159 }, { "epoch": 4.32, "learning_rate": 9.553246144497829e-07, "loss": 0.0586, "step": 8160 }, { "epoch": 4.32, "learning_rate": 9.538623991632368e-07, "loss": 0.0792, "step": 8161 }, { "epoch": 4.32, "learning_rate": 9.524012477187705e-07, "loss": 0.0604, "step": 8162 }, { "epoch": 4.32, "learning_rate": 9.509411602882157e-07, "loss": 0.0534, "step": 8163 }, { "epoch": 4.32, "learning_rate": 9.494821370432849e-07, "loss": 0.0782, "step": 8164 }, { "epoch": 4.32, "learning_rate": 9.480241781555577e-07, "loss": 0.0716, "step": 8165 }, { "epoch": 4.32, "learning_rate": 9.465672837964957e-07, "loss": 0.0667, "step": 8166 }, { "epoch": 4.32, "learning_rate": 9.451114541374351e-07, "loss": 0.0345, "step": 8167 }, { "epoch": 4.32, "learning_rate": 9.436566893495769e-07, "loss": 0.0653, "step": 8168 }, { "epoch": 4.32, "learning_rate": 9.422029896040064e-07, "loss": 0.0656, "step": 8169 }, { "epoch": 4.32, "learning_rate": 9.407503550716856e-07, "loss": 0.0981, "step": 8170 }, { "epoch": 4.32, "learning_rate": 9.392987859234414e-07, "loss": 0.0694, "step": 8171 }, { "epoch": 4.33, "learning_rate": 9.378482823299861e-07, "loss": 0.0528, "step": 8172 }, { "epoch": 4.33, "learning_rate": 9.363988444618966e-07, "loss": 0.0702, "step": 8173 }, { "epoch": 4.33, "learning_rate": 9.349504724896341e-07, "loss": 0.0414, "step": 8174 }, { "epoch": 4.33, "learning_rate": 9.33503166583527e-07, "loss": 0.067, "step": 8175 }, { "epoch": 4.33, "learning_rate": 9.320569269137791e-07, "loss": 0.0489, "step": 8176 }, { "epoch": 4.33, "learning_rate": 9.306117536504756e-07, "loss": 0.0526, "step": 8177 }, { "epoch": 4.33, "learning_rate": 9.291676469635669e-07, "loss": 0.1105, "step": 8178 }, { "epoch": 4.33, "learning_rate": 9.277246070228862e-07, "loss": 0.0792, "step": 8179 }, { "epoch": 4.33, "learning_rate": 9.262826339981345e-07, "loss": 0.0589, "step": 8180 }, { "epoch": 4.33, "learning_rate": 9.248417280588917e-07, "loss": 0.0643, "step": 8181 }, { "epoch": 4.33, "learning_rate": 9.234018893746132e-07, "loss": 0.0728, "step": 8182 }, { "epoch": 4.33, "learning_rate": 9.219631181146249e-07, "loss": 0.0727, "step": 8183 }, { "epoch": 4.33, "learning_rate": 9.205254144481268e-07, "loss": 0.0621, "step": 8184 }, { "epoch": 4.33, "learning_rate": 9.190887785441982e-07, "loss": 0.0641, "step": 8185 }, { "epoch": 4.33, "learning_rate": 9.176532105717883e-07, "loss": 0.0812, "step": 8186 }, { "epoch": 4.33, "learning_rate": 9.162187106997245e-07, "loss": 0.084, "step": 8187 }, { "epoch": 4.33, "learning_rate": 9.147852790967038e-07, "loss": 0.0647, "step": 8188 }, { "epoch": 4.33, "learning_rate": 9.133529159313037e-07, "loss": 0.0482, "step": 8189 }, { "epoch": 4.34, "learning_rate": 9.119216213719695e-07, "loss": 0.059, "step": 8190 }, { "epoch": 4.34, "learning_rate": 9.104913955870254e-07, "loss": 0.0635, "step": 8191 }, { "epoch": 4.34, "learning_rate": 9.090622387446723e-07, "loss": 0.0676, "step": 8192 }, { "epoch": 4.34, "learning_rate": 9.076341510129738e-07, "loss": 0.0555, "step": 8193 }, { "epoch": 4.34, "learning_rate": 9.062071325598809e-07, "loss": 0.0622, "step": 8194 }, { "epoch": 4.34, "learning_rate": 9.047811835532117e-07, "loss": 0.0677, "step": 8195 }, { "epoch": 4.34, "learning_rate": 9.033563041606608e-07, "loss": 0.0626, "step": 8196 }, { "epoch": 4.34, "learning_rate": 9.019324945497976e-07, "loss": 0.0665, "step": 8197 }, { "epoch": 4.34, "learning_rate": 9.005097548880626e-07, "loss": 0.0639, "step": 8198 }, { "epoch": 4.34, "learning_rate": 8.990880853427763e-07, "loss": 0.0493, "step": 8199 }, { "epoch": 4.34, "learning_rate": 8.976674860811263e-07, "loss": 0.0711, "step": 8200 }, { "epoch": 4.34, "learning_rate": 8.962479572701765e-07, "loss": 0.0612, "step": 8201 }, { "epoch": 4.34, "learning_rate": 8.948294990768702e-07, "loss": 0.0441, "step": 8202 }, { "epoch": 4.34, "learning_rate": 8.934121116680161e-07, "loss": 0.0721, "step": 8203 }, { "epoch": 4.34, "learning_rate": 8.919957952103064e-07, "loss": 0.0615, "step": 8204 }, { "epoch": 4.34, "learning_rate": 8.905805498702969e-07, "loss": 0.0859, "step": 8205 }, { "epoch": 4.34, "learning_rate": 8.891663758144286e-07, "loss": 0.0984, "step": 8206 }, { "epoch": 4.34, "learning_rate": 8.877532732090067e-07, "loss": 0.0608, "step": 8207 }, { "epoch": 4.34, "learning_rate": 8.863412422202178e-07, "loss": 0.0783, "step": 8208 }, { "epoch": 4.35, "learning_rate": 8.849302830141182e-07, "loss": 0.094, "step": 8209 }, { "epoch": 4.35, "learning_rate": 8.835203957566363e-07, "loss": 0.1023, "step": 8210 }, { "epoch": 4.35, "learning_rate": 8.821115806135794e-07, "loss": 0.0768, "step": 8211 }, { "epoch": 4.35, "learning_rate": 8.807038377506294e-07, "loss": 0.0716, "step": 8212 }, { "epoch": 4.35, "learning_rate": 8.792971673333351e-07, "loss": 0.1073, "step": 8213 }, { "epoch": 4.35, "learning_rate": 8.778915695271284e-07, "loss": 0.0657, "step": 8214 }, { "epoch": 4.35, "learning_rate": 8.76487044497304e-07, "loss": 0.113, "step": 8215 }, { "epoch": 4.35, "learning_rate": 8.750835924090429e-07, "loss": 0.0918, "step": 8216 }, { "epoch": 4.35, "learning_rate": 8.736812134273898e-07, "loss": 0.091, "step": 8217 }, { "epoch": 4.35, "learning_rate": 8.72279907717265e-07, "loss": 0.075, "step": 8218 }, { "epoch": 4.35, "learning_rate": 8.708796754434701e-07, "loss": 0.1161, "step": 8219 }, { "epoch": 4.35, "learning_rate": 8.694805167706711e-07, "loss": 0.0601, "step": 8220 }, { "epoch": 4.35, "learning_rate": 8.680824318634129e-07, "loss": 0.0774, "step": 8221 }, { "epoch": 4.35, "learning_rate": 8.666854208861119e-07, "loss": 0.0827, "step": 8222 }, { "epoch": 4.35, "learning_rate": 8.65289484003059e-07, "loss": 0.0733, "step": 8223 }, { "epoch": 4.35, "learning_rate": 8.638946213784227e-07, "loss": 0.0635, "step": 8224 }, { "epoch": 4.35, "learning_rate": 8.625008331762352e-07, "loss": 0.0574, "step": 8225 }, { "epoch": 4.35, "learning_rate": 8.61108119560412e-07, "loss": 0.0573, "step": 8226 }, { "epoch": 4.35, "learning_rate": 8.5971648069474e-07, "loss": 0.0603, "step": 8227 }, { "epoch": 4.36, "learning_rate": 8.583259167428737e-07, "loss": 0.0663, "step": 8228 }, { "epoch": 4.36, "learning_rate": 8.569364278683522e-07, "loss": 0.1108, "step": 8229 }, { "epoch": 4.36, "learning_rate": 8.555480142345762e-07, "loss": 0.0842, "step": 8230 }, { "epoch": 4.36, "learning_rate": 8.541606760048304e-07, "loss": 0.0403, "step": 8231 }, { "epoch": 4.36, "learning_rate": 8.527744133422655e-07, "loss": 0.0883, "step": 8232 }, { "epoch": 4.36, "learning_rate": 8.513892264099067e-07, "loss": 0.0524, "step": 8233 }, { "epoch": 4.36, "learning_rate": 8.50005115370659e-07, "loss": 0.0505, "step": 8234 }, { "epoch": 4.36, "learning_rate": 8.486220803872924e-07, "loss": 0.0913, "step": 8235 }, { "epoch": 4.36, "learning_rate": 8.472401216224579e-07, "loss": 0.07, "step": 8236 }, { "epoch": 4.36, "learning_rate": 8.458592392386722e-07, "loss": 0.0566, "step": 8237 }, { "epoch": 4.36, "learning_rate": 8.444794333983319e-07, "loss": 0.0767, "step": 8238 }, { "epoch": 4.36, "learning_rate": 8.431007042637073e-07, "loss": 0.0745, "step": 8239 }, { "epoch": 4.36, "learning_rate": 8.417230519969355e-07, "loss": 0.0934, "step": 8240 }, { "epoch": 4.36, "learning_rate": 8.403464767600322e-07, "loss": 0.0405, "step": 8241 }, { "epoch": 4.36, "learning_rate": 8.389709787148836e-07, "loss": 0.0825, "step": 8242 }, { "epoch": 4.36, "learning_rate": 8.375965580232514e-07, "loss": 0.046, "step": 8243 }, { "epoch": 4.36, "learning_rate": 8.362232148467731e-07, "loss": 0.0497, "step": 8244 }, { "epoch": 4.36, "learning_rate": 8.348509493469525e-07, "loss": 0.0845, "step": 8245 }, { "epoch": 4.36, "learning_rate": 8.334797616851731e-07, "loss": 0.1146, "step": 8246 }, { "epoch": 4.37, "learning_rate": 8.321096520226868e-07, "loss": 0.0683, "step": 8247 }, { "epoch": 4.37, "learning_rate": 8.307406205206226e-07, "loss": 0.0566, "step": 8248 }, { "epoch": 4.37, "learning_rate": 8.293726673399816e-07, "loss": 0.0987, "step": 8249 }, { "epoch": 4.37, "learning_rate": 8.28005792641634e-07, "loss": 0.0729, "step": 8250 }, { "epoch": 4.37, "learning_rate": 8.266399965863303e-07, "loss": 0.0805, "step": 8251 }, { "epoch": 4.37, "learning_rate": 8.252752793346885e-07, "loss": 0.0722, "step": 8252 }, { "epoch": 4.37, "learning_rate": 8.239116410472025e-07, "loss": 0.0705, "step": 8253 }, { "epoch": 4.37, "learning_rate": 8.225490818842419e-07, "loss": 0.0767, "step": 8254 }, { "epoch": 4.37, "learning_rate": 8.211876020060394e-07, "loss": 0.0589, "step": 8255 }, { "epoch": 4.37, "learning_rate": 8.19827201572716e-07, "loss": 0.0588, "step": 8256 }, { "epoch": 4.37, "learning_rate": 8.184678807442481e-07, "loss": 0.0478, "step": 8257 }, { "epoch": 4.37, "learning_rate": 8.17109639680499e-07, "loss": 0.0891, "step": 8258 }, { "epoch": 4.37, "learning_rate": 8.157524785412018e-07, "loss": 0.0681, "step": 8259 }, { "epoch": 4.37, "learning_rate": 8.14396397485957e-07, "loss": 0.0531, "step": 8260 }, { "epoch": 4.37, "learning_rate": 8.130413966742467e-07, "loss": 0.07, "step": 8261 }, { "epoch": 4.37, "learning_rate": 8.11687476265417e-07, "loss": 0.0789, "step": 8262 }, { "epoch": 4.37, "learning_rate": 8.103346364186948e-07, "loss": 0.0777, "step": 8263 }, { "epoch": 4.37, "learning_rate": 8.089828772931752e-07, "loss": 0.0652, "step": 8264 }, { "epoch": 4.37, "learning_rate": 8.076321990478265e-07, "loss": 0.0633, "step": 8265 }, { "epoch": 4.38, "learning_rate": 8.062826018414938e-07, "loss": 0.0659, "step": 8266 }, { "epoch": 4.38, "learning_rate": 8.049340858328869e-07, "loss": 0.0554, "step": 8267 }, { "epoch": 4.38, "learning_rate": 8.035866511806001e-07, "loss": 0.0947, "step": 8268 }, { "epoch": 4.38, "learning_rate": 8.022402980430899e-07, "loss": 0.0469, "step": 8269 }, { "epoch": 4.38, "learning_rate": 8.008950265786897e-07, "loss": 0.0609, "step": 8270 }, { "epoch": 4.38, "learning_rate": 7.995508369456096e-07, "loss": 0.0528, "step": 8271 }, { "epoch": 4.38, "learning_rate": 7.982077293019264e-07, "loss": 0.0701, "step": 8272 }, { "epoch": 4.38, "learning_rate": 7.968657038055894e-07, "loss": 0.0829, "step": 8273 }, { "epoch": 4.38, "learning_rate": 7.955247606144278e-07, "loss": 0.0942, "step": 8274 }, { "epoch": 4.38, "learning_rate": 7.941848998861357e-07, "loss": 0.0695, "step": 8275 }, { "epoch": 4.38, "learning_rate": 7.928461217782857e-07, "loss": 0.1028, "step": 8276 }, { "epoch": 4.38, "learning_rate": 7.915084264483175e-07, "loss": 0.0764, "step": 8277 }, { "epoch": 4.38, "learning_rate": 7.901718140535497e-07, "loss": 0.0576, "step": 8278 }, { "epoch": 4.38, "learning_rate": 7.888362847511666e-07, "loss": 0.0513, "step": 8279 }, { "epoch": 4.38, "learning_rate": 7.875018386982336e-07, "loss": 0.0564, "step": 8280 }, { "epoch": 4.38, "learning_rate": 7.861684760516807e-07, "loss": 0.0588, "step": 8281 }, { "epoch": 4.38, "learning_rate": 7.848361969683127e-07, "loss": 0.0611, "step": 8282 }, { "epoch": 4.38, "learning_rate": 7.835050016048107e-07, "loss": 0.0747, "step": 8283 }, { "epoch": 4.38, "learning_rate": 7.82174890117724e-07, "loss": 0.0791, "step": 8284 }, { "epoch": 4.39, "learning_rate": 7.808458626634763e-07, "loss": 0.1078, "step": 8285 }, { "epoch": 4.39, "learning_rate": 7.79517919398366e-07, "loss": 0.0824, "step": 8286 }, { "epoch": 4.39, "learning_rate": 7.78191060478558e-07, "loss": 0.0676, "step": 8287 }, { "epoch": 4.39, "learning_rate": 7.768652860600967e-07, "loss": 0.0638, "step": 8288 }, { "epoch": 4.39, "learning_rate": 7.755405962988938e-07, "loss": 0.0588, "step": 8289 }, { "epoch": 4.39, "learning_rate": 7.742169913507336e-07, "loss": 0.0747, "step": 8290 }, { "epoch": 4.39, "learning_rate": 7.728944713712772e-07, "loss": 0.0728, "step": 8291 }, { "epoch": 4.39, "learning_rate": 7.715730365160534e-07, "loss": 0.0671, "step": 8292 }, { "epoch": 4.39, "learning_rate": 7.70252686940468e-07, "loss": 0.0805, "step": 8293 }, { "epoch": 4.39, "learning_rate": 7.689334227997924e-07, "loss": 0.0564, "step": 8294 }, { "epoch": 4.39, "learning_rate": 7.676152442491769e-07, "loss": 0.071, "step": 8295 }, { "epoch": 4.39, "learning_rate": 7.662981514436452e-07, "loss": 0.0947, "step": 8296 }, { "epoch": 4.39, "learning_rate": 7.649821445380822e-07, "loss": 0.0574, "step": 8297 }, { "epoch": 4.39, "learning_rate": 7.636672236872578e-07, "loss": 0.0439, "step": 8298 }, { "epoch": 4.39, "learning_rate": 7.623533890458068e-07, "loss": 0.0811, "step": 8299 }, { "epoch": 4.39, "learning_rate": 7.610406407682391e-07, "loss": 0.0779, "step": 8300 }, { "epoch": 4.39, "learning_rate": 7.59728979008939e-07, "loss": 0.0727, "step": 8301 }, { "epoch": 4.39, "learning_rate": 7.584184039221554e-07, "loss": 0.0637, "step": 8302 }, { "epoch": 4.39, "learning_rate": 7.571089156620193e-07, "loss": 0.0708, "step": 8303 }, { "epoch": 4.4, "learning_rate": 7.558005143825254e-07, "loss": 0.0537, "step": 8304 }, { "epoch": 4.4, "learning_rate": 7.544932002375449e-07, "loss": 0.0825, "step": 8305 }, { "epoch": 4.4, "learning_rate": 7.531869733808206e-07, "loss": 0.0842, "step": 8306 }, { "epoch": 4.4, "learning_rate": 7.518818339659661e-07, "loss": 0.0705, "step": 8307 }, { "epoch": 4.4, "learning_rate": 7.50577782146471e-07, "loss": 0.1064, "step": 8308 }, { "epoch": 4.4, "learning_rate": 7.492748180756903e-07, "loss": 0.0728, "step": 8309 }, { "epoch": 4.4, "learning_rate": 7.479729419068582e-07, "loss": 0.1036, "step": 8310 }, { "epoch": 4.4, "learning_rate": 7.466721537930744e-07, "loss": 0.0378, "step": 8311 }, { "epoch": 4.4, "learning_rate": 7.453724538873186e-07, "loss": 0.0526, "step": 8312 }, { "epoch": 4.4, "learning_rate": 7.440738423424343e-07, "loss": 0.0725, "step": 8313 }, { "epoch": 4.4, "learning_rate": 7.427763193111392e-07, "loss": 0.0851, "step": 8314 }, { "epoch": 4.4, "learning_rate": 7.41479884946027e-07, "loss": 0.071, "step": 8315 }, { "epoch": 4.4, "learning_rate": 7.401845393995621e-07, "loss": 0.064, "step": 8316 }, { "epoch": 4.4, "learning_rate": 7.388902828240763e-07, "loss": 0.0529, "step": 8317 }, { "epoch": 4.4, "learning_rate": 7.375971153717787e-07, "loss": 0.1242, "step": 8318 }, { "epoch": 4.4, "learning_rate": 7.363050371947456e-07, "loss": 0.0735, "step": 8319 }, { "epoch": 4.4, "learning_rate": 7.350140484449308e-07, "loss": 0.0518, "step": 8320 }, { "epoch": 4.4, "learning_rate": 7.337241492741553e-07, "loss": 0.0677, "step": 8321 }, { "epoch": 4.4, "learning_rate": 7.324353398341133e-07, "loss": 0.0549, "step": 8322 }, { "epoch": 4.41, "learning_rate": 7.311476202763712e-07, "loss": 0.0529, "step": 8323 }, { "epoch": 4.41, "learning_rate": 7.298609907523668e-07, "loss": 0.0862, "step": 8324 }, { "epoch": 4.41, "learning_rate": 7.285754514134124e-07, "loss": 0.0996, "step": 8325 }, { "epoch": 4.41, "learning_rate": 7.272910024106872e-07, "loss": 0.0908, "step": 8326 }, { "epoch": 4.41, "learning_rate": 7.260076438952446e-07, "loss": 0.0588, "step": 8327 }, { "epoch": 4.41, "learning_rate": 7.247253760180117e-07, "loss": 0.0707, "step": 8328 }, { "epoch": 4.41, "learning_rate": 7.234441989297858e-07, "loss": 0.0669, "step": 8329 }, { "epoch": 4.41, "learning_rate": 7.221641127812329e-07, "loss": 0.0897, "step": 8330 }, { "epoch": 4.41, "learning_rate": 7.208851177228959e-07, "loss": 0.0551, "step": 8331 }, { "epoch": 4.41, "learning_rate": 7.196072139051846e-07, "loss": 0.0503, "step": 8332 }, { "epoch": 4.41, "learning_rate": 7.183304014783866e-07, "loss": 0.0694, "step": 8333 }, { "epoch": 4.41, "learning_rate": 7.170546805926526e-07, "loss": 0.0975, "step": 8334 }, { "epoch": 4.41, "learning_rate": 7.157800513980139e-07, "loss": 0.0851, "step": 8335 }, { "epoch": 4.41, "learning_rate": 7.145065140443652e-07, "loss": 0.0528, "step": 8336 }, { "epoch": 4.41, "learning_rate": 7.132340686814809e-07, "loss": 0.0453, "step": 8337 }, { "epoch": 4.41, "learning_rate": 7.119627154590014e-07, "loss": 0.1155, "step": 8338 }, { "epoch": 4.41, "learning_rate": 7.106924545264371e-07, "loss": 0.0689, "step": 8339 }, { "epoch": 4.41, "learning_rate": 7.094232860331773e-07, "loss": 0.0554, "step": 8340 }, { "epoch": 4.41, "learning_rate": 7.08155210128475e-07, "loss": 0.0549, "step": 8341 }, { "epoch": 4.42, "learning_rate": 7.068882269614608e-07, "loss": 0.0797, "step": 8342 }, { "epoch": 4.42, "learning_rate": 7.056223366811332e-07, "loss": 0.0673, "step": 8343 }, { "epoch": 4.42, "learning_rate": 7.043575394363633e-07, "loss": 0.0651, "step": 8344 }, { "epoch": 4.42, "learning_rate": 7.030938353758965e-07, "loss": 0.0793, "step": 8345 }, { "epoch": 4.42, "learning_rate": 7.018312246483394e-07, "loss": 0.0599, "step": 8346 }, { "epoch": 4.42, "learning_rate": 7.005697074021822e-07, "loss": 0.0864, "step": 8347 }, { "epoch": 4.42, "learning_rate": 6.993092837857829e-07, "loss": 0.0604, "step": 8348 }, { "epoch": 4.42, "learning_rate": 6.980499539473662e-07, "loss": 0.0879, "step": 8349 }, { "epoch": 4.42, "learning_rate": 6.967917180350337e-07, "loss": 0.0489, "step": 8350 }, { "epoch": 4.42, "learning_rate": 6.955345761967546e-07, "loss": 0.0519, "step": 8351 }, { "epoch": 4.42, "learning_rate": 6.942785285803732e-07, "loss": 0.0568, "step": 8352 }, { "epoch": 4.42, "learning_rate": 6.930235753336023e-07, "loss": 0.0661, "step": 8353 }, { "epoch": 4.42, "learning_rate": 6.917697166040238e-07, "loss": 0.0938, "step": 8354 }, { "epoch": 4.42, "learning_rate": 6.905169525390986e-07, "loss": 0.0899, "step": 8355 }, { "epoch": 4.42, "learning_rate": 6.892652832861491e-07, "loss": 0.0517, "step": 8356 }, { "epoch": 4.42, "learning_rate": 6.880147089923761e-07, "loss": 0.097, "step": 8357 }, { "epoch": 4.42, "learning_rate": 6.867652298048522e-07, "loss": 0.0683, "step": 8358 }, { "epoch": 4.42, "learning_rate": 6.855168458705131e-07, "loss": 0.0563, "step": 8359 }, { "epoch": 4.43, "learning_rate": 6.842695573361746e-07, "loss": 0.0638, "step": 8360 }, { "epoch": 4.43, "learning_rate": 6.830233643485207e-07, "loss": 0.0691, "step": 8361 }, { "epoch": 4.43, "learning_rate": 6.817782670541018e-07, "loss": 0.0714, "step": 8362 }, { "epoch": 4.43, "learning_rate": 6.805342655993475e-07, "loss": 0.081, "step": 8363 }, { "epoch": 4.43, "learning_rate": 6.792913601305529e-07, "loss": 0.1089, "step": 8364 }, { "epoch": 4.43, "learning_rate": 6.780495507938867e-07, "loss": 0.0651, "step": 8365 }, { "epoch": 4.43, "learning_rate": 6.768088377353854e-07, "loss": 0.0666, "step": 8366 }, { "epoch": 4.43, "learning_rate": 6.755692211009635e-07, "loss": 0.0779, "step": 8367 }, { "epoch": 4.43, "learning_rate": 6.743307010363986e-07, "loss": 0.0866, "step": 8368 }, { "epoch": 4.43, "learning_rate": 6.730932776873456e-07, "loss": 0.0787, "step": 8369 }, { "epoch": 4.43, "learning_rate": 6.718569511993256e-07, "loss": 0.0524, "step": 8370 }, { "epoch": 4.43, "learning_rate": 6.706217217177324e-07, "loss": 0.0791, "step": 8371 }, { "epoch": 4.43, "learning_rate": 6.693875893878332e-07, "loss": 0.0688, "step": 8372 }, { "epoch": 4.43, "learning_rate": 6.681545543547619e-07, "loss": 0.0722, "step": 8373 }, { "epoch": 4.43, "learning_rate": 6.669226167635279e-07, "loss": 0.1108, "step": 8374 }, { "epoch": 4.43, "learning_rate": 6.656917767590087e-07, "loss": 0.0672, "step": 8375 }, { "epoch": 4.43, "learning_rate": 6.64462034485952e-07, "loss": 0.0874, "step": 8376 }, { "epoch": 4.43, "learning_rate": 6.632333900889798e-07, "loss": 0.0527, "step": 8377 }, { "epoch": 4.43, "learning_rate": 6.62005843712582e-07, "loss": 0.0581, "step": 8378 }, { "epoch": 4.44, "learning_rate": 6.607793955011188e-07, "loss": 0.0956, "step": 8379 }, { "epoch": 4.44, "learning_rate": 6.595540455988247e-07, "loss": 0.097, "step": 8380 }, { "epoch": 4.44, "learning_rate": 6.583297941498013e-07, "loss": 0.0514, "step": 8381 }, { "epoch": 4.44, "learning_rate": 6.571066412980253e-07, "loss": 0.0608, "step": 8382 }, { "epoch": 4.44, "learning_rate": 6.558845871873387e-07, "loss": 0.0624, "step": 8383 }, { "epoch": 4.44, "learning_rate": 6.546636319614586e-07, "loss": 0.0851, "step": 8384 }, { "epoch": 4.44, "learning_rate": 6.534437757639744e-07, "loss": 0.0457, "step": 8385 }, { "epoch": 4.44, "learning_rate": 6.522250187383383e-07, "loss": 0.0772, "step": 8386 }, { "epoch": 4.44, "learning_rate": 6.510073610278822e-07, "loss": 0.0861, "step": 8387 }, { "epoch": 4.44, "learning_rate": 6.497908027758004e-07, "loss": 0.0852, "step": 8388 }, { "epoch": 4.44, "learning_rate": 6.485753441251663e-07, "loss": 0.0703, "step": 8389 }, { "epoch": 4.44, "learning_rate": 6.473609852189211e-07, "loss": 0.0896, "step": 8390 }, { "epoch": 4.44, "learning_rate": 6.461477261998705e-07, "loss": 0.0636, "step": 8391 }, { "epoch": 4.44, "learning_rate": 6.449355672107016e-07, "loss": 0.0799, "step": 8392 }, { "epoch": 4.44, "learning_rate": 6.437245083939636e-07, "loss": 0.0769, "step": 8393 }, { "epoch": 4.44, "learning_rate": 6.42514549892077e-07, "loss": 0.0617, "step": 8394 }, { "epoch": 4.44, "learning_rate": 6.413056918473404e-07, "loss": 0.0619, "step": 8395 }, { "epoch": 4.44, "learning_rate": 6.40097934401912e-07, "loss": 0.0646, "step": 8396 }, { "epoch": 4.44, "learning_rate": 6.388912776978307e-07, "loss": 0.1014, "step": 8397 }, { "epoch": 4.45, "learning_rate": 6.376857218769994e-07, "loss": 0.0602, "step": 8398 }, { "epoch": 4.45, "learning_rate": 6.364812670811926e-07, "loss": 0.0991, "step": 8399 }, { "epoch": 4.45, "learning_rate": 6.352779134520603e-07, "loss": 0.0934, "step": 8400 }, { "epoch": 4.45, "learning_rate": 6.34075661131116e-07, "loss": 0.0621, "step": 8401 }, { "epoch": 4.45, "learning_rate": 6.328745102597478e-07, "loss": 0.0739, "step": 8402 }, { "epoch": 4.45, "learning_rate": 6.316744609792114e-07, "loss": 0.0619, "step": 8403 }, { "epoch": 4.45, "learning_rate": 6.304755134306362e-07, "loss": 0.058, "step": 8404 }, { "epoch": 4.45, "learning_rate": 6.292776677550216e-07, "loss": 0.0584, "step": 8405 }, { "epoch": 4.45, "learning_rate": 6.280809240932328e-07, "loss": 0.0814, "step": 8406 }, { "epoch": 4.45, "learning_rate": 6.268852825860139e-07, "loss": 0.0851, "step": 8407 }, { "epoch": 4.45, "learning_rate": 6.256907433739712e-07, "loss": 0.0575, "step": 8408 }, { "epoch": 4.45, "learning_rate": 6.244973065975856e-07, "loss": 0.0498, "step": 8409 }, { "epoch": 4.45, "learning_rate": 6.233049723972084e-07, "loss": 0.0675, "step": 8410 }, { "epoch": 4.45, "learning_rate": 6.221137409130573e-07, "loss": 0.1041, "step": 8411 }, { "epoch": 4.45, "learning_rate": 6.209236122852269e-07, "loss": 0.0926, "step": 8412 }, { "epoch": 4.45, "learning_rate": 6.197345866536753e-07, "loss": 0.0696, "step": 8413 }, { "epoch": 4.45, "learning_rate": 6.185466641582371e-07, "loss": 0.1305, "step": 8414 }, { "epoch": 4.45, "learning_rate": 6.173598449386109e-07, "loss": 0.0561, "step": 8415 }, { "epoch": 4.45, "learning_rate": 6.161741291343703e-07, "loss": 0.1008, "step": 8416 }, { "epoch": 4.46, "learning_rate": 6.149895168849607e-07, "loss": 0.0904, "step": 8417 }, { "epoch": 4.46, "learning_rate": 6.138060083296893e-07, "loss": 0.0971, "step": 8418 }, { "epoch": 4.46, "learning_rate": 6.126236036077405e-07, "loss": 0.0631, "step": 8419 }, { "epoch": 4.46, "learning_rate": 6.114423028581695e-07, "loss": 0.0655, "step": 8420 }, { "epoch": 4.46, "learning_rate": 6.102621062198966e-07, "loss": 0.0996, "step": 8421 }, { "epoch": 4.46, "learning_rate": 6.090830138317172e-07, "loss": 0.0695, "step": 8422 }, { "epoch": 4.46, "learning_rate": 6.079050258322927e-07, "loss": 0.0938, "step": 8423 }, { "epoch": 4.46, "learning_rate": 6.06728142360159e-07, "loss": 0.0532, "step": 8424 }, { "epoch": 4.46, "learning_rate": 6.055523635537186e-07, "loss": 0.0817, "step": 8425 }, { "epoch": 4.46, "learning_rate": 6.043776895512432e-07, "loss": 0.1099, "step": 8426 }, { "epoch": 4.46, "learning_rate": 6.032041204908801e-07, "loss": 0.0605, "step": 8427 }, { "epoch": 4.46, "learning_rate": 6.020316565106399e-07, "loss": 0.0616, "step": 8428 }, { "epoch": 4.46, "learning_rate": 6.008602977484102e-07, "loss": 0.0804, "step": 8429 }, { "epoch": 4.46, "learning_rate": 5.996900443419395e-07, "loss": 0.103, "step": 8430 }, { "epoch": 4.46, "learning_rate": 5.985208964288569e-07, "loss": 0.0583, "step": 8431 }, { "epoch": 4.46, "learning_rate": 5.973528541466544e-07, "loss": 0.0722, "step": 8432 }, { "epoch": 4.46, "learning_rate": 5.961859176326957e-07, "loss": 0.0483, "step": 8433 }, { "epoch": 4.46, "learning_rate": 5.95020087024214e-07, "loss": 0.0774, "step": 8434 }, { "epoch": 4.46, "learning_rate": 5.938553624583155e-07, "loss": 0.0747, "step": 8435 }, { "epoch": 4.47, "learning_rate": 5.926917440719693e-07, "loss": 0.0585, "step": 8436 }, { "epoch": 4.47, "learning_rate": 5.91529232002025e-07, "loss": 0.0825, "step": 8437 }, { "epoch": 4.47, "learning_rate": 5.903678263851898e-07, "loss": 0.0411, "step": 8438 }, { "epoch": 4.47, "learning_rate": 5.892075273580533e-07, "loss": 0.0637, "step": 8439 }, { "epoch": 4.47, "learning_rate": 5.880483350570632e-07, "loss": 0.1099, "step": 8440 }, { "epoch": 4.47, "learning_rate": 5.86890249618548e-07, "loss": 0.0649, "step": 8441 }, { "epoch": 4.47, "learning_rate": 5.857332711786967e-07, "loss": 0.0638, "step": 8442 }, { "epoch": 4.47, "learning_rate": 5.845773998735727e-07, "loss": 0.1062, "step": 8443 }, { "epoch": 4.47, "learning_rate": 5.834226358391093e-07, "loss": 0.0557, "step": 8444 }, { "epoch": 4.47, "learning_rate": 5.822689792111081e-07, "loss": 0.106, "step": 8445 }, { "epoch": 4.47, "learning_rate": 5.811164301252403e-07, "loss": 0.0596, "step": 8446 }, { "epoch": 4.47, "learning_rate": 5.799649887170522e-07, "loss": 0.0668, "step": 8447 }, { "epoch": 4.47, "learning_rate": 5.788146551219497e-07, "loss": 0.1, "step": 8448 }, { "epoch": 4.47, "learning_rate": 5.776654294752181e-07, "loss": 0.0715, "step": 8449 }, { "epoch": 4.47, "learning_rate": 5.765173119120071e-07, "loss": 0.0882, "step": 8450 }, { "epoch": 4.47, "learning_rate": 5.753703025673363e-07, "loss": 0.0832, "step": 8451 }, { "epoch": 4.47, "learning_rate": 5.742244015760979e-07, "loss": 0.0765, "step": 8452 }, { "epoch": 4.47, "learning_rate": 5.730796090730506e-07, "loss": 0.0587, "step": 8453 }, { "epoch": 4.47, "learning_rate": 5.719359251928247e-07, "loss": 0.049, "step": 8454 }, { "epoch": 4.48, "learning_rate": 5.707933500699192e-07, "loss": 0.0795, "step": 8455 }, { "epoch": 4.48, "learning_rate": 5.696518838387033e-07, "loss": 0.0792, "step": 8456 }, { "epoch": 4.48, "learning_rate": 5.68511526633414e-07, "loss": 0.0893, "step": 8457 }, { "epoch": 4.48, "learning_rate": 5.67372278588163e-07, "loss": 0.0912, "step": 8458 }, { "epoch": 4.48, "learning_rate": 5.662341398369253e-07, "loss": 0.0784, "step": 8459 }, { "epoch": 4.48, "learning_rate": 5.650971105135461e-07, "loss": 0.0785, "step": 8460 }, { "epoch": 4.48, "learning_rate": 5.639611907517451e-07, "loss": 0.0604, "step": 8461 }, { "epoch": 4.48, "learning_rate": 5.628263806851086e-07, "loss": 0.0877, "step": 8462 }, { "epoch": 4.48, "learning_rate": 5.616926804470913e-07, "loss": 0.1025, "step": 8463 }, { "epoch": 4.48, "learning_rate": 5.605600901710184e-07, "loss": 0.078, "step": 8464 }, { "epoch": 4.48, "learning_rate": 5.594286099900836e-07, "loss": 0.0886, "step": 8465 }, { "epoch": 4.48, "learning_rate": 5.582982400373549e-07, "loss": 0.0623, "step": 8466 }, { "epoch": 4.48, "learning_rate": 5.571689804457614e-07, "loss": 0.107, "step": 8467 }, { "epoch": 4.48, "learning_rate": 5.56040831348107e-07, "loss": 0.0583, "step": 8468 }, { "epoch": 4.48, "learning_rate": 5.549137928770664e-07, "loss": 0.0679, "step": 8469 }, { "epoch": 4.48, "learning_rate": 5.537878651651774e-07, "loss": 0.0991, "step": 8470 }, { "epoch": 4.48, "learning_rate": 5.526630483448558e-07, "loss": 0.0727, "step": 8471 }, { "epoch": 4.48, "learning_rate": 5.515393425483784e-07, "loss": 0.0597, "step": 8472 }, { "epoch": 4.48, "learning_rate": 5.504167479078959e-07, "loss": 0.073, "step": 8473 }, { "epoch": 4.49, "learning_rate": 5.492952645554306e-07, "loss": 0.0777, "step": 8474 }, { "epoch": 4.49, "learning_rate": 5.481748926228659e-07, "loss": 0.0734, "step": 8475 }, { "epoch": 4.49, "learning_rate": 5.470556322419618e-07, "loss": 0.0465, "step": 8476 }, { "epoch": 4.49, "learning_rate": 5.459374835443465e-07, "loss": 0.0778, "step": 8477 }, { "epoch": 4.49, "learning_rate": 5.448204466615147e-07, "loss": 0.1225, "step": 8478 }, { "epoch": 4.49, "learning_rate": 5.437045217248338e-07, "loss": 0.0809, "step": 8479 }, { "epoch": 4.49, "learning_rate": 5.425897088655352e-07, "loss": 0.075, "step": 8480 }, { "epoch": 4.49, "learning_rate": 5.414760082147264e-07, "loss": 0.0785, "step": 8481 }, { "epoch": 4.49, "learning_rate": 5.403634199033802e-07, "loss": 0.1178, "step": 8482 }, { "epoch": 4.49, "learning_rate": 5.392519440623356e-07, "loss": 0.0594, "step": 8483 }, { "epoch": 4.49, "learning_rate": 5.381415808223089e-07, "loss": 0.0676, "step": 8484 }, { "epoch": 4.49, "learning_rate": 5.370323303138758e-07, "loss": 0.0709, "step": 8485 }, { "epoch": 4.49, "learning_rate": 5.359241926674918e-07, "loss": 0.0763, "step": 8486 }, { "epoch": 4.49, "learning_rate": 5.348171680134706e-07, "loss": 0.0509, "step": 8487 }, { "epoch": 4.49, "learning_rate": 5.337112564820024e-07, "loss": 0.0744, "step": 8488 }, { "epoch": 4.49, "learning_rate": 5.326064582031465e-07, "loss": 0.0752, "step": 8489 }, { "epoch": 4.49, "learning_rate": 5.315027733068279e-07, "loss": 0.0556, "step": 8490 }, { "epoch": 4.49, "learning_rate": 5.304002019228405e-07, "loss": 0.1397, "step": 8491 }, { "epoch": 4.49, "learning_rate": 5.292987441808484e-07, "loss": 0.0512, "step": 8492 }, { "epoch": 4.5, "learning_rate": 5.281984002103857e-07, "loss": 0.0617, "step": 8493 }, { "epoch": 4.5, "learning_rate": 5.270991701408568e-07, "loss": 0.0736, "step": 8494 }, { "epoch": 4.5, "learning_rate": 5.260010541015304e-07, "loss": 0.0521, "step": 8495 }, { "epoch": 4.5, "learning_rate": 5.249040522215487e-07, "loss": 0.0665, "step": 8496 }, { "epoch": 4.5, "learning_rate": 5.238081646299198e-07, "loss": 0.0671, "step": 8497 }, { "epoch": 4.5, "learning_rate": 5.227133914555249e-07, "loss": 0.069, "step": 8498 }, { "epoch": 4.5, "learning_rate": 5.216197328271078e-07, "loss": 0.0777, "step": 8499 }, { "epoch": 4.5, "learning_rate": 5.205271888732854e-07, "loss": 0.0737, "step": 8500 }, { "epoch": 4.5, "learning_rate": 5.194357597225452e-07, "loss": 0.0512, "step": 8501 }, { "epoch": 4.5, "learning_rate": 5.183454455032377e-07, "loss": 0.0734, "step": 8502 }, { "epoch": 4.5, "learning_rate": 5.172562463435871e-07, "loss": 0.0897, "step": 8503 }, { "epoch": 4.5, "learning_rate": 5.161681623716874e-07, "loss": 0.093, "step": 8504 }, { "epoch": 4.5, "learning_rate": 5.150811937154965e-07, "loss": 0.0863, "step": 8505 }, { "epoch": 4.5, "learning_rate": 5.139953405028486e-07, "loss": 0.0516, "step": 8506 }, { "epoch": 4.5, "learning_rate": 5.129106028614339e-07, "loss": 0.0679, "step": 8507 }, { "epoch": 4.5, "learning_rate": 5.118269809188248e-07, "loss": 0.0678, "step": 8508 }, { "epoch": 4.5, "learning_rate": 5.107444748024582e-07, "loss": 0.0807, "step": 8509 }, { "epoch": 4.5, "learning_rate": 5.096630846396355e-07, "loss": 0.0637, "step": 8510 }, { "epoch": 4.5, "learning_rate": 5.085828105575319e-07, "loss": 0.0793, "step": 8511 }, { "epoch": 4.51, "learning_rate": 5.07503652683189e-07, "loss": 0.0681, "step": 8512 }, { "epoch": 4.51, "learning_rate": 5.064256111435184e-07, "loss": 0.0544, "step": 8513 }, { "epoch": 4.51, "learning_rate": 5.05348686065299e-07, "loss": 0.0448, "step": 8514 }, { "epoch": 4.51, "learning_rate": 5.042728775751793e-07, "loss": 0.0891, "step": 8515 }, { "epoch": 4.51, "learning_rate": 5.03198185799677e-07, "loss": 0.1016, "step": 8516 }, { "epoch": 4.51, "learning_rate": 5.021246108651756e-07, "loss": 0.0902, "step": 8517 }, { "epoch": 4.51, "learning_rate": 5.010521528979329e-07, "loss": 0.0727, "step": 8518 }, { "epoch": 4.51, "learning_rate": 4.99980812024069e-07, "loss": 0.1026, "step": 8519 }, { "epoch": 4.51, "learning_rate": 4.989105883695755e-07, "loss": 0.0766, "step": 8520 }, { "epoch": 4.51, "learning_rate": 4.978414820603161e-07, "loss": 0.0718, "step": 8521 }, { "epoch": 4.51, "learning_rate": 4.967734932220158e-07, "loss": 0.0786, "step": 8522 }, { "epoch": 4.51, "learning_rate": 4.957066219802731e-07, "loss": 0.0884, "step": 8523 }, { "epoch": 4.51, "learning_rate": 4.946408684605564e-07, "loss": 0.0508, "step": 8524 }, { "epoch": 4.51, "learning_rate": 4.935762327881955e-07, "loss": 0.0811, "step": 8525 }, { "epoch": 4.51, "learning_rate": 4.92512715088398e-07, "loss": 0.0439, "step": 8526 }, { "epoch": 4.51, "learning_rate": 4.914503154862316e-07, "loss": 0.0745, "step": 8527 }, { "epoch": 4.51, "learning_rate": 4.903890341066408e-07, "loss": 0.063, "step": 8528 }, { "epoch": 4.51, "learning_rate": 4.893288710744292e-07, "loss": 0.0585, "step": 8529 }, { "epoch": 4.52, "learning_rate": 4.882698265142793e-07, "loss": 0.0792, "step": 8530 }, { "epoch": 4.52, "learning_rate": 4.872119005507325e-07, "loss": 0.0869, "step": 8531 }, { "epoch": 4.52, "learning_rate": 4.861550933082038e-07, "loss": 0.073, "step": 8532 }, { "epoch": 4.52, "learning_rate": 4.850994049109769e-07, "loss": 0.0761, "step": 8533 }, { "epoch": 4.52, "learning_rate": 4.840448354832006e-07, "loss": 0.0637, "step": 8534 }, { "epoch": 4.52, "learning_rate": 4.829913851488954e-07, "loss": 0.0616, "step": 8535 }, { "epoch": 4.52, "learning_rate": 4.81939054031949e-07, "loss": 0.0677, "step": 8536 }, { "epoch": 4.52, "learning_rate": 4.80887842256117e-07, "loss": 0.0688, "step": 8537 }, { "epoch": 4.52, "learning_rate": 4.798377499450257e-07, "loss": 0.0371, "step": 8538 }, { "epoch": 4.52, "learning_rate": 4.787887772221655e-07, "loss": 0.0606, "step": 8539 }, { "epoch": 4.52, "learning_rate": 4.777409242108977e-07, "loss": 0.0663, "step": 8540 }, { "epoch": 4.52, "learning_rate": 4.7669419103445357e-07, "loss": 0.0864, "step": 8541 }, { "epoch": 4.52, "learning_rate": 4.75648577815927e-07, "loss": 0.0905, "step": 8542 }, { "epoch": 4.52, "learning_rate": 4.7460408467828844e-07, "loss": 0.0715, "step": 8543 }, { "epoch": 4.52, "learning_rate": 4.735607117443686e-07, "loss": 0.0925, "step": 8544 }, { "epoch": 4.52, "learning_rate": 4.7251845913687255e-07, "loss": 0.0653, "step": 8545 }, { "epoch": 4.52, "learning_rate": 4.7147732697837237e-07, "loss": 0.0661, "step": 8546 }, { "epoch": 4.52, "learning_rate": 4.7043731539130113e-07, "loss": 0.0942, "step": 8547 }, { "epoch": 4.52, "learning_rate": 4.693984244979721e-07, "loss": 0.2136, "step": 8548 }, { "epoch": 4.53, "learning_rate": 4.6836065442055655e-07, "loss": 0.107, "step": 8549 }, { "epoch": 4.53, "learning_rate": 4.673240052811001e-07, "loss": 0.0579, "step": 8550 }, { "epoch": 4.53, "learning_rate": 4.662884772015164e-07, "loss": 0.0707, "step": 8551 }, { "epoch": 4.53, "learning_rate": 4.652540703035813e-07, "loss": 0.0382, "step": 8552 }, { "epoch": 4.53, "learning_rate": 4.642207847089464e-07, "loss": 0.0645, "step": 8553 }, { "epoch": 4.53, "learning_rate": 4.631886205391267e-07, "loss": 0.0706, "step": 8554 }, { "epoch": 4.53, "learning_rate": 4.621575779155063e-07, "loss": 0.0636, "step": 8555 }, { "epoch": 4.53, "learning_rate": 4.6112765695933704e-07, "loss": 0.0492, "step": 8556 }, { "epoch": 4.53, "learning_rate": 4.600988577917409e-07, "loss": 0.047, "step": 8557 }, { "epoch": 4.53, "learning_rate": 4.590711805337056e-07, "loss": 0.0385, "step": 8558 }, { "epoch": 4.53, "learning_rate": 4.5804462530608773e-07, "loss": 0.0517, "step": 8559 }, { "epoch": 4.53, "learning_rate": 4.57019192229613e-07, "loss": 0.0661, "step": 8560 }, { "epoch": 4.53, "learning_rate": 4.559948814248738e-07, "loss": 0.1063, "step": 8561 }, { "epoch": 4.53, "learning_rate": 4.549716930123305e-07, "loss": 0.0762, "step": 8562 }, { "epoch": 4.53, "learning_rate": 4.5394962711231226e-07, "loss": 0.0664, "step": 8563 }, { "epoch": 4.53, "learning_rate": 4.529286838450153e-07, "loss": 0.0816, "step": 8564 }, { "epoch": 4.53, "learning_rate": 4.5190886333050467e-07, "loss": 0.056, "step": 8565 }, { "epoch": 4.53, "learning_rate": 4.5089016568871434e-07, "loss": 0.0912, "step": 8566 }, { "epoch": 4.53, "learning_rate": 4.498725910394419e-07, "loss": 0.0905, "step": 8567 }, { "epoch": 4.54, "learning_rate": 4.488561395023605e-07, "loss": 0.0542, "step": 8568 }, { "epoch": 4.54, "learning_rate": 4.478408111970023e-07, "loss": 0.0945, "step": 8569 }, { "epoch": 4.54, "learning_rate": 4.4682660624277394e-07, "loss": 0.052, "step": 8570 }, { "epoch": 4.54, "learning_rate": 4.45813524758949e-07, "loss": 0.0535, "step": 8571 }, { "epoch": 4.54, "learning_rate": 4.448015668646632e-07, "loss": 0.0924, "step": 8572 }, { "epoch": 4.54, "learning_rate": 4.4379073267893035e-07, "loss": 0.0627, "step": 8573 }, { "epoch": 4.54, "learning_rate": 4.427810223206208e-07, "loss": 0.0981, "step": 8574 }, { "epoch": 4.54, "learning_rate": 4.41772435908483e-07, "loss": 0.0741, "step": 8575 }, { "epoch": 4.54, "learning_rate": 4.4076497356112526e-07, "loss": 0.0702, "step": 8576 }, { "epoch": 4.54, "learning_rate": 4.3975863539702847e-07, "loss": 0.0829, "step": 8577 }, { "epoch": 4.54, "learning_rate": 4.3875342153454127e-07, "loss": 0.0692, "step": 8578 }, { "epoch": 4.54, "learning_rate": 4.377493320918758e-07, "loss": 0.0567, "step": 8579 }, { "epoch": 4.54, "learning_rate": 4.367463671871153e-07, "loss": 0.0764, "step": 8580 }, { "epoch": 4.54, "learning_rate": 4.3574452693821214e-07, "loss": 0.0627, "step": 8581 }, { "epoch": 4.54, "learning_rate": 4.3474381146298206e-07, "loss": 0.0639, "step": 8582 }, { "epoch": 4.54, "learning_rate": 4.337442208791132e-07, "loss": 0.0935, "step": 8583 }, { "epoch": 4.54, "learning_rate": 4.3274575530415695e-07, "loss": 0.0705, "step": 8584 }, { "epoch": 4.54, "learning_rate": 4.317484148555373e-07, "loss": 0.0459, "step": 8585 }, { "epoch": 4.54, "learning_rate": 4.3075219965054147e-07, "loss": 0.056, "step": 8586 }, { "epoch": 4.55, "learning_rate": 4.297571098063269e-07, "loss": 0.0698, "step": 8587 }, { "epoch": 4.55, "learning_rate": 4.287631454399177e-07, "loss": 0.0764, "step": 8588 }, { "epoch": 4.55, "learning_rate": 4.2777030666820483e-07, "loss": 0.0699, "step": 8589 }, { "epoch": 4.55, "learning_rate": 4.267785936079505e-07, "loss": 0.0751, "step": 8590 }, { "epoch": 4.55, "learning_rate": 4.257880063757791e-07, "loss": 0.1178, "step": 8591 }, { "epoch": 4.55, "learning_rate": 4.247985450881864e-07, "loss": 0.1029, "step": 8592 }, { "epoch": 4.55, "learning_rate": 4.23810209861536e-07, "loss": 0.0876, "step": 8593 }, { "epoch": 4.55, "learning_rate": 4.2282300081205705e-07, "loss": 0.0998, "step": 8594 }, { "epoch": 4.55, "learning_rate": 4.21836918055849e-07, "loss": 0.085, "step": 8595 }, { "epoch": 4.55, "learning_rate": 4.2085196170887243e-07, "loss": 0.0805, "step": 8596 }, { "epoch": 4.55, "learning_rate": 4.198681318869624e-07, "loss": 0.0559, "step": 8597 }, { "epoch": 4.55, "learning_rate": 4.1888542870582085e-07, "loss": 0.0458, "step": 8598 }, { "epoch": 4.55, "learning_rate": 4.1790385228101303e-07, "loss": 0.0445, "step": 8599 }, { "epoch": 4.55, "learning_rate": 4.169234027279767e-07, "loss": 0.0708, "step": 8600 }, { "epoch": 4.55, "learning_rate": 4.159440801620107e-07, "loss": 0.0986, "step": 8601 }, { "epoch": 4.55, "learning_rate": 4.1496588469828847e-07, "loss": 0.0793, "step": 8602 }, { "epoch": 4.55, "learning_rate": 4.1398881645184576e-07, "loss": 0.0408, "step": 8603 }, { "epoch": 4.55, "learning_rate": 4.130128755375884e-07, "loss": 0.0666, "step": 8604 }, { "epoch": 4.55, "learning_rate": 4.1203806207028906e-07, "loss": 0.0866, "step": 8605 }, { "epoch": 4.56, "learning_rate": 4.1106437616458493e-07, "loss": 0.053, "step": 8606 }, { "epoch": 4.56, "learning_rate": 4.1009181793498664e-07, "loss": 0.0836, "step": 8607 }, { "epoch": 4.56, "learning_rate": 4.0912038749586714e-07, "loss": 0.0718, "step": 8608 }, { "epoch": 4.56, "learning_rate": 4.0815008496146837e-07, "loss": 0.0487, "step": 8609 }, { "epoch": 4.56, "learning_rate": 4.0718091044590127e-07, "loss": 0.0674, "step": 8610 }, { "epoch": 4.56, "learning_rate": 4.062128640631413e-07, "loss": 0.0554, "step": 8611 }, { "epoch": 4.56, "learning_rate": 4.0524594592703083e-07, "loss": 0.0911, "step": 8612 }, { "epoch": 4.56, "learning_rate": 4.042801561512844e-07, "loss": 0.0518, "step": 8613 }, { "epoch": 4.56, "learning_rate": 4.0331549484947775e-07, "loss": 0.051, "step": 8614 }, { "epoch": 4.56, "learning_rate": 4.023519621350591e-07, "loss": 0.0372, "step": 8615 }, { "epoch": 4.56, "learning_rate": 4.0138955812133896e-07, "loss": 0.0723, "step": 8616 }, { "epoch": 4.56, "learning_rate": 4.004282829215012e-07, "loss": 0.0602, "step": 8617 }, { "epoch": 4.56, "learning_rate": 3.9946813664858973e-07, "loss": 0.0917, "step": 8618 }, { "epoch": 4.56, "learning_rate": 3.985091194155233e-07, "loss": 0.0789, "step": 8619 }, { "epoch": 4.56, "learning_rate": 3.9755123133508155e-07, "loss": 0.0697, "step": 8620 }, { "epoch": 4.56, "learning_rate": 3.965944725199133e-07, "loss": 0.0598, "step": 8621 }, { "epoch": 4.56, "learning_rate": 3.9563884308253754e-07, "loss": 0.0438, "step": 8622 }, { "epoch": 4.56, "learning_rate": 3.9468434313533534e-07, "loss": 0.0912, "step": 8623 }, { "epoch": 4.56, "learning_rate": 3.9373097279056025e-07, "loss": 0.0908, "step": 8624 }, { "epoch": 4.57, "learning_rate": 3.9277873216032937e-07, "loss": 0.0763, "step": 8625 }, { "epoch": 4.57, "learning_rate": 3.918276213566263e-07, "loss": 0.0622, "step": 8626 }, { "epoch": 4.57, "learning_rate": 3.9087764049130614e-07, "loss": 0.0726, "step": 8627 }, { "epoch": 4.57, "learning_rate": 3.8992878967608725e-07, "loss": 0.0558, "step": 8628 }, { "epoch": 4.57, "learning_rate": 3.8898106902255484e-07, "loss": 0.0392, "step": 8629 }, { "epoch": 4.57, "learning_rate": 3.8803447864216416e-07, "loss": 0.0776, "step": 8630 }, { "epoch": 4.57, "learning_rate": 3.8708901864623393e-07, "loss": 0.0539, "step": 8631 }, { "epoch": 4.57, "learning_rate": 3.8614468914595527e-07, "loss": 0.0789, "step": 8632 }, { "epoch": 4.57, "learning_rate": 3.852014902523804e-07, "loss": 0.074, "step": 8633 }, { "epoch": 4.57, "learning_rate": 3.8425942207643175e-07, "loss": 0.0869, "step": 8634 }, { "epoch": 4.57, "learning_rate": 3.833184847289007e-07, "loss": 0.0661, "step": 8635 }, { "epoch": 4.57, "learning_rate": 3.8237867832043863e-07, "loss": 0.0535, "step": 8636 }, { "epoch": 4.57, "learning_rate": 3.8144000296157167e-07, "loss": 0.0925, "step": 8637 }, { "epoch": 4.57, "learning_rate": 3.8050245876268823e-07, "loss": 0.0872, "step": 8638 }, { "epoch": 4.57, "learning_rate": 3.7956604583404555e-07, "loss": 0.0487, "step": 8639 }, { "epoch": 4.57, "learning_rate": 3.7863076428576897e-07, "loss": 0.109, "step": 8640 }, { "epoch": 4.57, "learning_rate": 3.77696614227846e-07, "loss": 0.0513, "step": 8641 }, { "epoch": 4.57, "learning_rate": 3.767635957701388e-07, "loss": 0.0649, "step": 8642 }, { "epoch": 4.57, "learning_rate": 3.7583170902236956e-07, "loss": 0.0479, "step": 8643 }, { "epoch": 4.58, "learning_rate": 3.749009540941284e-07, "loss": 0.0895, "step": 8644 }, { "epoch": 4.58, "learning_rate": 3.739713310948767e-07, "loss": 0.0852, "step": 8645 }, { "epoch": 4.58, "learning_rate": 3.7304284013393697e-07, "loss": 0.051, "step": 8646 }, { "epoch": 4.58, "learning_rate": 3.7211548132050413e-07, "loss": 0.0815, "step": 8647 }, { "epoch": 4.58, "learning_rate": 3.711892547636353e-07, "loss": 0.0442, "step": 8648 }, { "epoch": 4.58, "learning_rate": 3.7026416057225677e-07, "loss": 0.0681, "step": 8649 }, { "epoch": 4.58, "learning_rate": 3.693401988551626e-07, "loss": 0.0423, "step": 8650 }, { "epoch": 4.58, "learning_rate": 3.684173697210114e-07, "loss": 0.0672, "step": 8651 }, { "epoch": 4.58, "learning_rate": 3.674956732783297e-07, "loss": 0.0827, "step": 8652 }, { "epoch": 4.58, "learning_rate": 3.665751096355097e-07, "loss": 0.0756, "step": 8653 }, { "epoch": 4.58, "learning_rate": 3.6565567890081144e-07, "loss": 0.0741, "step": 8654 }, { "epoch": 4.58, "learning_rate": 3.647373811823629e-07, "loss": 0.0775, "step": 8655 }, { "epoch": 4.58, "learning_rate": 3.6382021658815657e-07, "loss": 0.0941, "step": 8656 }, { "epoch": 4.58, "learning_rate": 3.6290418522605397e-07, "loss": 0.0941, "step": 8657 }, { "epoch": 4.58, "learning_rate": 3.6198928720378e-07, "loss": 0.1171, "step": 8658 }, { "epoch": 4.58, "learning_rate": 3.6107552262893087e-07, "loss": 0.0963, "step": 8659 }, { "epoch": 4.58, "learning_rate": 3.60162891608965e-07, "loss": 0.0604, "step": 8660 }, { "epoch": 4.58, "learning_rate": 3.592513942512088e-07, "loss": 0.08, "step": 8661 }, { "epoch": 4.58, "learning_rate": 3.5834103066285876e-07, "loss": 0.0551, "step": 8662 }, { "epoch": 4.59, "learning_rate": 3.574318009509714e-07, "loss": 0.0876, "step": 8663 }, { "epoch": 4.59, "learning_rate": 3.5652370522247793e-07, "loss": 0.099, "step": 8664 }, { "epoch": 4.59, "learning_rate": 3.5561674358416955e-07, "loss": 0.0623, "step": 8665 }, { "epoch": 4.59, "learning_rate": 3.547109161427065e-07, "loss": 0.0686, "step": 8666 }, { "epoch": 4.59, "learning_rate": 3.53806223004618e-07, "loss": 0.0565, "step": 8667 }, { "epoch": 4.59, "learning_rate": 3.529026642762945e-07, "loss": 0.0891, "step": 8668 }, { "epoch": 4.59, "learning_rate": 3.520002400639977e-07, "loss": 0.0429, "step": 8669 }, { "epoch": 4.59, "learning_rate": 3.510989504738549e-07, "loss": 0.0505, "step": 8670 }, { "epoch": 4.59, "learning_rate": 3.50198795611858e-07, "loss": 0.0387, "step": 8671 }, { "epoch": 4.59, "learning_rate": 3.4929977558386787e-07, "loss": 0.101, "step": 8672 }, { "epoch": 4.59, "learning_rate": 3.4840189049560993e-07, "loss": 0.059, "step": 8673 }, { "epoch": 4.59, "learning_rate": 3.4750514045267857e-07, "loss": 0.0822, "step": 8674 }, { "epoch": 4.59, "learning_rate": 3.4660952556053287e-07, "loss": 0.0675, "step": 8675 }, { "epoch": 4.59, "learning_rate": 3.4571504592449625e-07, "loss": 0.0719, "step": 8676 }, { "epoch": 4.59, "learning_rate": 3.448217016497646e-07, "loss": 0.0807, "step": 8677 }, { "epoch": 4.59, "learning_rate": 3.43929492841395e-07, "loss": 0.0469, "step": 8678 }, { "epoch": 4.59, "learning_rate": 3.430384196043135e-07, "loss": 0.0733, "step": 8679 }, { "epoch": 4.59, "learning_rate": 3.4214848204331073e-07, "loss": 0.0747, "step": 8680 }, { "epoch": 4.59, "learning_rate": 3.412596802630452e-07, "loss": 0.063, "step": 8681 }, { "epoch": 4.6, "learning_rate": 3.4037201436804336e-07, "loss": 0.0697, "step": 8682 }, { "epoch": 4.6, "learning_rate": 3.3948548446269604e-07, "loss": 0.1007, "step": 8683 }, { "epoch": 4.6, "learning_rate": 3.386000906512577e-07, "loss": 0.0694, "step": 8684 }, { "epoch": 4.6, "learning_rate": 3.3771583303785603e-07, "loss": 0.0687, "step": 8685 }, { "epoch": 4.6, "learning_rate": 3.36832711726478e-07, "loss": 0.0542, "step": 8686 }, { "epoch": 4.6, "learning_rate": 3.3595072682098385e-07, "loss": 0.1156, "step": 8687 }, { "epoch": 4.6, "learning_rate": 3.350698784250928e-07, "loss": 0.0868, "step": 8688 }, { "epoch": 4.6, "learning_rate": 3.3419016664239743e-07, "loss": 0.0646, "step": 8689 }, { "epoch": 4.6, "learning_rate": 3.333115915763507e-07, "loss": 0.0781, "step": 8690 }, { "epoch": 4.6, "learning_rate": 3.3243415333027773e-07, "loss": 0.067, "step": 8691 }, { "epoch": 4.6, "learning_rate": 3.3155785200736477e-07, "loss": 0.0563, "step": 8692 }, { "epoch": 4.6, "learning_rate": 3.3068268771066615e-07, "loss": 0.0784, "step": 8693 }, { "epoch": 4.6, "learning_rate": 3.298086605431039e-07, "loss": 0.0678, "step": 8694 }, { "epoch": 4.6, "learning_rate": 3.2893577060746364e-07, "loss": 0.0649, "step": 8695 }, { "epoch": 4.6, "learning_rate": 3.2806401800639987e-07, "loss": 0.0711, "step": 8696 }, { "epoch": 4.6, "learning_rate": 3.2719340284243394e-07, "loss": 0.0367, "step": 8697 }, { "epoch": 4.6, "learning_rate": 3.2632392521794843e-07, "loss": 0.0793, "step": 8698 }, { "epoch": 4.6, "learning_rate": 3.2545558523519816e-07, "loss": 0.0424, "step": 8699 }, { "epoch": 4.61, "learning_rate": 3.2458838299630035e-07, "loss": 0.0997, "step": 8700 }, { "epoch": 4.61, "learning_rate": 3.23722318603239e-07, "loss": 0.0528, "step": 8701 }, { "epoch": 4.61, "learning_rate": 3.22857392157867e-07, "loss": 0.0551, "step": 8702 }, { "epoch": 4.61, "learning_rate": 3.219936037618976e-07, "loss": 0.0477, "step": 8703 }, { "epoch": 4.61, "learning_rate": 3.2113095351691714e-07, "loss": 0.0745, "step": 8704 }, { "epoch": 4.61, "learning_rate": 3.2026944152437346e-07, "loss": 0.085, "step": 8705 }, { "epoch": 4.61, "learning_rate": 3.1940906788558213e-07, "loss": 0.0862, "step": 8706 }, { "epoch": 4.61, "learning_rate": 3.185498327017245e-07, "loss": 0.0985, "step": 8707 }, { "epoch": 4.61, "learning_rate": 3.176917360738496e-07, "loss": 0.0909, "step": 8708 }, { "epoch": 4.61, "learning_rate": 3.168347781028691e-07, "loss": 0.0561, "step": 8709 }, { "epoch": 4.61, "learning_rate": 3.159789588895623e-07, "loss": 0.1185, "step": 8710 }, { "epoch": 4.61, "learning_rate": 3.151242785345776e-07, "loss": 0.0494, "step": 8711 }, { "epoch": 4.61, "learning_rate": 3.142707371384257e-07, "loss": 0.1072, "step": 8712 }, { "epoch": 4.61, "learning_rate": 3.1341833480148407e-07, "loss": 0.0874, "step": 8713 }, { "epoch": 4.61, "learning_rate": 3.125670716239981e-07, "loss": 0.0842, "step": 8714 }, { "epoch": 4.61, "learning_rate": 3.1171694770607664e-07, "loss": 0.0854, "step": 8715 }, { "epoch": 4.61, "learning_rate": 3.108679631476963e-07, "loss": 0.0692, "step": 8716 }, { "epoch": 4.61, "learning_rate": 3.100201180486995e-07, "loss": 0.0716, "step": 8717 }, { "epoch": 4.61, "learning_rate": 3.0917341250879305e-07, "loss": 0.0503, "step": 8718 }, { "epoch": 4.62, "learning_rate": 3.08327846627553e-07, "loss": 0.0595, "step": 8719 }, { "epoch": 4.62, "learning_rate": 3.074834205044175e-07, "loss": 0.0366, "step": 8720 }, { "epoch": 4.62, "learning_rate": 3.066401342386949e-07, "loss": 0.0501, "step": 8721 }, { "epoch": 4.62, "learning_rate": 3.057979879295536e-07, "loss": 0.0762, "step": 8722 }, { "epoch": 4.62, "learning_rate": 3.0495698167603447e-07, "loss": 0.0912, "step": 8723 }, { "epoch": 4.62, "learning_rate": 3.041171155770428e-07, "loss": 0.0812, "step": 8724 }, { "epoch": 4.62, "learning_rate": 3.0327838973134405e-07, "loss": 0.0773, "step": 8725 }, { "epoch": 4.62, "learning_rate": 3.024408042375759e-07, "loss": 0.0467, "step": 8726 }, { "epoch": 4.62, "learning_rate": 3.016043591942408e-07, "loss": 0.1183, "step": 8727 }, { "epoch": 4.62, "learning_rate": 3.007690546997044e-07, "loss": 0.0553, "step": 8728 }, { "epoch": 4.62, "learning_rate": 2.999348908522026e-07, "loss": 0.0859, "step": 8729 }, { "epoch": 4.62, "learning_rate": 2.991018677498314e-07, "loss": 0.0592, "step": 8730 }, { "epoch": 4.62, "learning_rate": 2.9826998549055795e-07, "loss": 0.1101, "step": 8731 }, { "epoch": 4.62, "learning_rate": 2.9743924417221403e-07, "loss": 0.0598, "step": 8732 }, { "epoch": 4.62, "learning_rate": 2.966096438924926e-07, "loss": 0.0782, "step": 8733 }, { "epoch": 4.62, "learning_rate": 2.9578118474896e-07, "loss": 0.0694, "step": 8734 }, { "epoch": 4.62, "learning_rate": 2.949538668390406e-07, "loss": 0.0575, "step": 8735 }, { "epoch": 4.62, "learning_rate": 2.9412769026003317e-07, "loss": 0.0867, "step": 8736 }, { "epoch": 4.62, "learning_rate": 2.933026551090934e-07, "loss": 0.0479, "step": 8737 }, { "epoch": 4.63, "learning_rate": 2.9247876148324917e-07, "loss": 0.0497, "step": 8738 }, { "epoch": 4.63, "learning_rate": 2.916560094793919e-07, "loss": 0.0845, "step": 8739 }, { "epoch": 4.63, "learning_rate": 2.908343991942786e-07, "loss": 0.0948, "step": 8740 }, { "epoch": 4.63, "learning_rate": 2.90013930724532e-07, "loss": 0.0819, "step": 8741 }, { "epoch": 4.63, "learning_rate": 2.8919460416663937e-07, "loss": 0.0718, "step": 8742 }, { "epoch": 4.63, "learning_rate": 2.8837641961695474e-07, "loss": 0.0642, "step": 8743 }, { "epoch": 4.63, "learning_rate": 2.8755937717170113e-07, "loss": 0.1072, "step": 8744 }, { "epoch": 4.63, "learning_rate": 2.8674347692696167e-07, "loss": 0.0335, "step": 8745 }, { "epoch": 4.63, "learning_rate": 2.859287189786886e-07, "loss": 0.049, "step": 8746 }, { "epoch": 4.63, "learning_rate": 2.851151034226984e-07, "loss": 0.0629, "step": 8747 }, { "epoch": 4.63, "learning_rate": 2.8430263035467587e-07, "loss": 0.071, "step": 8748 }, { "epoch": 4.63, "learning_rate": 2.8349129987016664e-07, "loss": 0.063, "step": 8749 }, { "epoch": 4.63, "learning_rate": 2.8268111206458446e-07, "loss": 0.0943, "step": 8750 }, { "epoch": 4.63, "learning_rate": 2.818720670332109e-07, "loss": 0.0688, "step": 8751 }, { "epoch": 4.63, "learning_rate": 2.810641648711887e-07, "loss": 0.0699, "step": 8752 }, { "epoch": 4.63, "learning_rate": 2.8025740567353075e-07, "loss": 0.0815, "step": 8753 }, { "epoch": 4.63, "learning_rate": 2.7945178953511344e-07, "loss": 0.0896, "step": 8754 }, { "epoch": 4.63, "learning_rate": 2.786473165506765e-07, "loss": 0.0556, "step": 8755 }, { "epoch": 4.63, "learning_rate": 2.7784398681483194e-07, "loss": 0.0521, "step": 8756 }, { "epoch": 4.64, "learning_rate": 2.770418004220465e-07, "loss": 0.0425, "step": 8757 }, { "epoch": 4.64, "learning_rate": 2.7624075746666257e-07, "loss": 0.0753, "step": 8758 }, { "epoch": 4.64, "learning_rate": 2.7544085804288467e-07, "loss": 0.0577, "step": 8759 }, { "epoch": 4.64, "learning_rate": 2.7464210224477985e-07, "loss": 0.104, "step": 8760 }, { "epoch": 4.64, "learning_rate": 2.7384449016628623e-07, "loss": 0.0609, "step": 8761 }, { "epoch": 4.64, "learning_rate": 2.7304802190120107e-07, "loss": 0.0564, "step": 8762 }, { "epoch": 4.64, "learning_rate": 2.722526975431938e-07, "loss": 0.0738, "step": 8763 }, { "epoch": 4.64, "learning_rate": 2.714585171857953e-07, "loss": 0.1226, "step": 8764 }, { "epoch": 4.64, "learning_rate": 2.7066548092239965e-07, "loss": 0.0754, "step": 8765 }, { "epoch": 4.64, "learning_rate": 2.6987358884627333e-07, "loss": 0.1004, "step": 8766 }, { "epoch": 4.64, "learning_rate": 2.690828410505408e-07, "loss": 0.0833, "step": 8767 }, { "epoch": 4.64, "learning_rate": 2.6829323762819883e-07, "loss": 0.1138, "step": 8768 }, { "epoch": 4.64, "learning_rate": 2.67504778672103e-07, "loss": 0.0838, "step": 8769 }, { "epoch": 4.64, "learning_rate": 2.6671746427497925e-07, "loss": 0.0606, "step": 8770 }, { "epoch": 4.64, "learning_rate": 2.659312945294168e-07, "loss": 0.076, "step": 8771 }, { "epoch": 4.64, "learning_rate": 2.651462695278717e-07, "loss": 0.1012, "step": 8772 }, { "epoch": 4.64, "learning_rate": 2.6436238936266235e-07, "loss": 0.0495, "step": 8773 }, { "epoch": 4.64, "learning_rate": 2.635796541259761e-07, "loss": 0.0824, "step": 8774 }, { "epoch": 4.64, "learning_rate": 2.6279806390986263e-07, "loss": 0.061, "step": 8775 }, { "epoch": 4.65, "learning_rate": 2.6201761880624064e-07, "loss": 0.1126, "step": 8776 }, { "epoch": 4.65, "learning_rate": 2.612383189068879e-07, "loss": 0.0872, "step": 8777 }, { "epoch": 4.65, "learning_rate": 2.6046016430345543e-07, "loss": 0.0698, "step": 8778 }, { "epoch": 4.65, "learning_rate": 2.5968315508745347e-07, "loss": 0.0632, "step": 8779 }, { "epoch": 4.65, "learning_rate": 2.5890729135026215e-07, "loss": 0.0834, "step": 8780 }, { "epoch": 4.65, "learning_rate": 2.581325731831208e-07, "loss": 0.0447, "step": 8781 }, { "epoch": 4.65, "learning_rate": 2.5735900067713984e-07, "loss": 0.0525, "step": 8782 }, { "epoch": 4.65, "learning_rate": 2.5658657392329313e-07, "loss": 0.0559, "step": 8783 }, { "epoch": 4.65, "learning_rate": 2.558152930124169e-07, "loss": 0.0383, "step": 8784 }, { "epoch": 4.65, "learning_rate": 2.550451580352176e-07, "loss": 0.0719, "step": 8785 }, { "epoch": 4.65, "learning_rate": 2.5427616908226596e-07, "loss": 0.0575, "step": 8786 }, { "epoch": 4.65, "learning_rate": 2.5350832624399193e-07, "loss": 0.0758, "step": 8787 }, { "epoch": 4.65, "learning_rate": 2.527416296106999e-07, "loss": 0.0715, "step": 8788 }, { "epoch": 4.65, "learning_rate": 2.519760792725512e-07, "loss": 0.0637, "step": 8789 }, { "epoch": 4.65, "learning_rate": 2.5121167531957703e-07, "loss": 0.0681, "step": 8790 }, { "epoch": 4.65, "learning_rate": 2.504484178416744e-07, "loss": 0.0716, "step": 8791 }, { "epoch": 4.65, "learning_rate": 2.4968630692860153e-07, "loss": 0.0386, "step": 8792 }, { "epoch": 4.65, "learning_rate": 2.4892534266998446e-07, "loss": 0.0821, "step": 8793 }, { "epoch": 4.65, "learning_rate": 2.48165525155315e-07, "loss": 0.0696, "step": 8794 }, { "epoch": 4.66, "learning_rate": 2.474068544739472e-07, "loss": 0.1138, "step": 8795 }, { "epoch": 4.66, "learning_rate": 2.4664933071510524e-07, "loss": 0.0597, "step": 8796 }, { "epoch": 4.66, "learning_rate": 2.4589295396787226e-07, "loss": 0.0652, "step": 8797 }, { "epoch": 4.66, "learning_rate": 2.4513772432120054e-07, "loss": 0.0603, "step": 8798 }, { "epoch": 4.66, "learning_rate": 2.4438364186390674e-07, "loss": 0.0769, "step": 8799 }, { "epoch": 4.66, "learning_rate": 2.43630706684671e-07, "loss": 0.0625, "step": 8800 }, { "epoch": 4.66, "learning_rate": 2.428789188720426e-07, "loss": 0.0722, "step": 8801 }, { "epoch": 4.66, "learning_rate": 2.4212827851442964e-07, "loss": 0.0666, "step": 8802 }, { "epoch": 4.66, "learning_rate": 2.4137878570011264e-07, "loss": 0.0785, "step": 8803 }, { "epoch": 4.66, "learning_rate": 2.4063044051722994e-07, "loss": 0.0705, "step": 8804 }, { "epoch": 4.66, "learning_rate": 2.3988324305379005e-07, "loss": 0.109, "step": 8805 }, { "epoch": 4.66, "learning_rate": 2.3913719339766497e-07, "loss": 0.066, "step": 8806 }, { "epoch": 4.66, "learning_rate": 2.3839229163658883e-07, "loss": 0.0786, "step": 8807 }, { "epoch": 4.66, "learning_rate": 2.3764853785816722e-07, "loss": 0.0884, "step": 8808 }, { "epoch": 4.66, "learning_rate": 2.3690593214986457e-07, "loss": 0.0542, "step": 8809 }, { "epoch": 4.66, "learning_rate": 2.3616447459901325e-07, "loss": 0.0592, "step": 8810 }, { "epoch": 4.66, "learning_rate": 2.354241652928091e-07, "loss": 0.061, "step": 8811 }, { "epoch": 4.66, "learning_rate": 2.3468500431831575e-07, "loss": 0.0395, "step": 8812 }, { "epoch": 4.66, "learning_rate": 2.3394699176245927e-07, "loss": 0.0748, "step": 8813 }, { "epoch": 4.67, "learning_rate": 2.3321012771202912e-07, "loss": 0.068, "step": 8814 }, { "epoch": 4.67, "learning_rate": 2.3247441225368373e-07, "loss": 0.0961, "step": 8815 }, { "epoch": 4.67, "learning_rate": 2.3173984547394612e-07, "loss": 0.0524, "step": 8816 }, { "epoch": 4.67, "learning_rate": 2.3100642745919833e-07, "loss": 0.0764, "step": 8817 }, { "epoch": 4.67, "learning_rate": 2.3027415829569688e-07, "loss": 0.0724, "step": 8818 }, { "epoch": 4.67, "learning_rate": 2.295430380695529e-07, "loss": 0.0635, "step": 8819 }, { "epoch": 4.67, "learning_rate": 2.2881306686675097e-07, "loss": 0.0787, "step": 8820 }, { "epoch": 4.67, "learning_rate": 2.2808424477313684e-07, "loss": 0.0543, "step": 8821 }, { "epoch": 4.67, "learning_rate": 2.2735657187441863e-07, "loss": 0.0806, "step": 8822 }, { "epoch": 4.67, "learning_rate": 2.2663004825617453e-07, "loss": 0.0453, "step": 8823 }, { "epoch": 4.67, "learning_rate": 2.2590467400384285e-07, "loss": 0.0757, "step": 8824 }, { "epoch": 4.67, "learning_rate": 2.2518044920273206e-07, "loss": 0.0711, "step": 8825 }, { "epoch": 4.67, "learning_rate": 2.244573739380096e-07, "loss": 0.0479, "step": 8826 }, { "epoch": 4.67, "learning_rate": 2.237354482947107e-07, "loss": 0.1057, "step": 8827 }, { "epoch": 4.67, "learning_rate": 2.2301467235773644e-07, "loss": 0.0854, "step": 8828 }, { "epoch": 4.67, "learning_rate": 2.2229504621185117e-07, "loss": 0.0616, "step": 8829 }, { "epoch": 4.67, "learning_rate": 2.2157656994168276e-07, "loss": 0.0637, "step": 8830 }, { "epoch": 4.67, "learning_rate": 2.2085924363172695e-07, "loss": 0.0467, "step": 8831 }, { "epoch": 4.67, "learning_rate": 2.2014306736634074e-07, "loss": 0.0648, "step": 8832 }, { "epoch": 4.68, "learning_rate": 2.1942804122975002e-07, "loss": 0.0882, "step": 8833 }, { "epoch": 4.68, "learning_rate": 2.1871416530604094e-07, "loss": 0.1164, "step": 8834 }, { "epoch": 4.68, "learning_rate": 2.1800143967916854e-07, "loss": 0.0512, "step": 8835 }, { "epoch": 4.68, "learning_rate": 2.1728986443294908e-07, "loss": 0.047, "step": 8836 }, { "epoch": 4.68, "learning_rate": 2.1657943965106677e-07, "loss": 0.0525, "step": 8837 }, { "epoch": 4.68, "learning_rate": 2.1587016541706807e-07, "loss": 0.0539, "step": 8838 }, { "epoch": 4.68, "learning_rate": 2.1516204181436296e-07, "loss": 0.0498, "step": 8839 }, { "epoch": 4.68, "learning_rate": 2.1445506892623146e-07, "loss": 0.0709, "step": 8840 }, { "epoch": 4.68, "learning_rate": 2.1374924683581266e-07, "loss": 0.0724, "step": 8841 }, { "epoch": 4.68, "learning_rate": 2.1304457562611236e-07, "loss": 0.0552, "step": 8842 }, { "epoch": 4.68, "learning_rate": 2.1234105538000206e-07, "loss": 0.0783, "step": 8843 }, { "epoch": 4.68, "learning_rate": 2.1163868618021667e-07, "loss": 0.0814, "step": 8844 }, { "epoch": 4.68, "learning_rate": 2.10937468109359e-07, "loss": 0.0569, "step": 8845 }, { "epoch": 4.68, "learning_rate": 2.1023740124988756e-07, "loss": 0.1282, "step": 8846 }, { "epoch": 4.68, "learning_rate": 2.0953848568413537e-07, "loss": 0.0344, "step": 8847 }, { "epoch": 4.68, "learning_rate": 2.0884072149429558e-07, "loss": 0.0372, "step": 8848 }, { "epoch": 4.68, "learning_rate": 2.0814410876242585e-07, "loss": 0.0848, "step": 8849 }, { "epoch": 4.68, "learning_rate": 2.0744864757045068e-07, "loss": 0.075, "step": 8850 }, { "epoch": 4.68, "learning_rate": 2.0675433800015576e-07, "loss": 0.0722, "step": 8851 }, { "epoch": 4.69, "learning_rate": 2.0606118013319466e-07, "loss": 0.0999, "step": 8852 }, { "epoch": 4.69, "learning_rate": 2.0536917405108325e-07, "loss": 0.0533, "step": 8853 }, { "epoch": 4.69, "learning_rate": 2.0467831983520204e-07, "loss": 0.0687, "step": 8854 }, { "epoch": 4.69, "learning_rate": 2.039886175667982e-07, "loss": 0.0657, "step": 8855 }, { "epoch": 4.69, "learning_rate": 2.0330006732697915e-07, "loss": 0.0692, "step": 8856 }, { "epoch": 4.69, "learning_rate": 2.026126691967234e-07, "loss": 0.1048, "step": 8857 }, { "epoch": 4.69, "learning_rate": 2.0192642325686852e-07, "loss": 0.0571, "step": 8858 }, { "epoch": 4.69, "learning_rate": 2.012413295881166e-07, "loss": 0.0775, "step": 8859 }, { "epoch": 4.69, "learning_rate": 2.0055738827103876e-07, "loss": 0.0531, "step": 8860 }, { "epoch": 4.69, "learning_rate": 1.9987459938606736e-07, "loss": 0.0641, "step": 8861 }, { "epoch": 4.69, "learning_rate": 1.99192963013497e-07, "loss": 0.0564, "step": 8862 }, { "epoch": 4.69, "learning_rate": 1.9851247923349137e-07, "loss": 0.0687, "step": 8863 }, { "epoch": 4.69, "learning_rate": 1.9783314812607646e-07, "loss": 0.0903, "step": 8864 }, { "epoch": 4.69, "learning_rate": 1.9715496977114278e-07, "loss": 0.0828, "step": 8865 }, { "epoch": 4.69, "learning_rate": 1.9647794424844435e-07, "loss": 0.0768, "step": 8866 }, { "epoch": 4.69, "learning_rate": 1.958020716376019e-07, "loss": 0.0583, "step": 8867 }, { "epoch": 4.69, "learning_rate": 1.951273520180985e-07, "loss": 0.0588, "step": 8868 }, { "epoch": 4.69, "learning_rate": 1.9445378546928406e-07, "loss": 0.0864, "step": 8869 }, { "epoch": 4.69, "learning_rate": 1.937813720703685e-07, "loss": 0.0692, "step": 8870 }, { "epoch": 4.7, "learning_rate": 1.9311011190043082e-07, "loss": 0.0639, "step": 8871 }, { "epoch": 4.7, "learning_rate": 1.924400050384101e-07, "loss": 0.0649, "step": 8872 }, { "epoch": 4.7, "learning_rate": 1.9177105156311547e-07, "loss": 0.0785, "step": 8873 }, { "epoch": 4.7, "learning_rate": 1.911032515532152e-07, "loss": 0.0394, "step": 8874 }, { "epoch": 4.7, "learning_rate": 1.9043660508724415e-07, "loss": 0.0942, "step": 8875 }, { "epoch": 4.7, "learning_rate": 1.8977111224359968e-07, "loss": 0.0589, "step": 8876 }, { "epoch": 4.7, "learning_rate": 1.8910677310054802e-07, "loss": 0.0496, "step": 8877 }, { "epoch": 4.7, "learning_rate": 1.8844358773621452e-07, "loss": 0.0886, "step": 8878 }, { "epoch": 4.7, "learning_rate": 1.8778155622859006e-07, "loss": 0.0525, "step": 8879 }, { "epoch": 4.7, "learning_rate": 1.871206786555324e-07, "loss": 0.1144, "step": 8880 }, { "epoch": 4.7, "learning_rate": 1.8646095509476159e-07, "loss": 0.0421, "step": 8881 }, { "epoch": 4.7, "learning_rate": 1.8580238562386333e-07, "loss": 0.0893, "step": 8882 }, { "epoch": 4.7, "learning_rate": 1.8514497032028344e-07, "loss": 0.0712, "step": 8883 }, { "epoch": 4.7, "learning_rate": 1.844887092613379e-07, "loss": 0.0702, "step": 8884 }, { "epoch": 4.7, "learning_rate": 1.838336025242049e-07, "loss": 0.0768, "step": 8885 }, { "epoch": 4.7, "learning_rate": 1.8317965018592397e-07, "loss": 0.1, "step": 8886 }, { "epoch": 4.7, "learning_rate": 1.8252685232340138e-07, "loss": 0.0589, "step": 8887 }, { "epoch": 4.7, "learning_rate": 1.8187520901340795e-07, "loss": 0.0861, "step": 8888 }, { "epoch": 4.71, "learning_rate": 1.8122472033257677e-07, "loss": 0.0766, "step": 8889 }, { "epoch": 4.71, "learning_rate": 1.8057538635741e-07, "loss": 0.0767, "step": 8890 }, { "epoch": 4.71, "learning_rate": 1.7992720716426659e-07, "loss": 0.0754, "step": 8891 }, { "epoch": 4.71, "learning_rate": 1.7928018282937554e-07, "loss": 0.0652, "step": 8892 }, { "epoch": 4.71, "learning_rate": 1.786343134288293e-07, "loss": 0.0695, "step": 8893 }, { "epoch": 4.71, "learning_rate": 1.7798959903857938e-07, "loss": 0.0965, "step": 8894 }, { "epoch": 4.71, "learning_rate": 1.773460397344484e-07, "loss": 0.0859, "step": 8895 }, { "epoch": 4.71, "learning_rate": 1.7670363559211922e-07, "loss": 0.0438, "step": 8896 }, { "epoch": 4.71, "learning_rate": 1.7606238668714026e-07, "loss": 0.0352, "step": 8897 }, { "epoch": 4.71, "learning_rate": 1.7542229309492232e-07, "loss": 0.0823, "step": 8898 }, { "epoch": 4.71, "learning_rate": 1.7478335489074294e-07, "loss": 0.0472, "step": 8899 }, { "epoch": 4.71, "learning_rate": 1.7414557214974204e-07, "loss": 0.0438, "step": 8900 }, { "epoch": 4.71, "learning_rate": 1.7350894494692405e-07, "loss": 0.1118, "step": 8901 }, { "epoch": 4.71, "learning_rate": 1.72873473357158e-07, "loss": 0.0592, "step": 8902 }, { "epoch": 4.71, "learning_rate": 1.7223915745517406e-07, "loss": 0.0638, "step": 8903 }, { "epoch": 4.71, "learning_rate": 1.7160599731557038e-07, "loss": 0.0697, "step": 8904 }, { "epoch": 4.71, "learning_rate": 1.7097399301280847e-07, "loss": 0.0711, "step": 8905 }, { "epoch": 4.71, "learning_rate": 1.7034314462121338e-07, "loss": 0.0897, "step": 8906 }, { "epoch": 4.71, "learning_rate": 1.697134522149724e-07, "loss": 0.1083, "step": 8907 }, { "epoch": 4.72, "learning_rate": 1.6908491586813957e-07, "loss": 0.0637, "step": 8908 }, { "epoch": 4.72, "learning_rate": 1.6845753565463253e-07, "loss": 0.1088, "step": 8909 }, { "epoch": 4.72, "learning_rate": 1.6783131164823106e-07, "loss": 0.0714, "step": 8910 }, { "epoch": 4.72, "learning_rate": 1.6720624392257966e-07, "loss": 0.0599, "step": 8911 }, { "epoch": 4.72, "learning_rate": 1.665823325511895e-07, "loss": 0.0553, "step": 8912 }, { "epoch": 4.72, "learning_rate": 1.6595957760743188e-07, "loss": 0.0757, "step": 8913 }, { "epoch": 4.72, "learning_rate": 1.6533797916454597e-07, "loss": 0.086, "step": 8914 }, { "epoch": 4.72, "learning_rate": 1.6471753729563e-07, "loss": 0.0598, "step": 8915 }, { "epoch": 4.72, "learning_rate": 1.6409825207365116e-07, "loss": 0.0588, "step": 8916 }, { "epoch": 4.72, "learning_rate": 1.6348012357143895e-07, "loss": 0.1207, "step": 8917 }, { "epoch": 4.72, "learning_rate": 1.6286315186168411e-07, "loss": 0.0735, "step": 8918 }, { "epoch": 4.72, "learning_rate": 1.6224733701694528e-07, "loss": 0.0626, "step": 8919 }, { "epoch": 4.72, "learning_rate": 1.6163267910964341e-07, "loss": 0.0682, "step": 8920 }, { "epoch": 4.72, "learning_rate": 1.6101917821206182e-07, "loss": 0.0856, "step": 8921 }, { "epoch": 4.72, "learning_rate": 1.6040683439635162e-07, "loss": 0.1031, "step": 8922 }, { "epoch": 4.72, "learning_rate": 1.5979564773452416e-07, "loss": 0.0683, "step": 8923 }, { "epoch": 4.72, "learning_rate": 1.5918561829845634e-07, "loss": 0.0694, "step": 8924 }, { "epoch": 4.72, "learning_rate": 1.585767461598886e-07, "loss": 0.0995, "step": 8925 }, { "epoch": 4.72, "learning_rate": 1.5796903139042473e-07, "loss": 0.0873, "step": 8926 }, { "epoch": 4.73, "learning_rate": 1.5736247406153538e-07, "loss": 0.0826, "step": 8927 }, { "epoch": 4.73, "learning_rate": 1.5675707424455012e-07, "loss": 0.0612, "step": 8928 }, { "epoch": 4.73, "learning_rate": 1.5615283201066646e-07, "loss": 0.0826, "step": 8929 }, { "epoch": 4.73, "learning_rate": 1.555497474309442e-07, "loss": 0.0538, "step": 8930 }, { "epoch": 4.73, "learning_rate": 1.5494782057630663e-07, "loss": 0.0839, "step": 8931 }, { "epoch": 4.73, "learning_rate": 1.543470515175427e-07, "loss": 0.0842, "step": 8932 }, { "epoch": 4.73, "learning_rate": 1.5374744032530365e-07, "loss": 0.0789, "step": 8933 }, { "epoch": 4.73, "learning_rate": 1.5314898707010306e-07, "loss": 0.0818, "step": 8934 }, { "epoch": 4.73, "learning_rate": 1.5255169182232242e-07, "loss": 0.0707, "step": 8935 }, { "epoch": 4.73, "learning_rate": 1.5195555465220225e-07, "loss": 0.078, "step": 8936 }, { "epoch": 4.73, "learning_rate": 1.5136057562985306e-07, "loss": 0.0621, "step": 8937 }, { "epoch": 4.73, "learning_rate": 1.5076675482524117e-07, "loss": 0.077, "step": 8938 }, { "epoch": 4.73, "learning_rate": 1.5017409230820402e-07, "loss": 0.0952, "step": 8939 }, { "epoch": 4.73, "learning_rate": 1.4958258814843806e-07, "loss": 0.0914, "step": 8940 }, { "epoch": 4.73, "learning_rate": 1.4899224241550658e-07, "loss": 0.0534, "step": 8941 }, { "epoch": 4.73, "learning_rate": 1.48403055178834e-07, "loss": 0.0897, "step": 8942 }, { "epoch": 4.73, "learning_rate": 1.478150265077105e-07, "loss": 0.0793, "step": 8943 }, { "epoch": 4.73, "learning_rate": 1.4722815647129073e-07, "loss": 0.0928, "step": 8944 }, { "epoch": 4.73, "learning_rate": 1.4664244513858838e-07, "loss": 0.0874, "step": 8945 }, { "epoch": 4.74, "learning_rate": 1.4605789257848613e-07, "loss": 0.1258, "step": 8946 }, { "epoch": 4.74, "learning_rate": 1.4547449885972897e-07, "loss": 0.0772, "step": 8947 }, { "epoch": 4.74, "learning_rate": 1.4489226405092426e-07, "loss": 0.053, "step": 8948 }, { "epoch": 4.74, "learning_rate": 1.443111882205439e-07, "loss": 0.0716, "step": 8949 }, { "epoch": 4.74, "learning_rate": 1.4373127143692433e-07, "loss": 0.0759, "step": 8950 }, { "epoch": 4.74, "learning_rate": 1.431525137682621e-07, "loss": 0.0882, "step": 8951 }, { "epoch": 4.74, "learning_rate": 1.425749152826239e-07, "loss": 0.0852, "step": 8952 }, { "epoch": 4.74, "learning_rate": 1.4199847604793315e-07, "loss": 0.0743, "step": 8953 }, { "epoch": 4.74, "learning_rate": 1.414231961319823e-07, "loss": 0.0611, "step": 8954 }, { "epoch": 4.74, "learning_rate": 1.408490756024239e-07, "loss": 0.1274, "step": 8955 }, { "epoch": 4.74, "learning_rate": 1.4027611452677725e-07, "loss": 0.1331, "step": 8956 }, { "epoch": 4.74, "learning_rate": 1.397043129724207e-07, "loss": 0.076, "step": 8957 }, { "epoch": 4.74, "learning_rate": 1.3913367100660268e-07, "loss": 0.086, "step": 8958 }, { "epoch": 4.74, "learning_rate": 1.3856418869642953e-07, "loss": 0.0891, "step": 8959 }, { "epoch": 4.74, "learning_rate": 1.3799586610887317e-07, "loss": 0.0705, "step": 8960 }, { "epoch": 4.74, "learning_rate": 1.3742870331077018e-07, "loss": 0.0971, "step": 8961 }, { "epoch": 4.74, "learning_rate": 1.3686270036882054e-07, "loss": 0.0576, "step": 8962 }, { "epoch": 4.74, "learning_rate": 1.3629785734958546e-07, "loss": 0.0514, "step": 8963 }, { "epoch": 4.74, "learning_rate": 1.3573417431949398e-07, "loss": 0.0925, "step": 8964 }, { "epoch": 4.75, "learning_rate": 1.3517165134483311e-07, "loss": 0.071, "step": 8965 }, { "epoch": 4.75, "learning_rate": 1.3461028849175993e-07, "loss": 0.0775, "step": 8966 }, { "epoch": 4.75, "learning_rate": 1.3405008582628943e-07, "loss": 0.0554, "step": 8967 }, { "epoch": 4.75, "learning_rate": 1.334910434143022e-07, "loss": 0.0854, "step": 8968 }, { "epoch": 4.75, "learning_rate": 1.3293316132154344e-07, "loss": 0.0511, "step": 8969 }, { "epoch": 4.75, "learning_rate": 1.323764396136218e-07, "loss": 0.0849, "step": 8970 }, { "epoch": 4.75, "learning_rate": 1.3182087835600822e-07, "loss": 0.0425, "step": 8971 }, { "epoch": 4.75, "learning_rate": 1.3126647761403598e-07, "loss": 0.0681, "step": 8972 }, { "epoch": 4.75, "learning_rate": 1.3071323745290632e-07, "loss": 0.0708, "step": 8973 }, { "epoch": 4.75, "learning_rate": 1.3016115793768158e-07, "loss": 0.0955, "step": 8974 }, { "epoch": 4.75, "learning_rate": 1.296102391332843e-07, "loss": 0.0762, "step": 8975 }, { "epoch": 4.75, "learning_rate": 1.2906048110450488e-07, "loss": 0.0607, "step": 8976 }, { "epoch": 4.75, "learning_rate": 1.285118839159971e-07, "loss": 0.0806, "step": 8977 }, { "epoch": 4.75, "learning_rate": 1.2796444763227388e-07, "loss": 0.0797, "step": 8978 }, { "epoch": 4.75, "learning_rate": 1.2741817231771925e-07, "loss": 0.0779, "step": 8979 }, { "epoch": 4.75, "learning_rate": 1.2687305803657178e-07, "loss": 0.0497, "step": 8980 }, { "epoch": 4.75, "learning_rate": 1.2632910485294137e-07, "loss": 0.0762, "step": 8981 }, { "epoch": 4.75, "learning_rate": 1.2578631283079568e-07, "loss": 0.079, "step": 8982 }, { "epoch": 4.75, "learning_rate": 1.252446820339681e-07, "loss": 0.0715, "step": 8983 }, { "epoch": 4.76, "learning_rate": 1.247042125261577e-07, "loss": 0.0811, "step": 8984 }, { "epoch": 4.76, "learning_rate": 1.2416490437092144e-07, "loss": 0.0521, "step": 8985 }, { "epoch": 4.76, "learning_rate": 1.2362675763168408e-07, "loss": 0.0767, "step": 8986 }, { "epoch": 4.76, "learning_rate": 1.2308977237173393e-07, "loss": 0.0844, "step": 8987 }, { "epoch": 4.76, "learning_rate": 1.2255394865421931e-07, "loss": 0.0724, "step": 8988 }, { "epoch": 4.76, "learning_rate": 1.2201928654215655e-07, "loss": 0.05, "step": 8989 }, { "epoch": 4.76, "learning_rate": 1.21485786098422e-07, "loss": 0.109, "step": 8990 }, { "epoch": 4.76, "learning_rate": 1.2095344738575542e-07, "loss": 0.071, "step": 8991 }, { "epoch": 4.76, "learning_rate": 1.2042227046676014e-07, "loss": 0.0617, "step": 8992 }, { "epoch": 4.76, "learning_rate": 1.1989225540390504e-07, "loss": 0.0733, "step": 8993 }, { "epoch": 4.76, "learning_rate": 1.1936340225952026e-07, "loss": 0.1103, "step": 8994 }, { "epoch": 4.76, "learning_rate": 1.188357110958005e-07, "loss": 0.0456, "step": 8995 }, { "epoch": 4.76, "learning_rate": 1.1830918197480278e-07, "loss": 0.092, "step": 8996 }, { "epoch": 4.76, "learning_rate": 1.1778381495844759e-07, "loss": 0.0648, "step": 8997 }, { "epoch": 4.76, "learning_rate": 1.1725961010851882e-07, "loss": 0.0613, "step": 8998 }, { "epoch": 4.76, "learning_rate": 1.1673656748666606e-07, "loss": 0.081, "step": 8999 }, { "epoch": 4.76, "learning_rate": 1.1621468715439677e-07, "loss": 0.0768, "step": 9000 }, { "epoch": 4.76, "learning_rate": 1.1569396917308628e-07, "loss": 0.0715, "step": 9001 }, { "epoch": 4.76, "learning_rate": 1.1517441360397341e-07, "loss": 0.0552, "step": 9002 }, { "epoch": 4.77, "learning_rate": 1.1465602050815595e-07, "loss": 0.0665, "step": 9003 }, { "epoch": 4.77, "learning_rate": 1.1413878994660177e-07, "loss": 0.0725, "step": 9004 }, { "epoch": 4.77, "learning_rate": 1.1362272198013447e-07, "loss": 0.0783, "step": 9005 }, { "epoch": 4.77, "learning_rate": 1.131078166694477e-07, "loss": 0.1038, "step": 9006 }, { "epoch": 4.77, "learning_rate": 1.1259407407509193e-07, "loss": 0.1086, "step": 9007 }, { "epoch": 4.77, "learning_rate": 1.120814942574866e-07, "loss": 0.086, "step": 9008 }, { "epoch": 4.77, "learning_rate": 1.1157007727691127e-07, "loss": 0.0843, "step": 9009 }, { "epoch": 4.77, "learning_rate": 1.1105982319350895e-07, "loss": 0.093, "step": 9010 }, { "epoch": 4.77, "learning_rate": 1.1055073206728829e-07, "loss": 0.0384, "step": 9011 }, { "epoch": 4.77, "learning_rate": 1.1004280395811695e-07, "loss": 0.0495, "step": 9012 }, { "epoch": 4.77, "learning_rate": 1.095360389257305e-07, "loss": 0.0532, "step": 9013 }, { "epoch": 4.77, "learning_rate": 1.0903043702972349e-07, "loss": 0.0827, "step": 9014 }, { "epoch": 4.77, "learning_rate": 1.0852599832955612e-07, "loss": 0.0789, "step": 9015 }, { "epoch": 4.77, "learning_rate": 1.0802272288455207e-07, "loss": 0.0692, "step": 9016 }, { "epoch": 4.77, "learning_rate": 1.075206107538962e-07, "loss": 0.0665, "step": 9017 }, { "epoch": 4.77, "learning_rate": 1.0701966199663905e-07, "loss": 0.0815, "step": 9018 }, { "epoch": 4.77, "learning_rate": 1.0651987667169128e-07, "loss": 0.0575, "step": 9019 }, { "epoch": 4.77, "learning_rate": 1.0602125483783143e-07, "loss": 0.0668, "step": 9020 }, { "epoch": 4.77, "learning_rate": 1.0552379655369594e-07, "loss": 0.0933, "step": 9021 }, { "epoch": 4.78, "learning_rate": 1.0502750187778798e-07, "loss": 0.0943, "step": 9022 }, { "epoch": 4.78, "learning_rate": 1.04532370868472e-07, "loss": 0.0594, "step": 9023 }, { "epoch": 4.78, "learning_rate": 1.0403840358397699e-07, "loss": 0.0692, "step": 9024 }, { "epoch": 4.78, "learning_rate": 1.0354560008239312e-07, "loss": 0.0756, "step": 9025 }, { "epoch": 4.78, "learning_rate": 1.0305396042167626e-07, "loss": 0.0786, "step": 9026 }, { "epoch": 4.78, "learning_rate": 1.025634846596435e-07, "loss": 0.0715, "step": 9027 }, { "epoch": 4.78, "learning_rate": 1.0207417285397647e-07, "loss": 0.0542, "step": 9028 }, { "epoch": 4.78, "learning_rate": 1.0158602506221804e-07, "loss": 0.1134, "step": 9029 }, { "epoch": 4.78, "learning_rate": 1.0109904134177561e-07, "loss": 0.1114, "step": 9030 }, { "epoch": 4.78, "learning_rate": 1.0061322174992005e-07, "loss": 0.0492, "step": 9031 }, { "epoch": 4.78, "learning_rate": 1.0012856634378343e-07, "loss": 0.0877, "step": 9032 }, { "epoch": 4.78, "learning_rate": 9.964507518036349e-08, "loss": 0.0918, "step": 9033 }, { "epoch": 4.78, "learning_rate": 9.916274831651807e-08, "loss": 0.0499, "step": 9034 }, { "epoch": 4.78, "learning_rate": 9.868158580897069e-08, "loss": 0.0714, "step": 9035 }, { "epoch": 4.78, "learning_rate": 9.820158771430722e-08, "loss": 0.0765, "step": 9036 }, { "epoch": 4.78, "learning_rate": 9.772275408897469e-08, "loss": 0.0729, "step": 9037 }, { "epoch": 4.78, "learning_rate": 9.724508498928698e-08, "loss": 0.063, "step": 9038 }, { "epoch": 4.78, "learning_rate": 9.676858047141802e-08, "loss": 0.0648, "step": 9039 }, { "epoch": 4.78, "learning_rate": 9.62932405914041e-08, "loss": 0.05, "step": 9040 }, { "epoch": 4.79, "learning_rate": 9.581906540514829e-08, "loss": 0.0659, "step": 9041 }, { "epoch": 4.79, "learning_rate": 9.534605496841376e-08, "loss": 0.0714, "step": 9042 }, { "epoch": 4.79, "learning_rate": 9.487420933682601e-08, "loss": 0.0399, "step": 9043 }, { "epoch": 4.79, "learning_rate": 9.44035285658762e-08, "loss": 0.0955, "step": 9044 }, { "epoch": 4.79, "learning_rate": 9.393401271091784e-08, "loss": 0.0867, "step": 9045 }, { "epoch": 4.79, "learning_rate": 9.346566182716566e-08, "loss": 0.0539, "step": 9046 }, { "epoch": 4.79, "learning_rate": 9.299847596969891e-08, "loss": 0.0841, "step": 9047 }, { "epoch": 4.79, "learning_rate": 9.253245519346032e-08, "loss": 0.09, "step": 9048 }, { "epoch": 4.79, "learning_rate": 9.206759955325273e-08, "loss": 0.08, "step": 9049 }, { "epoch": 4.79, "learning_rate": 9.160390910374461e-08, "loss": 0.1025, "step": 9050 }, { "epoch": 4.79, "learning_rate": 9.114138389946903e-08, "loss": 0.0661, "step": 9051 }, { "epoch": 4.79, "learning_rate": 9.068002399481801e-08, "loss": 0.0813, "step": 9052 }, { "epoch": 4.79, "learning_rate": 9.021982944404817e-08, "loss": 0.0575, "step": 9053 }, { "epoch": 4.79, "learning_rate": 8.976080030127954e-08, "loss": 0.0838, "step": 9054 }, { "epoch": 4.79, "learning_rate": 8.93029366204945e-08, "loss": 0.0675, "step": 9055 }, { "epoch": 4.79, "learning_rate": 8.884623845553997e-08, "loss": 0.0679, "step": 9056 }, { "epoch": 4.79, "learning_rate": 8.839070586012189e-08, "loss": 0.1027, "step": 9057 }, { "epoch": 4.79, "learning_rate": 8.793633888781406e-08, "loss": 0.081, "step": 9058 }, { "epoch": 4.8, "learning_rate": 8.74831375920504e-08, "loss": 0.0553, "step": 9059 }, { "epoch": 4.8, "learning_rate": 8.703110202612719e-08, "loss": 0.0732, "step": 9060 }, { "epoch": 4.8, "learning_rate": 8.658023224320521e-08, "loss": 0.0732, "step": 9061 }, { "epoch": 4.8, "learning_rate": 8.613052829630875e-08, "loss": 0.047, "step": 9062 }, { "epoch": 4.8, "learning_rate": 8.568199023832213e-08, "loss": 0.0796, "step": 9063 }, { "epoch": 4.8, "learning_rate": 8.523461812199429e-08, "loss": 0.0889, "step": 9064 }, { "epoch": 4.8, "learning_rate": 8.478841199993759e-08, "loss": 0.0494, "step": 9065 }, { "epoch": 4.8, "learning_rate": 8.434337192462671e-08, "loss": 0.0586, "step": 9066 }, { "epoch": 4.8, "learning_rate": 8.38994979483987e-08, "loss": 0.0616, "step": 9067 }, { "epoch": 4.8, "learning_rate": 8.345679012345509e-08, "loss": 0.0724, "step": 9068 }, { "epoch": 4.8, "learning_rate": 8.30152485018576e-08, "loss": 0.0496, "step": 9069 }, { "epoch": 4.8, "learning_rate": 8.257487313553358e-08, "loss": 0.0742, "step": 9070 }, { "epoch": 4.8, "learning_rate": 8.21356640762716e-08, "loss": 0.0567, "step": 9071 }, { "epoch": 4.8, "learning_rate": 8.169762137572257e-08, "loss": 0.099, "step": 9072 }, { "epoch": 4.8, "learning_rate": 8.126074508540194e-08, "loss": 0.0445, "step": 9073 }, { "epoch": 4.8, "learning_rate": 8.082503525668639e-08, "loss": 0.092, "step": 9074 }, { "epoch": 4.8, "learning_rate": 8.039049194081828e-08, "loss": 0.0504, "step": 9075 }, { "epoch": 4.8, "learning_rate": 7.995711518889781e-08, "loss": 0.0573, "step": 9076 }, { "epoch": 4.8, "learning_rate": 7.952490505189204e-08, "loss": 0.0597, "step": 9077 }, { "epoch": 4.81, "learning_rate": 7.909386158063138e-08, "loss": 0.1089, "step": 9078 }, { "epoch": 4.81, "learning_rate": 7.866398482580529e-08, "loss": 0.0878, "step": 9079 }, { "epoch": 4.81, "learning_rate": 7.823527483796778e-08, "loss": 0.1155, "step": 9080 }, { "epoch": 4.81, "learning_rate": 7.780773166753741e-08, "loss": 0.1109, "step": 9081 }, { "epoch": 4.81, "learning_rate": 7.738135536479286e-08, "loss": 0.0794, "step": 9082 }, { "epoch": 4.81, "learning_rate": 7.695614597987733e-08, "loss": 0.0675, "step": 9083 }, { "epoch": 4.81, "learning_rate": 7.65321035627964e-08, "loss": 0.0752, "step": 9084 }, { "epoch": 4.81, "learning_rate": 7.610922816341792e-08, "loss": 0.09, "step": 9085 }, { "epoch": 4.81, "learning_rate": 7.568751983147216e-08, "loss": 0.1153, "step": 9086 }, { "epoch": 4.81, "learning_rate": 7.526697861655385e-08, "loss": 0.0614, "step": 9087 }, { "epoch": 4.81, "learning_rate": 7.484760456811902e-08, "loss": 0.1422, "step": 9088 }, { "epoch": 4.81, "learning_rate": 7.442939773548708e-08, "loss": 0.0687, "step": 9089 }, { "epoch": 4.81, "learning_rate": 7.40123581678387e-08, "loss": 0.1118, "step": 9090 }, { "epoch": 4.81, "learning_rate": 7.359648591421909e-08, "loss": 0.0788, "step": 9091 }, { "epoch": 4.81, "learning_rate": 7.31817810235369e-08, "loss": 0.0742, "step": 9092 }, { "epoch": 4.81, "learning_rate": 7.276824354456091e-08, "loss": 0.0811, "step": 9093 }, { "epoch": 4.81, "learning_rate": 7.235587352592333e-08, "loss": 0.0799, "step": 9094 }, { "epoch": 4.81, "learning_rate": 7.194467101612201e-08, "loss": 0.0787, "step": 9095 }, { "epoch": 4.81, "learning_rate": 7.153463606351163e-08, "loss": 0.064, "step": 9096 }, { "epoch": 4.82, "learning_rate": 7.112576871631582e-08, "loss": 0.0706, "step": 9097 }, { "epoch": 4.82, "learning_rate": 7.071806902261725e-08, "loss": 0.0781, "step": 9098 }, { "epoch": 4.82, "learning_rate": 7.031153703036087e-08, "loss": 0.0779, "step": 9099 }, { "epoch": 4.82, "learning_rate": 6.990617278735845e-08, "loss": 0.0694, "step": 9100 }, { "epoch": 4.82, "learning_rate": 6.950197634127854e-08, "loss": 0.0661, "step": 9101 }, { "epoch": 4.82, "learning_rate": 6.909894773965753e-08, "loss": 0.0998, "step": 9102 }, { "epoch": 4.82, "learning_rate": 6.869708702989197e-08, "loss": 0.0421, "step": 9103 }, { "epoch": 4.82, "learning_rate": 6.829639425923961e-08, "loss": 0.0615, "step": 9104 }, { "epoch": 4.82, "learning_rate": 6.789686947482388e-08, "loss": 0.0482, "step": 9105 }, { "epoch": 4.82, "learning_rate": 6.749851272363051e-08, "loss": 0.079, "step": 9106 }, { "epoch": 4.82, "learning_rate": 6.710132405250536e-08, "loss": 0.0563, "step": 9107 }, { "epoch": 4.82, "learning_rate": 6.670530350815996e-08, "loss": 0.0664, "step": 9108 }, { "epoch": 4.82, "learning_rate": 6.631045113716595e-08, "loss": 0.066, "step": 9109 }, { "epoch": 4.82, "learning_rate": 6.59167669859595e-08, "loss": 0.0845, "step": 9110 }, { "epoch": 4.82, "learning_rate": 6.552425110083915e-08, "loss": 0.0394, "step": 9111 }, { "epoch": 4.82, "learning_rate": 6.513290352796242e-08, "loss": 0.0654, "step": 9112 }, { "epoch": 4.82, "learning_rate": 6.474272431335582e-08, "loss": 0.088, "step": 9113 }, { "epoch": 4.82, "learning_rate": 6.435371350290376e-08, "loss": 0.095, "step": 9114 }, { "epoch": 4.82, "learning_rate": 6.396587114235519e-08, "loss": 0.0905, "step": 9115 }, { "epoch": 4.83, "learning_rate": 6.35791972773192e-08, "loss": 0.0706, "step": 9116 }, { "epoch": 4.83, "learning_rate": 6.319369195327274e-08, "loss": 0.0427, "step": 9117 }, { "epoch": 4.83, "learning_rate": 6.280935521554843e-08, "loss": 0.0579, "step": 9118 }, { "epoch": 4.83, "learning_rate": 6.242618710934677e-08, "loss": 0.0592, "step": 9119 }, { "epoch": 4.83, "learning_rate": 6.204418767972953e-08, "loss": 0.0571, "step": 9120 }, { "epoch": 4.83, "learning_rate": 6.166335697161852e-08, "loss": 0.0621, "step": 9121 }, { "epoch": 4.83, "learning_rate": 6.128369502980125e-08, "loss": 0.09, "step": 9122 }, { "epoch": 4.83, "learning_rate": 6.090520189892756e-08, "loss": 0.0613, "step": 9123 }, { "epoch": 4.83, "learning_rate": 6.052787762350854e-08, "loss": 0.0538, "step": 9124 }, { "epoch": 4.83, "learning_rate": 6.015172224791755e-08, "loss": 0.0816, "step": 9125 }, { "epoch": 4.83, "learning_rate": 5.977673581639032e-08, "loss": 0.074, "step": 9126 }, { "epoch": 4.83, "learning_rate": 5.9402918373028254e-08, "loss": 0.0654, "step": 9127 }, { "epoch": 4.83, "learning_rate": 5.903026996179062e-08, "loss": 0.0835, "step": 9128 }, { "epoch": 4.83, "learning_rate": 5.865879062650237e-08, "loss": 0.0802, "step": 9129 }, { "epoch": 4.83, "learning_rate": 5.828848041085189e-08, "loss": 0.0621, "step": 9130 }, { "epoch": 4.83, "learning_rate": 5.791933935838545e-08, "loss": 0.0661, "step": 9131 }, { "epoch": 4.83, "learning_rate": 5.755136751251611e-08, "loss": 0.0722, "step": 9132 }, { "epoch": 4.83, "learning_rate": 5.7184564916518136e-08, "loss": 0.069, "step": 9133 }, { "epoch": 4.83, "learning_rate": 5.681893161352814e-08, "loss": 0.1032, "step": 9134 }, { "epoch": 4.84, "learning_rate": 5.645446764654616e-08, "loss": 0.0464, "step": 9135 }, { "epoch": 4.84, "learning_rate": 5.609117305843126e-08, "loss": 0.0849, "step": 9136 }, { "epoch": 4.84, "learning_rate": 5.572904789191036e-08, "loss": 0.0747, "step": 9137 }, { "epoch": 4.84, "learning_rate": 5.536809218956718e-08, "loss": 0.0743, "step": 9138 }, { "epoch": 4.84, "learning_rate": 5.500830599385332e-08, "loss": 0.0608, "step": 9139 }, { "epoch": 4.84, "learning_rate": 5.464968934707826e-08, "loss": 0.0731, "step": 9140 }, { "epoch": 4.84, "learning_rate": 5.4292242291418276e-08, "loss": 0.0638, "step": 9141 }, { "epoch": 4.84, "learning_rate": 5.393596486890751e-08, "loss": 0.087, "step": 9142 }, { "epoch": 4.84, "learning_rate": 5.3580857121446895e-08, "loss": 0.0595, "step": 9143 }, { "epoch": 4.84, "learning_rate": 5.322691909079525e-08, "loss": 0.0721, "step": 9144 }, { "epoch": 4.84, "learning_rate": 5.287415081857927e-08, "loss": 0.0563, "step": 9145 }, { "epoch": 4.84, "learning_rate": 5.252255234628245e-08, "loss": 0.0694, "step": 9146 }, { "epoch": 4.84, "learning_rate": 5.2172123715255044e-08, "loss": 0.0852, "step": 9147 }, { "epoch": 4.84, "learning_rate": 5.182286496670741e-08, "loss": 0.0837, "step": 9148 }, { "epoch": 4.84, "learning_rate": 5.1474776141712256e-08, "loss": 0.0502, "step": 9149 }, { "epoch": 4.84, "learning_rate": 5.112785728120795e-08, "loss": 0.0796, "step": 9150 }, { "epoch": 4.84, "learning_rate": 5.078210842599074e-08, "loss": 0.0622, "step": 9151 }, { "epoch": 4.84, "learning_rate": 5.0437529616721437e-08, "loss": 0.0833, "step": 9152 }, { "epoch": 4.84, "learning_rate": 5.009412089392318e-08, "loss": 0.0451, "step": 9153 }, { "epoch": 4.85, "learning_rate": 4.9751882297982554e-08, "loss": 0.0885, "step": 9154 }, { "epoch": 4.85, "learning_rate": 4.941081386914626e-08, "loss": 0.0769, "step": 9155 }, { "epoch": 4.85, "learning_rate": 4.907091564752442e-08, "loss": 0.0789, "step": 9156 }, { "epoch": 4.85, "learning_rate": 4.873218767309173e-08, "loss": 0.0402, "step": 9157 }, { "epoch": 4.85, "learning_rate": 4.8394629985679676e-08, "loss": 0.0916, "step": 9158 }, { "epoch": 4.85, "learning_rate": 4.80582426249887e-08, "loss": 0.0816, "step": 9159 }, { "epoch": 4.85, "learning_rate": 4.772302563057829e-08, "loss": 0.0531, "step": 9160 }, { "epoch": 4.85, "learning_rate": 4.738897904186801e-08, "loss": 0.0445, "step": 9161 }, { "epoch": 4.85, "learning_rate": 4.705610289814533e-08, "loss": 0.0804, "step": 9162 }, { "epoch": 4.85, "learning_rate": 4.67243972385556e-08, "loss": 0.0781, "step": 9163 }, { "epoch": 4.85, "learning_rate": 4.6393862102108724e-08, "loss": 0.1397, "step": 9164 }, { "epoch": 4.85, "learning_rate": 4.606449752767472e-08, "loss": 0.0718, "step": 9165 }, { "epoch": 4.85, "learning_rate": 4.573630355398928e-08, "loss": 0.077, "step": 9166 }, { "epoch": 4.85, "learning_rate": 4.5409280219648186e-08, "loss": 0.0549, "step": 9167 }, { "epoch": 4.85, "learning_rate": 4.5083427563108464e-08, "loss": 0.075, "step": 9168 }, { "epoch": 4.85, "learning_rate": 4.4758745622692776e-08, "loss": 0.1005, "step": 9169 }, { "epoch": 4.85, "learning_rate": 4.443523443658393e-08, "loss": 0.0927, "step": 9170 }, { "epoch": 4.85, "learning_rate": 4.4112894042827037e-08, "loss": 0.0891, "step": 9171 }, { "epoch": 4.85, "learning_rate": 4.3791724479329556e-08, "loss": 0.0684, "step": 9172 }, { "epoch": 4.86, "learning_rate": 4.3471725783862384e-08, "loss": 0.0693, "step": 9173 }, { "epoch": 4.86, "learning_rate": 4.3152897994057644e-08, "loss": 0.0653, "step": 9174 }, { "epoch": 4.86, "learning_rate": 4.283524114740978e-08, "loss": 0.0481, "step": 9175 }, { "epoch": 4.86, "learning_rate": 4.251875528127669e-08, "loss": 0.0633, "step": 9176 }, { "epoch": 4.86, "learning_rate": 4.2203440432876385e-08, "loss": 0.1, "step": 9177 }, { "epoch": 4.86, "learning_rate": 4.188929663929142e-08, "loss": 0.0636, "step": 9178 }, { "epoch": 4.86, "learning_rate": 4.157632393746558e-08, "loss": 0.0791, "step": 9179 }, { "epoch": 4.86, "learning_rate": 4.1264522364204976e-08, "loss": 0.0623, "step": 9180 }, { "epoch": 4.86, "learning_rate": 4.095389195617694e-08, "loss": 0.0777, "step": 9181 }, { "epoch": 4.86, "learning_rate": 4.0644432749914474e-08, "loss": 0.0603, "step": 9182 }, { "epoch": 4.86, "learning_rate": 4.033614478180847e-08, "loss": 0.066, "step": 9183 }, { "epoch": 4.86, "learning_rate": 4.002902808811437e-08, "loss": 0.0473, "step": 9184 }, { "epoch": 4.86, "learning_rate": 3.972308270495107e-08, "loss": 0.0823, "step": 9185 }, { "epoch": 4.86, "learning_rate": 3.941830866829754e-08, "loss": 0.0809, "step": 9186 }, { "epoch": 4.86, "learning_rate": 3.9114706013995144e-08, "loss": 0.053, "step": 9187 }, { "epoch": 4.86, "learning_rate": 3.8812274777747517e-08, "loss": 0.0346, "step": 9188 }, { "epoch": 4.86, "learning_rate": 3.8511014995123997e-08, "loss": 0.0692, "step": 9189 }, { "epoch": 4.86, "learning_rate": 3.821092670154958e-08, "loss": 0.057, "step": 9190 }, { "epoch": 4.86, "learning_rate": 3.791200993231825e-08, "loss": 0.0958, "step": 9191 }, { "epoch": 4.87, "learning_rate": 3.76142647225819e-08, "loss": 0.0697, "step": 9192 }, { "epoch": 4.87, "learning_rate": 3.7317691107355835e-08, "loss": 0.0753, "step": 9193 }, { "epoch": 4.87, "learning_rate": 3.702228912151773e-08, "loss": 0.0723, "step": 9194 }, { "epoch": 4.87, "learning_rate": 3.6728058799807567e-08, "loss": 0.0868, "step": 9195 }, { "epoch": 4.87, "learning_rate": 3.6435000176826553e-08, "loss": 0.0866, "step": 9196 }, { "epoch": 4.87, "learning_rate": 3.6143113287040455e-08, "loss": 0.0606, "step": 9197 }, { "epoch": 4.87, "learning_rate": 3.585239816477404e-08, "loss": 0.0691, "step": 9198 }, { "epoch": 4.87, "learning_rate": 3.556285484421662e-08, "loss": 0.0641, "step": 9199 }, { "epoch": 4.87, "learning_rate": 3.5274483359419857e-08, "loss": 0.0686, "step": 9200 }, { "epoch": 4.87, "learning_rate": 3.4987283744295496e-08, "loss": 0.0633, "step": 9201 }, { "epoch": 4.87, "learning_rate": 3.470125603261987e-08, "loss": 0.054, "step": 9202 }, { "epoch": 4.87, "learning_rate": 3.441640025802828e-08, "loss": 0.0776, "step": 9203 }, { "epoch": 4.87, "learning_rate": 3.413271645402283e-08, "loss": 0.0814, "step": 9204 }, { "epoch": 4.87, "learning_rate": 3.38502046539646e-08, "loss": 0.0989, "step": 9205 }, { "epoch": 4.87, "learning_rate": 3.356886489107591e-08, "loss": 0.0431, "step": 9206 }, { "epoch": 4.87, "learning_rate": 3.328869719844474e-08, "loss": 0.0584, "step": 9207 }, { "epoch": 4.87, "learning_rate": 3.300970160901806e-08, "loss": 0.0571, "step": 9208 }, { "epoch": 4.87, "learning_rate": 3.27318781556063e-08, "loss": 0.0873, "step": 9209 }, { "epoch": 4.87, "learning_rate": 3.245522687088332e-08, "loss": 0.045, "step": 9210 }, { "epoch": 4.88, "learning_rate": 3.217974778738198e-08, "loss": 0.073, "step": 9211 }, { "epoch": 4.88, "learning_rate": 3.190544093750081e-08, "loss": 0.0779, "step": 9212 }, { "epoch": 4.88, "learning_rate": 3.163230635349734e-08, "loss": 0.0471, "step": 9213 }, { "epoch": 4.88, "learning_rate": 3.1360344067493664e-08, "loss": 0.0886, "step": 9214 }, { "epoch": 4.88, "learning_rate": 3.1089554111473075e-08, "loss": 0.0898, "step": 9215 }, { "epoch": 4.88, "learning_rate": 3.08199365172801e-08, "loss": 0.0712, "step": 9216 }, { "epoch": 4.88, "learning_rate": 3.055149131662272e-08, "loss": 0.0449, "step": 9217 }, { "epoch": 4.88, "learning_rate": 3.028421854107122e-08, "loss": 0.0842, "step": 9218 }, { "epoch": 4.88, "learning_rate": 3.001811822205602e-08, "loss": 0.0507, "step": 9219 }, { "epoch": 4.88, "learning_rate": 2.97531903908721e-08, "loss": 0.0927, "step": 9220 }, { "epoch": 4.88, "learning_rate": 2.948943507867452e-08, "loss": 0.0912, "step": 9221 }, { "epoch": 4.88, "learning_rate": 2.9226852316482922e-08, "loss": 0.0763, "step": 9222 }, { "epoch": 4.88, "learning_rate": 2.8965442135175937e-08, "loss": 0.0768, "step": 9223 }, { "epoch": 4.88, "learning_rate": 2.870520456549675e-08, "loss": 0.0678, "step": 9224 }, { "epoch": 4.88, "learning_rate": 2.8446139638049762e-08, "loss": 0.0425, "step": 9225 }, { "epoch": 4.88, "learning_rate": 2.8188247383299505e-08, "loss": 0.0778, "step": 9226 }, { "epoch": 4.88, "learning_rate": 2.7931527831578374e-08, "loss": 0.0914, "step": 9227 }, { "epoch": 4.88, "learning_rate": 2.7675981013073338e-08, "loss": 0.0798, "step": 9228 }, { "epoch": 4.89, "learning_rate": 2.742160695784035e-08, "loss": 0.0617, "step": 9229 }, { "epoch": 4.89, "learning_rate": 2.7168405695791044e-08, "loss": 0.064, "step": 9230 }, { "epoch": 4.89, "learning_rate": 2.6916377256704927e-08, "loss": 0.0756, "step": 9231 }, { "epoch": 4.89, "learning_rate": 2.666552167022052e-08, "loss": 0.0795, "step": 9232 }, { "epoch": 4.89, "learning_rate": 2.6415838965837548e-08, "loss": 0.0723, "step": 9233 }, { "epoch": 4.89, "learning_rate": 2.6167329172920308e-08, "loss": 0.0587, "step": 9234 }, { "epoch": 4.89, "learning_rate": 2.59199923206932e-08, "loss": 0.0551, "step": 9235 }, { "epoch": 4.89, "learning_rate": 2.5673828438245176e-08, "loss": 0.1137, "step": 9236 }, { "epoch": 4.89, "learning_rate": 2.542883755452308e-08, "loss": 0.1058, "step": 9237 }, { "epoch": 4.89, "learning_rate": 2.5185019698340528e-08, "loss": 0.0877, "step": 9238 }, { "epoch": 4.89, "learning_rate": 2.4942374898369036e-08, "loss": 0.0515, "step": 9239 }, { "epoch": 4.89, "learning_rate": 2.4700903183145774e-08, "loss": 0.0588, "step": 9240 }, { "epoch": 4.89, "learning_rate": 2.4460604581068026e-08, "loss": 0.0645, "step": 9241 }, { "epoch": 4.89, "learning_rate": 2.4221479120393186e-08, "loss": 0.1023, "step": 9242 }, { "epoch": 4.89, "learning_rate": 2.398352682924543e-08, "loss": 0.0487, "step": 9243 }, { "epoch": 4.89, "learning_rate": 2.3746747735607923e-08, "loss": 0.0717, "step": 9244 }, { "epoch": 4.89, "learning_rate": 2.351114186732506e-08, "loss": 0.0644, "step": 9245 }, { "epoch": 4.89, "learning_rate": 2.327670925210468e-08, "loss": 0.0927, "step": 9246 }, { "epoch": 4.89, "learning_rate": 2.3043449917518057e-08, "loss": 0.0514, "step": 9247 }, { "epoch": 4.9, "learning_rate": 2.2811363890995474e-08, "loss": 0.1015, "step": 9248 }, { "epoch": 4.9, "learning_rate": 2.2580451199831764e-08, "loss": 0.0639, "step": 9249 }, { "epoch": 4.9, "learning_rate": 2.2350711871180765e-08, "loss": 0.088, "step": 9250 }, { "epoch": 4.9, "learning_rate": 2.212214593206308e-08, "loss": 0.0629, "step": 9251 }, { "epoch": 4.9, "learning_rate": 2.1894753409354985e-08, "loss": 0.0512, "step": 9252 }, { "epoch": 4.9, "learning_rate": 2.166853432980176e-08, "loss": 0.0938, "step": 9253 }, { "epoch": 4.9, "learning_rate": 2.1443488720004345e-08, "loss": 0.0648, "step": 9254 }, { "epoch": 4.9, "learning_rate": 2.121961660643046e-08, "loss": 0.1022, "step": 9255 }, { "epoch": 4.9, "learning_rate": 2.099691801540682e-08, "loss": 0.038, "step": 9256 }, { "epoch": 4.9, "learning_rate": 2.0775392973123587e-08, "loss": 0.0814, "step": 9257 }, { "epoch": 4.9, "learning_rate": 2.0555041505632146e-08, "loss": 0.068, "step": 9258 }, { "epoch": 4.9, "learning_rate": 2.033586363884732e-08, "loss": 0.078, "step": 9259 }, { "epoch": 4.9, "learning_rate": 2.0117859398542938e-08, "loss": 0.0696, "step": 9260 }, { "epoch": 4.9, "learning_rate": 1.990102881035849e-08, "loss": 0.0419, "step": 9261 }, { "epoch": 4.9, "learning_rate": 1.9685371899792472e-08, "loss": 0.1113, "step": 9262 }, { "epoch": 4.9, "learning_rate": 1.9470888692207924e-08, "loss": 0.0576, "step": 9263 }, { "epoch": 4.9, "learning_rate": 1.9257579212826892e-08, "loss": 0.066, "step": 9264 }, { "epoch": 4.9, "learning_rate": 1.9045443486734872e-08, "loss": 0.074, "step": 9265 }, { "epoch": 4.9, "learning_rate": 1.883448153888079e-08, "loss": 0.0445, "step": 9266 }, { "epoch": 4.91, "learning_rate": 1.862469339407258e-08, "loss": 0.0673, "step": 9267 }, { "epoch": 4.91, "learning_rate": 1.841607907698384e-08, "loss": 0.0679, "step": 9268 }, { "epoch": 4.91, "learning_rate": 1.8208638612146057e-08, "loss": 0.0546, "step": 9269 }, { "epoch": 4.91, "learning_rate": 1.8002372023954162e-08, "loss": 0.0689, "step": 9270 }, { "epoch": 4.91, "learning_rate": 1.779727933666875e-08, "loss": 0.0869, "step": 9271 }, { "epoch": 4.91, "learning_rate": 1.7593360574406082e-08, "loss": 0.0754, "step": 9272 }, { "epoch": 4.91, "learning_rate": 1.739061576114698e-08, "loss": 0.0719, "step": 9273 }, { "epoch": 4.91, "learning_rate": 1.7189044920736808e-08, "loss": 0.0311, "step": 9274 }, { "epoch": 4.91, "learning_rate": 1.6988648076879944e-08, "loss": 0.0621, "step": 9275 }, { "epoch": 4.91, "learning_rate": 1.6789425253143088e-08, "loss": 0.0717, "step": 9276 }, { "epoch": 4.91, "learning_rate": 1.659137647295528e-08, "loss": 0.0522, "step": 9277 }, { "epoch": 4.91, "learning_rate": 1.6394501759606774e-08, "loss": 0.0626, "step": 9278 }, { "epoch": 4.91, "learning_rate": 1.6198801136251276e-08, "loss": 0.0773, "step": 9279 }, { "epoch": 4.91, "learning_rate": 1.6004274625903704e-08, "loss": 0.0823, "step": 9280 }, { "epoch": 4.91, "learning_rate": 1.581092225144021e-08, "loss": 0.0729, "step": 9281 }, { "epoch": 4.91, "learning_rate": 1.5618744035599266e-08, "loss": 0.0831, "step": 9282 }, { "epoch": 4.91, "learning_rate": 1.5427740000981682e-08, "loss": 0.0807, "step": 9283 }, { "epoch": 4.91, "learning_rate": 1.5237910170049496e-08, "loss": 0.075, "step": 9284 }, { "epoch": 4.91, "learning_rate": 1.504925456512707e-08, "loss": 0.0712, "step": 9285 }, { "epoch": 4.92, "learning_rate": 1.4861773208402208e-08, "loss": 0.0818, "step": 9286 }, { "epoch": 4.92, "learning_rate": 1.4675466121920613e-08, "loss": 0.0511, "step": 9287 }, { "epoch": 4.92, "learning_rate": 1.4490333327593642e-08, "loss": 0.0574, "step": 9288 }, { "epoch": 4.92, "learning_rate": 1.4306374847192773e-08, "loss": 0.0735, "step": 9289 }, { "epoch": 4.92, "learning_rate": 1.4123590702352918e-08, "loss": 0.0678, "step": 9290 }, { "epoch": 4.92, "learning_rate": 1.394198091456911e-08, "loss": 0.0618, "step": 9291 }, { "epoch": 4.92, "learning_rate": 1.3761545505197593e-08, "loss": 0.0597, "step": 9292 }, { "epoch": 4.92, "learning_rate": 1.3582284495460286e-08, "loss": 0.0606, "step": 9293 }, { "epoch": 4.92, "learning_rate": 1.340419790643699e-08, "loss": 0.0832, "step": 9294 }, { "epoch": 4.92, "learning_rate": 1.3227285759072061e-08, "loss": 0.0502, "step": 9295 }, { "epoch": 4.92, "learning_rate": 1.3051548074169973e-08, "loss": 0.0648, "step": 9296 }, { "epoch": 4.92, "learning_rate": 1.2876984872398634e-08, "loss": 0.0407, "step": 9297 }, { "epoch": 4.92, "learning_rate": 1.2703596174286071e-08, "loss": 0.0987, "step": 9298 }, { "epoch": 4.92, "learning_rate": 1.2531382000222637e-08, "loss": 0.0592, "step": 9299 }, { "epoch": 4.92, "learning_rate": 1.2360342370462131e-08, "loss": 0.0616, "step": 9300 }, { "epoch": 4.92, "learning_rate": 1.2190477305119575e-08, "loss": 0.0505, "step": 9301 }, { "epoch": 4.92, "learning_rate": 1.202178682416899e-08, "loss": 0.0869, "step": 9302 }, { "epoch": 4.92, "learning_rate": 1.1854270947452285e-08, "loss": 0.0794, "step": 9303 }, { "epoch": 4.92, "learning_rate": 1.1687929694665923e-08, "loss": 0.0855, "step": 9304 }, { "epoch": 4.93, "learning_rate": 1.1522763085375366e-08, "loss": 0.0604, "step": 9305 }, { "epoch": 4.93, "learning_rate": 1.1358771139001745e-08, "loss": 0.0599, "step": 9306 }, { "epoch": 4.93, "learning_rate": 1.1195953874831855e-08, "loss": 0.0439, "step": 9307 }, { "epoch": 4.93, "learning_rate": 1.1034311312013713e-08, "loss": 0.0919, "step": 9308 }, { "epoch": 4.93, "learning_rate": 1.0873843469555446e-08, "loss": 0.0715, "step": 9309 }, { "epoch": 4.93, "learning_rate": 1.0714550366329735e-08, "loss": 0.0814, "step": 9310 }, { "epoch": 4.93, "learning_rate": 1.0556432021069375e-08, "loss": 0.1003, "step": 9311 }, { "epoch": 4.93, "learning_rate": 1.0399488452370599e-08, "loss": 0.048, "step": 9312 }, { "epoch": 4.93, "learning_rate": 1.0243719678687537e-08, "loss": 0.0468, "step": 9313 }, { "epoch": 4.93, "learning_rate": 1.0089125718339976e-08, "loss": 0.0713, "step": 9314 }, { "epoch": 4.93, "learning_rate": 9.93570658950893e-09, "loss": 0.0647, "step": 9315 }, { "epoch": 4.93, "learning_rate": 9.783462310237746e-09, "loss": 0.0727, "step": 9316 }, { "epoch": 4.93, "learning_rate": 9.632392898428766e-09, "loss": 0.0716, "step": 9317 }, { "epoch": 4.93, "learning_rate": 9.482498371847782e-09, "loss": 0.0862, "step": 9318 }, { "epoch": 4.93, "learning_rate": 9.333778748125133e-09, "loss": 0.0617, "step": 9319 }, { "epoch": 4.93, "learning_rate": 9.186234044747944e-09, "loss": 0.0667, "step": 9320 }, { "epoch": 4.93, "learning_rate": 9.039864279069e-09, "loss": 0.0724, "step": 9321 }, { "epoch": 4.93, "learning_rate": 8.894669468302308e-09, "loss": 0.064, "step": 9322 }, { "epoch": 4.93, "learning_rate": 8.75064962952088e-09, "loss": 0.1355, "step": 9323 }, { "epoch": 4.94, "learning_rate": 8.607804779663386e-09, "loss": 0.0586, "step": 9324 }, { "epoch": 4.94, "learning_rate": 8.466134935528614e-09, "loss": 0.053, "step": 9325 }, { "epoch": 4.94, "learning_rate": 8.325640113776568e-09, "loss": 0.1021, "step": 9326 }, { "epoch": 4.94, "learning_rate": 8.186320330929588e-09, "loss": 0.0649, "step": 9327 }, { "epoch": 4.94, "learning_rate": 8.048175603372344e-09, "loss": 0.0559, "step": 9328 }, { "epoch": 4.94, "learning_rate": 7.911205947351842e-09, "loss": 0.06, "step": 9329 }, { "epoch": 4.94, "learning_rate": 7.775411378974085e-09, "loss": 0.0438, "step": 9330 }, { "epoch": 4.94, "learning_rate": 7.640791914209633e-09, "loss": 0.0801, "step": 9331 }, { "epoch": 4.94, "learning_rate": 7.507347568890267e-09, "loss": 0.0778, "step": 9332 }, { "epoch": 4.94, "learning_rate": 7.375078358708987e-09, "loss": 0.0665, "step": 9333 }, { "epoch": 4.94, "learning_rate": 7.243984299220019e-09, "loss": 0.0871, "step": 9334 }, { "epoch": 4.94, "learning_rate": 7.114065405842141e-09, "loss": 0.0564, "step": 9335 }, { "epoch": 4.94, "learning_rate": 6.985321693853131e-09, "loss": 0.0544, "step": 9336 }, { "epoch": 4.94, "learning_rate": 6.8577531783931e-09, "loss": 0.0844, "step": 9337 }, { "epoch": 4.94, "learning_rate": 6.731359874464493e-09, "loss": 0.0516, "step": 9338 }, { "epoch": 4.94, "learning_rate": 6.606141796930976e-09, "loss": 0.0614, "step": 9339 }, { "epoch": 4.94, "learning_rate": 6.482098960519656e-09, "loss": 0.0884, "step": 9340 }, { "epoch": 4.94, "learning_rate": 6.3592313798177544e-09, "loss": 0.1308, "step": 9341 }, { "epoch": 4.94, "learning_rate": 6.2375390692748225e-09, "loss": 0.0612, "step": 9342 }, { "epoch": 4.95, "learning_rate": 6.1170220432005264e-09, "loss": 0.0548, "step": 9343 }, { "epoch": 4.95, "learning_rate": 5.9976803157701934e-09, "loss": 0.0613, "step": 9344 }, { "epoch": 4.95, "learning_rate": 5.879513901017042e-09, "loss": 0.0804, "step": 9345 }, { "epoch": 4.95, "learning_rate": 5.762522812837734e-09, "loss": 0.0782, "step": 9346 }, { "epoch": 4.95, "learning_rate": 5.646707064991264e-09, "loss": 0.0612, "step": 9347 }, { "epoch": 4.95, "learning_rate": 5.532066671097846e-09, "loss": 0.1108, "step": 9348 }, { "epoch": 4.95, "learning_rate": 5.418601644640032e-09, "loss": 0.0793, "step": 9349 }, { "epoch": 4.95, "learning_rate": 5.306311998959368e-09, "loss": 0.1088, "step": 9350 }, { "epoch": 4.95, "learning_rate": 5.19519774726307e-09, "loss": 0.0941, "step": 9351 }, { "epoch": 4.95, "learning_rate": 5.085258902617351e-09, "loss": 0.1048, "step": 9352 }, { "epoch": 4.95, "learning_rate": 4.976495477951871e-09, "loss": 0.0656, "step": 9353 }, { "epoch": 4.95, "learning_rate": 4.868907486057506e-09, "loss": 0.0353, "step": 9354 }, { "epoch": 4.95, "learning_rate": 4.76249493958636e-09, "loss": 0.0472, "step": 9355 }, { "epoch": 4.95, "learning_rate": 4.6572578510528655e-09, "loss": 0.0616, "step": 9356 }, { "epoch": 4.95, "learning_rate": 4.5531962328326795e-09, "loss": 0.0791, "step": 9357 }, { "epoch": 4.95, "learning_rate": 4.4503100971649e-09, "loss": 0.0693, "step": 9358 }, { "epoch": 4.95, "learning_rate": 4.348599456148739e-09, "loss": 0.0621, "step": 9359 }, { "epoch": 4.95, "learning_rate": 4.248064321744627e-09, "loss": 0.0788, "step": 9360 }, { "epoch": 4.95, "learning_rate": 4.14870470577533e-09, "loss": 0.0941, "step": 9361 }, { "epoch": 4.96, "learning_rate": 4.050520619927056e-09, "loss": 0.0618, "step": 9362 }, { "epoch": 4.96, "learning_rate": 3.953512075747234e-09, "loss": 0.0702, "step": 9363 }, { "epoch": 4.96, "learning_rate": 3.857679084641186e-09, "loss": 0.0785, "step": 9364 }, { "epoch": 4.96, "learning_rate": 3.763021657882115e-09, "loss": 0.0921, "step": 9365 }, { "epoch": 4.96, "learning_rate": 3.669539806600009e-09, "loss": 0.066, "step": 9366 }, { "epoch": 4.96, "learning_rate": 3.5772335417894044e-09, "loss": 0.0643, "step": 9367 }, { "epoch": 4.96, "learning_rate": 3.486102874304953e-09, "loss": 0.072, "step": 9368 }, { "epoch": 4.96, "learning_rate": 3.396147814863637e-09, "loss": 0.056, "step": 9369 }, { "epoch": 4.96, "learning_rate": 3.307368374045883e-09, "loss": 0.0513, "step": 9370 }, { "epoch": 4.96, "learning_rate": 3.2197645622911167e-09, "loss": 0.079, "step": 9371 }, { "epoch": 4.96, "learning_rate": 3.1333363899022083e-09, "loss": 0.0614, "step": 9372 }, { "epoch": 4.96, "learning_rate": 3.04808386704325e-09, "loss": 0.0513, "step": 9373 }, { "epoch": 4.96, "learning_rate": 2.964007003738445e-09, "loss": 0.0763, "step": 9374 }, { "epoch": 4.96, "learning_rate": 2.88110580987766e-09, "loss": 0.0634, "step": 9375 }, { "epoch": 4.96, "learning_rate": 2.7993802952097636e-09, "loss": 0.0701, "step": 9376 }, { "epoch": 4.96, "learning_rate": 2.7188304693448465e-09, "loss": 0.0476, "step": 9377 }, { "epoch": 4.96, "learning_rate": 2.6394563417564412e-09, "loss": 0.0677, "step": 9378 }, { "epoch": 4.96, "learning_rate": 2.5612579217793033e-09, "loss": 0.0683, "step": 9379 }, { "epoch": 4.96, "learning_rate": 2.4842352186082995e-09, "loss": 0.0451, "step": 9380 }, { "epoch": 4.97, "learning_rate": 2.4083882413028502e-09, "loss": 0.0485, "step": 9381 }, { "epoch": 4.97, "learning_rate": 2.333716998782487e-09, "loss": 0.0711, "step": 9382 }, { "epoch": 4.97, "learning_rate": 2.260221499829074e-09, "loss": 0.0446, "step": 9383 }, { "epoch": 4.97, "learning_rate": 2.187901753084587e-09, "loss": 0.0738, "step": 9384 }, { "epoch": 4.97, "learning_rate": 2.116757767054445e-09, "loss": 0.0589, "step": 9385 }, { "epoch": 4.97, "learning_rate": 2.046789550105288e-09, "loss": 0.1024, "step": 9386 }, { "epoch": 4.97, "learning_rate": 1.97799711046609e-09, "loss": 0.0502, "step": 9387 }, { "epoch": 4.97, "learning_rate": 1.9103804562270458e-09, "loss": 0.0665, "step": 9388 }, { "epoch": 4.97, "learning_rate": 1.843939595338462e-09, "loss": 0.1186, "step": 9389 }, { "epoch": 4.97, "learning_rate": 1.7786745356163094e-09, "loss": 0.0831, "step": 9390 }, { "epoch": 4.97, "learning_rate": 1.7145852847333388e-09, "loss": 0.0915, "step": 9391 }, { "epoch": 4.97, "learning_rate": 1.6516718502279649e-09, "loss": 0.0437, "step": 9392 }, { "epoch": 4.97, "learning_rate": 1.5899342394987138e-09, "loss": 0.0637, "step": 9393 }, { "epoch": 4.97, "learning_rate": 1.5293724598064441e-09, "loss": 0.0809, "step": 9394 }, { "epoch": 4.97, "learning_rate": 1.4699865182721262e-09, "loss": 0.0756, "step": 9395 }, { "epoch": 4.97, "learning_rate": 1.4117764218812836e-09, "loss": 0.043, "step": 9396 }, { "epoch": 4.97, "learning_rate": 1.3547421774784408e-09, "loss": 0.0951, "step": 9397 }, { "epoch": 4.97, "learning_rate": 1.2988837917704555e-09, "loss": 0.089, "step": 9398 }, { "epoch": 4.97, "learning_rate": 1.2442012713276275e-09, "loss": 0.0496, "step": 9399 }, { "epoch": 4.98, "learning_rate": 1.1906946225803685e-09, "loss": 0.1059, "step": 9400 }, { "epoch": 4.98, "learning_rate": 1.1383638518214223e-09, "loss": 0.1016, "step": 9401 }, { "epoch": 4.98, "learning_rate": 1.0872089652036455e-09, "loss": 0.0653, "step": 9402 }, { "epoch": 4.98, "learning_rate": 1.0372299687444466e-09, "loss": 0.0831, "step": 9403 }, { "epoch": 4.98, "learning_rate": 9.884268683202358e-10, "loss": 0.0936, "step": 9404 }, { "epoch": 4.98, "learning_rate": 9.407996696708666e-10, "loss": 0.0786, "step": 9405 }, { "epoch": 4.98, "learning_rate": 8.943483783985241e-10, "loss": 0.0507, "step": 9406 }, { "epoch": 4.98, "learning_rate": 8.490729999643954e-10, "loss": 0.0819, "step": 9407 }, { "epoch": 4.98, "learning_rate": 8.049735396931102e-10, "loss": 0.0458, "step": 9408 }, { "epoch": 4.98, "learning_rate": 7.620500027705202e-10, "loss": 0.0829, "step": 9409 }, { "epoch": 4.98, "learning_rate": 7.2030239424703e-10, "loss": 0.0614, "step": 9410 }, { "epoch": 4.98, "learning_rate": 6.79730719028715e-10, "loss": 0.0552, "step": 9411 }, { "epoch": 4.98, "learning_rate": 6.403349818895343e-10, "loss": 0.0944, "step": 9412 }, { "epoch": 4.98, "learning_rate": 6.021151874602282e-10, "loss": 0.1005, "step": 9413 }, { "epoch": 4.98, "learning_rate": 5.650713402383101e-10, "loss": 0.0548, "step": 9414 }, { "epoch": 4.98, "learning_rate": 5.292034445780747e-10, "loss": 0.0786, "step": 9415 }, { "epoch": 4.98, "learning_rate": 4.945115046983695e-10, "loss": 0.0533, "step": 9416 }, { "epoch": 4.98, "learning_rate": 4.609955246781539e-10, "loss": 0.0501, "step": 9417 }, { "epoch": 4.99, "learning_rate": 4.2865550846094005e-10, "loss": 0.0827, "step": 9418 }, { "epoch": 4.99, "learning_rate": 3.9749145984813163e-10, "loss": 0.091, "step": 9419 }, { "epoch": 4.99, "learning_rate": 3.675033825045748e-10, "loss": 0.055, "step": 9420 }, { "epoch": 4.99, "learning_rate": 3.386912799585584e-10, "loss": 0.0583, "step": 9421 }, { "epoch": 4.99, "learning_rate": 3.1105515559737287e-10, "loss": 0.069, "step": 9422 }, { "epoch": 4.99, "learning_rate": 2.8459501267175117e-10, "loss": 0.0978, "step": 9423 }, { "epoch": 4.99, "learning_rate": 2.593108542925382e-10, "loss": 0.0552, "step": 9424 }, { "epoch": 4.99, "learning_rate": 2.352026834329113e-10, "loss": 0.1155, "step": 9425 }, { "epoch": 4.99, "learning_rate": 2.1227050292949024e-10, "loss": 0.0606, "step": 9426 }, { "epoch": 4.99, "learning_rate": 1.9051431547789655e-10, "loss": 0.0781, "step": 9427 }, { "epoch": 4.99, "learning_rate": 1.699341236371943e-10, "loss": 0.0773, "step": 9428 }, { "epoch": 4.99, "learning_rate": 1.5052992982766968e-10, "loss": 0.1153, "step": 9429 }, { "epoch": 4.99, "learning_rate": 1.3230173633194121e-10, "loss": 0.0416, "step": 9430 }, { "epoch": 4.99, "learning_rate": 1.1524954529273935e-10, "loss": 0.0407, "step": 9431 }, { "epoch": 4.99, "learning_rate": 9.937335871512688e-11, "loss": 0.1031, "step": 9432 }, { "epoch": 4.99, "learning_rate": 8.467317846649891e-11, "loss": 0.0735, "step": 9433 }, { "epoch": 4.99, "learning_rate": 7.114900627658295e-11, "loss": 0.0477, "step": 9434 }, { "epoch": 4.99, "learning_rate": 5.880084373521833e-11, "loss": 0.0567, "step": 9435 }, { "epoch": 4.99, "learning_rate": 4.7628692294576783e-11, "loss": 0.0908, "step": 9436 }, { "epoch": 5.0, "learning_rate": 3.763255326805215e-11, "loss": 0.0613, "step": 9437 }, { "epoch": 5.0, "learning_rate": 2.8812427831370616e-11, "loss": 0.082, "step": 9438 }, { "epoch": 5.0, "learning_rate": 2.116831702148048e-11, "loss": 0.0449, "step": 9439 }, { "epoch": 5.0, "learning_rate": 1.4700221738772612e-11, "loss": 0.1036, "step": 9440 }, { "epoch": 5.0, "learning_rate": 9.40814274263957e-12, "loss": 0.0573, "step": 9441 }, { "epoch": 5.0, "learning_rate": 5.292080655916465e-12, "loss": 0.0915, "step": 9442 }, { "epoch": 5.0, "learning_rate": 2.352035962660537e-12, "loss": 0.0812, "step": 9443 }, { "epoch": 5.0, "learning_rate": 5.880090081511469e-13, "loss": 0.071, "step": 9444 }, { "epoch": 5.0, "learning_rate": 0.0, "loss": 0.0659, "step": 9445 }, { "epoch": 5.0, "step": 9445, "total_flos": 3.382227231844598e+17, "train_loss": 0.1461414268314286, "train_runtime": 18556.1974, "train_samples_per_second": 14.252, "train_steps_per_second": 0.509 } ], "max_steps": 9445, "num_train_epochs": 5, "total_flos": 3.382227231844598e+17, "trial_name": null, "trial_params": null }