{ "best_metric": 0.24734805524349213, "best_model_checkpoint": "./vit-mae-demo/checkpoint-7705", "epoch": 1.0, "global_step": 7705, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.2143782383419688e-07, "loss": 1.633, "step": 10 }, { "epoch": 0.0, "learning_rate": 2.4287564766839375e-07, "loss": 1.7508, "step": 20 }, { "epoch": 0.0, "learning_rate": 3.6431347150259065e-07, "loss": 1.6418, "step": 30 }, { "epoch": 0.01, "learning_rate": 4.857512953367875e-07, "loss": 1.6749, "step": 40 }, { "epoch": 0.01, "learning_rate": 6.071891191709843e-07, "loss": 1.609, "step": 50 }, { "epoch": 0.01, "learning_rate": 7.286269430051813e-07, "loss": 1.5425, "step": 60 }, { "epoch": 0.01, "learning_rate": 8.500647668393782e-07, "loss": 1.4211, "step": 70 }, { "epoch": 0.01, "learning_rate": 9.71502590673575e-07, "loss": 1.4454, "step": 80 }, { "epoch": 0.01, "learning_rate": 1.092940414507772e-06, "loss": 1.2961, "step": 90 }, { "epoch": 0.01, "learning_rate": 1.2143782383419686e-06, "loss": 0.9962, "step": 100 }, { "epoch": 0.01, "learning_rate": 1.3358160621761657e-06, "loss": 0.8856, "step": 110 }, { "epoch": 0.02, "learning_rate": 1.4572538860103626e-06, "loss": 0.8172, "step": 120 }, { "epoch": 0.02, "learning_rate": 1.5786917098445593e-06, "loss": 0.7092, "step": 130 }, { "epoch": 0.02, "learning_rate": 1.7001295336787564e-06, "loss": 0.6926, "step": 140 }, { "epoch": 0.02, "learning_rate": 1.8215673575129531e-06, "loss": 0.6143, "step": 150 }, { "epoch": 0.02, "learning_rate": 1.94300518134715e-06, "loss": 0.5924, "step": 160 }, { "epoch": 0.02, "learning_rate": 2.064443005181347e-06, "loss": 0.5705, "step": 170 }, { "epoch": 0.02, "learning_rate": 2.185880829015544e-06, "loss": 0.5762, "step": 180 }, { "epoch": 0.02, "learning_rate": 2.3073186528497407e-06, "loss": 0.5635, "step": 190 }, { "epoch": 0.03, "learning_rate": 2.428756476683937e-06, "loss": 0.5498, "step": 200 }, { "epoch": 0.03, "learning_rate": 2.5501943005181345e-06, "loss": 0.5213, "step": 210 }, { "epoch": 0.03, "learning_rate": 2.6716321243523314e-06, "loss": 0.5022, "step": 220 }, { "epoch": 0.03, "learning_rate": 2.7930699481865283e-06, "loss": 0.52, "step": 230 }, { "epoch": 0.03, "learning_rate": 2.9145077720207252e-06, "loss": 0.5233, "step": 240 }, { "epoch": 0.03, "learning_rate": 3.0359455958549217e-06, "loss": 0.4757, "step": 250 }, { "epoch": 0.03, "learning_rate": 3.1573834196891186e-06, "loss": 0.4844, "step": 260 }, { "epoch": 0.04, "learning_rate": 3.278821243523316e-06, "loss": 0.4468, "step": 270 }, { "epoch": 0.04, "learning_rate": 3.400259067357513e-06, "loss": 0.4388, "step": 280 }, { "epoch": 0.04, "learning_rate": 3.5216968911917097e-06, "loss": 0.4308, "step": 290 }, { "epoch": 0.04, "learning_rate": 3.6431347150259062e-06, "loss": 0.4415, "step": 300 }, { "epoch": 0.04, "learning_rate": 3.764572538860103e-06, "loss": 0.4142, "step": 310 }, { "epoch": 0.04, "learning_rate": 3.8860103626943e-06, "loss": 0.3854, "step": 320 }, { "epoch": 0.04, "learning_rate": 4.007448186528497e-06, "loss": 0.3994, "step": 330 }, { "epoch": 0.04, "learning_rate": 4.128886010362694e-06, "loss": 0.3839, "step": 340 }, { "epoch": 0.05, "learning_rate": 4.250323834196891e-06, "loss": 0.4047, "step": 350 }, { "epoch": 0.05, "learning_rate": 4.371761658031088e-06, "loss": 0.4031, "step": 360 }, { "epoch": 0.05, "learning_rate": 4.4931994818652845e-06, "loss": 0.386, "step": 370 }, { "epoch": 0.05, "learning_rate": 4.6146373056994814e-06, "loss": 0.3897, "step": 380 }, { "epoch": 0.05, "learning_rate": 4.6874965454028015e-06, "loss": 0.374, "step": 390 }, { "epoch": 0.05, "learning_rate": 4.687457681301278e-06, "loss": 0.3712, "step": 400 }, { "epoch": 0.05, "learning_rate": 4.687375635570173e-06, "loss": 0.3602, "step": 410 }, { "epoch": 0.05, "learning_rate": 4.687250409721135e-06, "loss": 0.3419, "step": 420 }, { "epoch": 0.06, "learning_rate": 4.687082006061389e-06, "loss": 0.3584, "step": 430 }, { "epoch": 0.06, "learning_rate": 4.686870427693687e-06, "loss": 0.3601, "step": 440 }, { "epoch": 0.06, "learning_rate": 4.686615678516253e-06, "loss": 0.3556, "step": 450 }, { "epoch": 0.06, "learning_rate": 4.686317763222712e-06, "loss": 0.368, "step": 460 }, { "epoch": 0.06, "learning_rate": 4.685976687302003e-06, "loss": 0.3405, "step": 470 }, { "epoch": 0.06, "learning_rate": 4.685592457038277e-06, "loss": 0.3504, "step": 480 }, { "epoch": 0.06, "learning_rate": 4.6851650795107834e-06, "loss": 0.3321, "step": 490 }, { "epoch": 0.06, "learning_rate": 4.684694562593737e-06, "loss": 0.3395, "step": 500 }, { "epoch": 0.07, "learning_rate": 4.684180914956173e-06, "loss": 0.3264, "step": 510 }, { "epoch": 0.07, "learning_rate": 4.68362414606179e-06, "loss": 0.3453, "step": 520 }, { "epoch": 0.07, "learning_rate": 4.683024266168774e-06, "loss": 0.3395, "step": 530 }, { "epoch": 0.07, "learning_rate": 4.682381286329609e-06, "loss": 0.3219, "step": 540 }, { "epoch": 0.07, "learning_rate": 4.6816952183908725e-06, "loss": 0.3328, "step": 550 }, { "epoch": 0.07, "learning_rate": 4.680966074993021e-06, "loss": 0.3338, "step": 560 }, { "epoch": 0.07, "learning_rate": 4.6801938695701525e-06, "loss": 0.3177, "step": 570 }, { "epoch": 0.08, "learning_rate": 4.679378616349761e-06, "loss": 0.323, "step": 580 }, { "epoch": 0.08, "learning_rate": 4.678520330352476e-06, "loss": 0.3308, "step": 590 }, { "epoch": 0.08, "learning_rate": 4.677619027391783e-06, "loss": 0.3213, "step": 600 }, { "epoch": 0.08, "learning_rate": 4.676674724073732e-06, "loss": 0.313, "step": 610 }, { "epoch": 0.08, "learning_rate": 4.675687437796638e-06, "loss": 0.3182, "step": 620 }, { "epoch": 0.08, "learning_rate": 4.674657186750749e-06, "loss": 0.3359, "step": 630 }, { "epoch": 0.08, "learning_rate": 4.673583989917919e-06, "loss": 0.3065, "step": 640 }, { "epoch": 0.08, "learning_rate": 4.67246786707126e-06, "loss": 0.3058, "step": 650 }, { "epoch": 0.09, "learning_rate": 4.671308838774771e-06, "loss": 0.3154, "step": 660 }, { "epoch": 0.09, "learning_rate": 4.67010692638296e-06, "loss": 0.318, "step": 670 }, { "epoch": 0.09, "learning_rate": 4.668862152040458e-06, "loss": 0.2885, "step": 680 }, { "epoch": 0.09, "learning_rate": 4.667574538681602e-06, "loss": 0.2966, "step": 690 }, { "epoch": 0.09, "learning_rate": 4.666244110030018e-06, "loss": 0.3021, "step": 700 }, { "epoch": 0.09, "learning_rate": 4.664870890598181e-06, "loss": 0.3051, "step": 710 }, { "epoch": 0.09, "learning_rate": 4.663454905686965e-06, "loss": 0.3147, "step": 720 }, { "epoch": 0.09, "learning_rate": 4.661996181385176e-06, "loss": 0.3142, "step": 730 }, { "epoch": 0.1, "learning_rate": 4.660494744569075e-06, "loss": 0.3004, "step": 740 }, { "epoch": 0.1, "learning_rate": 4.658950622901873e-06, "loss": 0.3197, "step": 750 }, { "epoch": 0.1, "learning_rate": 4.6573638448332325e-06, "loss": 0.2984, "step": 760 }, { "epoch": 0.1, "learning_rate": 4.655734439598738e-06, "loss": 0.3315, "step": 770 }, { "epoch": 0.1, "learning_rate": 4.654062437219358e-06, "loss": 0.2924, "step": 780 }, { "epoch": 0.1, "learning_rate": 4.6523478685008895e-06, "loss": 0.2822, "step": 790 }, { "epoch": 0.1, "learning_rate": 4.650590765033395e-06, "loss": 0.3304, "step": 800 }, { "epoch": 0.11, "learning_rate": 4.64879115919062e-06, "loss": 0.3124, "step": 810 }, { "epoch": 0.11, "learning_rate": 4.64694908412939e-06, "loss": 0.2871, "step": 820 }, { "epoch": 0.11, "learning_rate": 4.645064573789009e-06, "loss": 0.2994, "step": 830 }, { "epoch": 0.11, "learning_rate": 4.643137662890627e-06, "loss": 0.292, "step": 840 }, { "epoch": 0.11, "learning_rate": 4.641168386936603e-06, "loss": 0.2934, "step": 850 }, { "epoch": 0.11, "learning_rate": 4.639156782209853e-06, "loss": 0.2974, "step": 860 }, { "epoch": 0.11, "learning_rate": 4.6371028857731744e-06, "loss": 0.3035, "step": 870 }, { "epoch": 0.11, "learning_rate": 4.6350067354685745e-06, "loss": 0.2723, "step": 880 }, { "epoch": 0.12, "learning_rate": 4.632868369916561e-06, "loss": 0.2703, "step": 890 }, { "epoch": 0.12, "learning_rate": 4.630687828515439e-06, "loss": 0.3041, "step": 900 }, { "epoch": 0.12, "learning_rate": 4.628465151440582e-06, "loss": 0.3032, "step": 910 }, { "epoch": 0.12, "learning_rate": 4.626200379643692e-06, "loss": 0.2793, "step": 920 }, { "epoch": 0.12, "learning_rate": 4.623893554852044e-06, "loss": 0.29, "step": 930 }, { "epoch": 0.12, "learning_rate": 4.62154471956772e-06, "loss": 0.2926, "step": 940 }, { "epoch": 0.12, "learning_rate": 4.619153917066823e-06, "loss": 0.2881, "step": 950 }, { "epoch": 0.12, "learning_rate": 4.61672119139868e-06, "loss": 0.2906, "step": 960 }, { "epoch": 0.13, "learning_rate": 4.614246587385033e-06, "loss": 0.3024, "step": 970 }, { "epoch": 0.13, "learning_rate": 4.61173015061921e-06, "loss": 0.2915, "step": 980 }, { "epoch": 0.13, "learning_rate": 4.6091719274652855e-06, "loss": 0.2686, "step": 990 }, { "epoch": 0.13, "learning_rate": 4.6065719650572305e-06, "loss": 0.2735, "step": 1000 }, { "epoch": 0.13, "learning_rate": 4.603930311298037e-06, "loss": 0.2904, "step": 1010 }, { "epoch": 0.13, "learning_rate": 4.60124701485884e-06, "loss": 0.2962, "step": 1020 }, { "epoch": 0.13, "learning_rate": 4.598522125178022e-06, "loss": 0.2818, "step": 1030 }, { "epoch": 0.13, "learning_rate": 4.595755692460297e-06, "loss": 0.2677, "step": 1040 }, { "epoch": 0.14, "learning_rate": 4.59294776767579e-06, "loss": 0.2872, "step": 1050 }, { "epoch": 0.14, "learning_rate": 4.590098402559097e-06, "loss": 0.2974, "step": 1060 }, { "epoch": 0.14, "learning_rate": 4.587207649608328e-06, "loss": 0.2918, "step": 1070 }, { "epoch": 0.14, "learning_rate": 4.584275562084148e-06, "loss": 0.2734, "step": 1080 }, { "epoch": 0.14, "learning_rate": 4.581302194008786e-06, "loss": 0.2886, "step": 1090 }, { "epoch": 0.14, "learning_rate": 4.578287600165046e-06, "loss": 0.2774, "step": 1100 }, { "epoch": 0.14, "learning_rate": 4.575231836095298e-06, "loss": 0.3036, "step": 1110 }, { "epoch": 0.15, "learning_rate": 4.5721349581004486e-06, "loss": 0.2828, "step": 1120 }, { "epoch": 0.15, "learning_rate": 4.56899702323891e-06, "loss": 0.2809, "step": 1130 }, { "epoch": 0.15, "learning_rate": 4.565818089325548e-06, "loss": 0.2933, "step": 1140 }, { "epoch": 0.15, "learning_rate": 4.562598214930612e-06, "loss": 0.2903, "step": 1150 }, { "epoch": 0.15, "learning_rate": 4.55933745937866e-06, "loss": 0.276, "step": 1160 }, { "epoch": 0.15, "learning_rate": 4.556035882747467e-06, "loss": 0.2784, "step": 1170 }, { "epoch": 0.15, "learning_rate": 4.552693545866911e-06, "loss": 0.2751, "step": 1180 }, { "epoch": 0.15, "learning_rate": 4.549310510317863e-06, "loss": 0.2766, "step": 1190 }, { "epoch": 0.16, "learning_rate": 4.545886838431042e-06, "loss": 0.2819, "step": 1200 }, { "epoch": 0.16, "learning_rate": 4.542422593285872e-06, "loss": 0.2866, "step": 1210 }, { "epoch": 0.16, "learning_rate": 4.538917838709319e-06, "loss": 0.2677, "step": 1220 }, { "epoch": 0.16, "learning_rate": 4.535372639274715e-06, "loss": 0.2785, "step": 1230 }, { "epoch": 0.16, "learning_rate": 4.531787060300568e-06, "loss": 0.2795, "step": 1240 }, { "epoch": 0.16, "learning_rate": 4.528161167849359e-06, "loss": 0.2861, "step": 1250 }, { "epoch": 0.16, "learning_rate": 4.524495028726325e-06, "loss": 0.2588, "step": 1260 }, { "epoch": 0.16, "learning_rate": 4.520788710478224e-06, "loss": 0.2642, "step": 1270 }, { "epoch": 0.17, "learning_rate": 4.517042281392097e-06, "loss": 0.2818, "step": 1280 }, { "epoch": 0.17, "learning_rate": 4.513255810494008e-06, "loss": 0.258, "step": 1290 }, { "epoch": 0.17, "learning_rate": 4.509429367547769e-06, "loss": 0.2768, "step": 1300 }, { "epoch": 0.17, "learning_rate": 4.505563023053659e-06, "loss": 0.2773, "step": 1310 }, { "epoch": 0.17, "learning_rate": 4.501656848247118e-06, "loss": 0.2791, "step": 1320 }, { "epoch": 0.17, "learning_rate": 4.497710915097446e-06, "loss": 0.2877, "step": 1330 }, { "epoch": 0.17, "learning_rate": 4.493725296306466e-06, "loss": 0.28, "step": 1340 }, { "epoch": 0.18, "learning_rate": 4.489700065307189e-06, "loss": 0.2788, "step": 1350 }, { "epoch": 0.18, "learning_rate": 4.485635296262463e-06, "loss": 0.2799, "step": 1360 }, { "epoch": 0.18, "learning_rate": 4.481531064063604e-06, "loss": 0.2745, "step": 1370 }, { "epoch": 0.18, "learning_rate": 4.477387444329015e-06, "loss": 0.2692, "step": 1380 }, { "epoch": 0.18, "learning_rate": 4.473204513402796e-06, "loss": 0.2897, "step": 1390 }, { "epoch": 0.18, "learning_rate": 4.4689823483533345e-06, "loss": 0.2748, "step": 1400 }, { "epoch": 0.18, "learning_rate": 4.464721026971889e-06, "loss": 0.2714, "step": 1410 }, { "epoch": 0.18, "learning_rate": 4.460420627771152e-06, "loss": 0.2849, "step": 1420 }, { "epoch": 0.19, "learning_rate": 4.456081229983805e-06, "loss": 0.2768, "step": 1430 }, { "epoch": 0.19, "learning_rate": 4.451702913561059e-06, "loss": 0.266, "step": 1440 }, { "epoch": 0.19, "learning_rate": 4.447285759171183e-06, "loss": 0.2758, "step": 1450 }, { "epoch": 0.19, "learning_rate": 4.442829848198013e-06, "loss": 0.2731, "step": 1460 }, { "epoch": 0.19, "learning_rate": 4.438335262739458e-06, "loss": 0.2435, "step": 1470 }, { "epoch": 0.19, "learning_rate": 4.433802085605981e-06, "loss": 0.2785, "step": 1480 }, { "epoch": 0.19, "learning_rate": 4.4292304003190824e-06, "loss": 0.2607, "step": 1490 }, { "epoch": 0.19, "learning_rate": 4.4246202911097526e-06, "loss": 0.2897, "step": 1500 }, { "epoch": 0.2, "learning_rate": 4.4199718429169235e-06, "loss": 0.2646, "step": 1510 }, { "epoch": 0.2, "learning_rate": 4.415285141385904e-06, "loss": 0.2753, "step": 1520 }, { "epoch": 0.2, "learning_rate": 4.410560272866801e-06, "loss": 0.2645, "step": 1530 }, { "epoch": 0.2, "learning_rate": 4.40579732441293e-06, "loss": 0.2605, "step": 1540 }, { "epoch": 0.2, "learning_rate": 4.400996383779208e-06, "loss": 0.2807, "step": 1550 }, { "epoch": 0.2, "learning_rate": 4.39615753942054e-06, "loss": 0.2784, "step": 1560 }, { "epoch": 0.2, "learning_rate": 4.39128088049019e-06, "loss": 0.2778, "step": 1570 }, { "epoch": 0.21, "learning_rate": 4.386366496838131e-06, "loss": 0.2675, "step": 1580 }, { "epoch": 0.21, "learning_rate": 4.381414479009401e-06, "loss": 0.2604, "step": 1590 }, { "epoch": 0.21, "learning_rate": 4.376424918242424e-06, "loss": 0.2737, "step": 1600 }, { "epoch": 0.21, "learning_rate": 4.371397906467338e-06, "loss": 0.2668, "step": 1610 }, { "epoch": 0.21, "learning_rate": 4.366333536304291e-06, "loss": 0.265, "step": 1620 }, { "epoch": 0.21, "learning_rate": 4.361231901061746e-06, "loss": 0.2777, "step": 1630 }, { "epoch": 0.21, "learning_rate": 4.356093094734755e-06, "loss": 0.2702, "step": 1640 }, { "epoch": 0.21, "learning_rate": 4.350917212003224e-06, "loss": 0.2735, "step": 1650 }, { "epoch": 0.22, "learning_rate": 4.34570434823018e-06, "loss": 0.2703, "step": 1660 }, { "epoch": 0.22, "learning_rate": 4.340454599460004e-06, "loss": 0.2686, "step": 1670 }, { "epoch": 0.22, "learning_rate": 4.335168062416661e-06, "loss": 0.2714, "step": 1680 }, { "epoch": 0.22, "learning_rate": 4.329844834501925e-06, "loss": 0.2799, "step": 1690 }, { "epoch": 0.22, "learning_rate": 4.324485013793582e-06, "loss": 0.2652, "step": 1700 }, { "epoch": 0.22, "learning_rate": 4.3190886990436184e-06, "loss": 0.259, "step": 1710 }, { "epoch": 0.22, "learning_rate": 4.313655989676407e-06, "loss": 0.2714, "step": 1720 }, { "epoch": 0.22, "learning_rate": 4.308186985786872e-06, "loss": 0.2619, "step": 1730 }, { "epoch": 0.23, "learning_rate": 4.3026817881386465e-06, "loss": 0.2629, "step": 1740 }, { "epoch": 0.23, "learning_rate": 4.297140498162214e-06, "loss": 0.2668, "step": 1750 }, { "epoch": 0.23, "learning_rate": 4.291563217953044e-06, "loss": 0.2698, "step": 1760 }, { "epoch": 0.23, "learning_rate": 4.285950050269703e-06, "loss": 0.272, "step": 1770 }, { "epoch": 0.23, "learning_rate": 4.2803010985319704e-06, "loss": 0.254, "step": 1780 }, { "epoch": 0.23, "learning_rate": 4.2746164668189275e-06, "loss": 0.2687, "step": 1790 }, { "epoch": 0.23, "learning_rate": 4.26889625986704e-06, "loss": 0.2757, "step": 1800 }, { "epoch": 0.23, "learning_rate": 4.2631405830682305e-06, "loss": 0.265, "step": 1810 }, { "epoch": 0.24, "learning_rate": 4.257349542467934e-06, "loss": 0.2615, "step": 1820 }, { "epoch": 0.24, "learning_rate": 4.2515232447631495e-06, "loss": 0.2758, "step": 1830 }, { "epoch": 0.24, "learning_rate": 4.2456617973004655e-06, "loss": 0.2491, "step": 1840 }, { "epoch": 0.24, "learning_rate": 4.23976530807409e-06, "loss": 0.2734, "step": 1850 }, { "epoch": 0.24, "learning_rate": 4.233833885723858e-06, "loss": 0.2618, "step": 1860 }, { "epoch": 0.24, "learning_rate": 4.227867639533227e-06, "loss": 0.2717, "step": 1870 }, { "epoch": 0.24, "learning_rate": 4.22186667942727e-06, "loss": 0.2487, "step": 1880 }, { "epoch": 0.25, "learning_rate": 4.215831115970642e-06, "loss": 0.2707, "step": 1890 }, { "epoch": 0.25, "learning_rate": 4.20976106036555e-06, "loss": 0.2583, "step": 1900 }, { "epoch": 0.25, "learning_rate": 4.203656624449703e-06, "loss": 0.2492, "step": 1910 }, { "epoch": 0.25, "learning_rate": 4.197517920694245e-06, "loss": 0.2519, "step": 1920 }, { "epoch": 0.25, "learning_rate": 4.1913450622016915e-06, "loss": 0.2744, "step": 1930 }, { "epoch": 0.25, "learning_rate": 4.185138162703841e-06, "loss": 0.2594, "step": 1940 }, { "epoch": 0.25, "learning_rate": 4.178897336559679e-06, "loss": 0.2559, "step": 1950 }, { "epoch": 0.25, "learning_rate": 4.1726226987532745e-06, "loss": 0.2725, "step": 1960 }, { "epoch": 0.26, "learning_rate": 4.166314364891657e-06, "loss": 0.2607, "step": 1970 }, { "epoch": 0.26, "learning_rate": 4.15997245120269e-06, "loss": 0.2624, "step": 1980 }, { "epoch": 0.26, "learning_rate": 4.153597074532928e-06, "loss": 0.2484, "step": 1990 }, { "epoch": 0.26, "learning_rate": 4.147188352345466e-06, "loss": 0.2567, "step": 2000 }, { "epoch": 0.26, "learning_rate": 4.1407464027177685e-06, "loss": 0.269, "step": 2010 }, { "epoch": 0.26, "learning_rate": 4.134271344339505e-06, "loss": 0.2597, "step": 2020 }, { "epoch": 0.26, "learning_rate": 4.127763296510351e-06, "loss": 0.265, "step": 2030 }, { "epoch": 0.26, "learning_rate": 4.121222379137801e-06, "loss": 0.2704, "step": 2040 }, { "epoch": 0.27, "learning_rate": 4.11464871273495e-06, "loss": 0.2509, "step": 2050 }, { "epoch": 0.27, "learning_rate": 4.108042418418282e-06, "loss": 0.2447, "step": 2060 }, { "epoch": 0.27, "learning_rate": 4.1014036179054275e-06, "loss": 0.2541, "step": 2070 }, { "epoch": 0.27, "learning_rate": 4.094732433512933e-06, "loss": 0.251, "step": 2080 }, { "epoch": 0.27, "learning_rate": 4.088028988154e-06, "loss": 0.2546, "step": 2090 }, { "epoch": 0.27, "learning_rate": 4.081293405336221e-06, "loss": 0.264, "step": 2100 }, { "epoch": 0.27, "learning_rate": 4.0745258091593045e-06, "loss": 0.2557, "step": 2110 }, { "epoch": 0.28, "learning_rate": 4.067726324312792e-06, "loss": 0.2577, "step": 2120 }, { "epoch": 0.28, "learning_rate": 4.060895076073753e-06, "loss": 0.2466, "step": 2130 }, { "epoch": 0.28, "learning_rate": 4.054032190304486e-06, "loss": 0.2501, "step": 2140 }, { "epoch": 0.28, "learning_rate": 4.047137793450191e-06, "loss": 0.2542, "step": 2150 }, { "epoch": 0.28, "learning_rate": 4.040212012536648e-06, "loss": 0.2525, "step": 2160 }, { "epoch": 0.28, "learning_rate": 4.033254975167869e-06, "loss": 0.2631, "step": 2170 }, { "epoch": 0.28, "learning_rate": 4.026266809523752e-06, "loss": 0.2559, "step": 2180 }, { "epoch": 0.28, "learning_rate": 4.0192476443577164e-06, "loss": 0.2494, "step": 2190 }, { "epoch": 0.29, "learning_rate": 4.012197608994335e-06, "loss": 0.2608, "step": 2200 }, { "epoch": 0.29, "learning_rate": 4.005116833326944e-06, "loss": 0.2649, "step": 2210 }, { "epoch": 0.29, "learning_rate": 3.9980054478152595e-06, "loss": 0.2531, "step": 2220 }, { "epoch": 0.29, "learning_rate": 3.990863583482963e-06, "loss": 0.2537, "step": 2230 }, { "epoch": 0.29, "learning_rate": 3.983691371915297e-06, "loss": 0.2651, "step": 2240 }, { "epoch": 0.29, "learning_rate": 3.976488945256635e-06, "loss": 0.2529, "step": 2250 }, { "epoch": 0.29, "learning_rate": 3.969256436208048e-06, "loss": 0.2774, "step": 2260 }, { "epoch": 0.29, "learning_rate": 3.961993978024859e-06, "loss": 0.2625, "step": 2270 }, { "epoch": 0.3, "learning_rate": 3.954701704514192e-06, "loss": 0.2573, "step": 2280 }, { "epoch": 0.3, "learning_rate": 3.9473797500325e-06, "loss": 0.2574, "step": 2290 }, { "epoch": 0.3, "learning_rate": 3.940028249483093e-06, "loss": 0.2556, "step": 2300 }, { "epoch": 0.3, "learning_rate": 3.932647338313655e-06, "loss": 0.2523, "step": 2310 }, { "epoch": 0.3, "learning_rate": 3.9252371525137435e-06, "loss": 0.2551, "step": 2320 }, { "epoch": 0.3, "learning_rate": 3.917797828612287e-06, "loss": 0.2768, "step": 2330 }, { "epoch": 0.3, "learning_rate": 3.910329503675069e-06, "loss": 0.2606, "step": 2340 }, { "epoch": 0.3, "learning_rate": 3.902832315302201e-06, "loss": 0.2547, "step": 2350 }, { "epoch": 0.31, "learning_rate": 3.895306401625592e-06, "loss": 0.2427, "step": 2360 }, { "epoch": 0.31, "learning_rate": 3.887751901306396e-06, "loss": 0.2469, "step": 2370 }, { "epoch": 0.31, "learning_rate": 3.8801689535324645e-06, "loss": 0.2551, "step": 2380 }, { "epoch": 0.31, "learning_rate": 3.872557698015779e-06, "loss": 0.2568, "step": 2390 }, { "epoch": 0.31, "learning_rate": 3.864918274989874e-06, "loss": 0.2474, "step": 2400 }, { "epoch": 0.31, "learning_rate": 3.857250825207259e-06, "loss": 0.2527, "step": 2410 }, { "epoch": 0.31, "learning_rate": 3.849555489936821e-06, "loss": 0.2523, "step": 2420 }, { "epoch": 0.32, "learning_rate": 3.841832410961225e-06, "loss": 0.2456, "step": 2430 }, { "epoch": 0.32, "learning_rate": 3.834081730574294e-06, "loss": 0.2627, "step": 2440 }, { "epoch": 0.32, "learning_rate": 3.826303591578402e-06, "loss": 0.2637, "step": 2450 }, { "epoch": 0.32, "learning_rate": 3.818498137281826e-06, "loss": 0.254, "step": 2460 }, { "epoch": 0.32, "learning_rate": 3.8106655114961175e-06, "loss": 0.2647, "step": 2470 }, { "epoch": 0.32, "learning_rate": 3.802805858533449e-06, "loss": 0.2613, "step": 2480 }, { "epoch": 0.32, "learning_rate": 3.7949193232039534e-06, "loss": 0.2436, "step": 2490 }, { "epoch": 0.32, "learning_rate": 3.7870060508130585e-06, "loss": 0.2554, "step": 2500 }, { "epoch": 0.33, "learning_rate": 3.7790661871588104e-06, "loss": 0.2663, "step": 2510 }, { "epoch": 0.33, "learning_rate": 3.7710998785291817e-06, "loss": 0.2506, "step": 2520 }, { "epoch": 0.33, "learning_rate": 3.7631072716993856e-06, "loss": 0.2605, "step": 2530 }, { "epoch": 0.33, "learning_rate": 3.7550885139291617e-06, "loss": 0.2492, "step": 2540 }, { "epoch": 0.33, "learning_rate": 3.7470437529600705e-06, "loss": 0.2715, "step": 2550 }, { "epoch": 0.33, "learning_rate": 3.738973137012767e-06, "loss": 0.2516, "step": 2560 }, { "epoch": 0.33, "learning_rate": 3.7308768147842717e-06, "loss": 0.2551, "step": 2570 }, { "epoch": 0.33, "learning_rate": 3.7227549354452296e-06, "loss": 0.2627, "step": 2580 }, { "epoch": 0.34, "learning_rate": 3.7146076486371625e-06, "loss": 0.2662, "step": 2590 }, { "epoch": 0.34, "learning_rate": 3.706435104469714e-06, "loss": 0.2686, "step": 2600 }, { "epoch": 0.34, "learning_rate": 3.6982374535178783e-06, "loss": 0.2518, "step": 2610 }, { "epoch": 0.34, "learning_rate": 3.6900148468192326e-06, "loss": 0.2499, "step": 2620 }, { "epoch": 0.34, "learning_rate": 3.6817674358711494e-06, "loss": 0.2721, "step": 2630 }, { "epoch": 0.34, "learning_rate": 3.6734953726280074e-06, "loss": 0.2497, "step": 2640 }, { "epoch": 0.34, "learning_rate": 3.665198809498391e-06, "loss": 0.2838, "step": 2650 }, { "epoch": 0.35, "learning_rate": 3.6568778993422835e-06, "loss": 0.2534, "step": 2660 }, { "epoch": 0.35, "learning_rate": 3.64853279546825e-06, "loss": 0.26, "step": 2670 }, { "epoch": 0.35, "learning_rate": 3.6401636516306112e-06, "loss": 0.2393, "step": 2680 }, { "epoch": 0.35, "learning_rate": 3.6317706220266126e-06, "loss": 0.2572, "step": 2690 }, { "epoch": 0.35, "learning_rate": 3.6233538612935835e-06, "loss": 0.2663, "step": 2700 }, { "epoch": 0.35, "learning_rate": 3.6149135245060863e-06, "loss": 0.2568, "step": 2710 }, { "epoch": 0.35, "learning_rate": 3.606449767173061e-06, "loss": 0.2632, "step": 2720 }, { "epoch": 0.35, "learning_rate": 3.5979627452349597e-06, "loss": 0.25, "step": 2730 }, { "epoch": 0.36, "learning_rate": 3.589452615060872e-06, "loss": 0.259, "step": 2740 }, { "epoch": 0.36, "learning_rate": 3.580919533445645e-06, "loss": 0.2613, "step": 2750 }, { "epoch": 0.36, "learning_rate": 3.5723636576069957e-06, "loss": 0.2566, "step": 2760 }, { "epoch": 0.36, "learning_rate": 3.5637851451826134e-06, "loss": 0.2526, "step": 2770 }, { "epoch": 0.36, "learning_rate": 3.5551841542272528e-06, "loss": 0.2494, "step": 2780 }, { "epoch": 0.36, "learning_rate": 3.5465608432098257e-06, "loss": 0.2591, "step": 2790 }, { "epoch": 0.36, "learning_rate": 3.5379153710104803e-06, "loss": 0.2481, "step": 2800 }, { "epoch": 0.36, "learning_rate": 3.5292478969176722e-06, "loss": 0.2358, "step": 2810 }, { "epoch": 0.37, "learning_rate": 3.5205585806252313e-06, "loss": 0.2473, "step": 2820 }, { "epoch": 0.37, "learning_rate": 3.5118475822294193e-06, "loss": 0.2682, "step": 2830 }, { "epoch": 0.37, "learning_rate": 3.5031150622259793e-06, "loss": 0.2587, "step": 2840 }, { "epoch": 0.37, "learning_rate": 3.4943611815071784e-06, "loss": 0.2593, "step": 2850 }, { "epoch": 0.37, "learning_rate": 3.4855861013588455e-06, "loss": 0.2606, "step": 2860 }, { "epoch": 0.37, "learning_rate": 3.4767899834573983e-06, "loss": 0.2562, "step": 2870 }, { "epoch": 0.37, "learning_rate": 3.4679729898668627e-06, "loss": 0.2386, "step": 2880 }, { "epoch": 0.38, "learning_rate": 3.4591352830358895e-06, "loss": 0.2657, "step": 2890 }, { "epoch": 0.38, "learning_rate": 3.450277025794762e-06, "loss": 0.2734, "step": 2900 }, { "epoch": 0.38, "learning_rate": 3.441398381352392e-06, "loss": 0.2563, "step": 2910 }, { "epoch": 0.38, "learning_rate": 3.4324995132933178e-06, "loss": 0.2384, "step": 2920 }, { "epoch": 0.38, "learning_rate": 3.423580585574684e-06, "loss": 0.2471, "step": 2930 }, { "epoch": 0.38, "learning_rate": 3.4146417625232287e-06, "loss": 0.2356, "step": 2940 }, { "epoch": 0.38, "learning_rate": 3.405683208832248e-06, "loss": 0.2389, "step": 2950 }, { "epoch": 0.38, "learning_rate": 3.3967050895585655e-06, "loss": 0.2518, "step": 2960 }, { "epoch": 0.39, "learning_rate": 3.387707570119493e-06, "loss": 0.2572, "step": 2970 }, { "epoch": 0.39, "learning_rate": 3.3786908162897785e-06, "loss": 0.2402, "step": 2980 }, { "epoch": 0.39, "learning_rate": 3.369654994198555e-06, "loss": 0.2277, "step": 2990 }, { "epoch": 0.39, "learning_rate": 3.3606002703262762e-06, "loss": 0.2557, "step": 3000 }, { "epoch": 0.39, "learning_rate": 3.3515268115016557e-06, "loss": 0.2608, "step": 3010 }, { "epoch": 0.39, "learning_rate": 3.3424347848985854e-06, "loss": 0.2436, "step": 3020 }, { "epoch": 0.39, "learning_rate": 3.3333243580330614e-06, "loss": 0.2605, "step": 3030 }, { "epoch": 0.39, "learning_rate": 3.3241956987600945e-06, "loss": 0.2567, "step": 3040 }, { "epoch": 0.4, "learning_rate": 3.315048975270618e-06, "loss": 0.2602, "step": 3050 }, { "epoch": 0.4, "learning_rate": 3.30588435608839e-06, "loss": 0.2576, "step": 3060 }, { "epoch": 0.4, "learning_rate": 3.2967020100668872e-06, "loss": 0.263, "step": 3070 }, { "epoch": 0.4, "learning_rate": 3.287502106386195e-06, "loss": 0.2484, "step": 3080 }, { "epoch": 0.4, "learning_rate": 3.2782848145498885e-06, "loss": 0.2772, "step": 3090 }, { "epoch": 0.4, "learning_rate": 3.2690503043819114e-06, "loss": 0.249, "step": 3100 }, { "epoch": 0.4, "learning_rate": 3.259798746023448e-06, "loss": 0.2561, "step": 3110 }, { "epoch": 0.4, "learning_rate": 3.2505303099297857e-06, "loss": 0.2531, "step": 3120 }, { "epoch": 0.41, "learning_rate": 3.241245166867174e-06, "loss": 0.2553, "step": 3130 }, { "epoch": 0.41, "learning_rate": 3.231943487909682e-06, "loss": 0.2543, "step": 3140 }, { "epoch": 0.41, "learning_rate": 3.2226254444360448e-06, "loss": 0.2654, "step": 3150 }, { "epoch": 0.41, "learning_rate": 3.213291208126502e-06, "loss": 0.2561, "step": 3160 }, { "epoch": 0.41, "learning_rate": 3.2039409509596418e-06, "loss": 0.2372, "step": 3170 }, { "epoch": 0.41, "learning_rate": 3.1945748452092277e-06, "loss": 0.2485, "step": 3180 }, { "epoch": 0.41, "learning_rate": 3.1851930634410233e-06, "loss": 0.2573, "step": 3190 }, { "epoch": 0.42, "learning_rate": 3.175795778509618e-06, "loss": 0.2615, "step": 3200 }, { "epoch": 0.42, "learning_rate": 3.166383163555236e-06, "loss": 0.2675, "step": 3210 }, { "epoch": 0.42, "learning_rate": 3.1569553920005527e-06, "loss": 0.2518, "step": 3220 }, { "epoch": 0.42, "learning_rate": 3.147512637547492e-06, "loss": 0.2627, "step": 3230 }, { "epoch": 0.42, "learning_rate": 3.1380550741740343e-06, "loss": 0.2515, "step": 3240 }, { "epoch": 0.42, "learning_rate": 3.1285828761310052e-06, "loss": 0.2485, "step": 3250 }, { "epoch": 0.42, "learning_rate": 3.1190962179388666e-06, "loss": 0.2463, "step": 3260 }, { "epoch": 0.42, "learning_rate": 3.109595274384501e-06, "loss": 0.2296, "step": 3270 }, { "epoch": 0.43, "learning_rate": 3.100080220517991e-06, "loss": 0.2504, "step": 3280 }, { "epoch": 0.43, "learning_rate": 3.0905512316493976e-06, "loss": 0.2451, "step": 3290 }, { "epoch": 0.43, "learning_rate": 3.0810084833455236e-06, "loss": 0.2522, "step": 3300 }, { "epoch": 0.43, "learning_rate": 3.0714521514266834e-06, "loss": 0.25, "step": 3310 }, { "epoch": 0.43, "learning_rate": 3.061882411963465e-06, "loss": 0.2476, "step": 3320 }, { "epoch": 0.43, "learning_rate": 3.05229944127348e-06, "loss": 0.2485, "step": 3330 }, { "epoch": 0.43, "learning_rate": 3.0427034159181205e-06, "loss": 0.2527, "step": 3340 }, { "epoch": 0.43, "learning_rate": 3.0330945126993043e-06, "loss": 0.2583, "step": 3350 }, { "epoch": 0.44, "learning_rate": 3.0234729086562184e-06, "loss": 0.2633, "step": 3360 }, { "epoch": 0.44, "learning_rate": 3.013838781062053e-06, "loss": 0.2532, "step": 3370 }, { "epoch": 0.44, "learning_rate": 3.004192307420741e-06, "loss": 0.2604, "step": 3380 }, { "epoch": 0.44, "learning_rate": 2.994533665463684e-06, "loss": 0.2601, "step": 3390 }, { "epoch": 0.44, "learning_rate": 2.9848630331464796e-06, "loss": 0.2446, "step": 3400 }, { "epoch": 0.44, "learning_rate": 2.9751805886456396e-06, "loss": 0.2511, "step": 3410 }, { "epoch": 0.44, "learning_rate": 2.9654865103553115e-06, "loss": 0.2419, "step": 3420 }, { "epoch": 0.45, "learning_rate": 2.95578097688399e-06, "loss": 0.2673, "step": 3430 }, { "epoch": 0.45, "learning_rate": 2.9460641670512215e-06, "loss": 0.245, "step": 3440 }, { "epoch": 0.45, "learning_rate": 2.936336259884318e-06, "loss": 0.2472, "step": 3450 }, { "epoch": 0.45, "learning_rate": 2.9265974346150538e-06, "loss": 0.2497, "step": 3460 }, { "epoch": 0.45, "learning_rate": 2.9168478706763613e-06, "loss": 0.2569, "step": 3470 }, { "epoch": 0.45, "learning_rate": 2.90708774769903e-06, "loss": 0.2444, "step": 3480 }, { "epoch": 0.45, "learning_rate": 2.8973172455083925e-06, "loss": 0.2546, "step": 3490 }, { "epoch": 0.45, "learning_rate": 2.8875365441210138e-06, "loss": 0.2419, "step": 3500 }, { "epoch": 0.46, "learning_rate": 2.8777458237413744e-06, "loss": 0.2416, "step": 3510 }, { "epoch": 0.46, "learning_rate": 2.8679452647585497e-06, "loss": 0.2543, "step": 3520 }, { "epoch": 0.46, "learning_rate": 2.858135047742886e-06, "loss": 0.2432, "step": 3530 }, { "epoch": 0.46, "learning_rate": 2.848315353442673e-06, "loss": 0.2471, "step": 3540 }, { "epoch": 0.46, "learning_rate": 2.838486362780817e-06, "loss": 0.2635, "step": 3550 }, { "epoch": 0.46, "learning_rate": 2.828648256851502e-06, "loss": 0.2373, "step": 3560 }, { "epoch": 0.46, "learning_rate": 2.818801216916859e-06, "loss": 0.261, "step": 3570 }, { "epoch": 0.46, "learning_rate": 2.808945424403622e-06, "loss": 0.2635, "step": 3580 }, { "epoch": 0.47, "learning_rate": 2.799081060899788e-06, "loss": 0.262, "step": 3590 }, { "epoch": 0.47, "learning_rate": 2.789208308151268e-06, "loss": 0.2522, "step": 3600 }, { "epoch": 0.47, "learning_rate": 2.779327348058543e-06, "loss": 0.2457, "step": 3610 }, { "epoch": 0.47, "learning_rate": 2.7694383626733088e-06, "loss": 0.2658, "step": 3620 }, { "epoch": 0.47, "learning_rate": 2.759541534195124e-06, "loss": 0.249, "step": 3630 }, { "epoch": 0.47, "learning_rate": 2.7496370449680507e-06, "loss": 0.2471, "step": 3640 }, { "epoch": 0.47, "learning_rate": 2.7397250774772976e-06, "loss": 0.2596, "step": 3650 }, { "epoch": 0.48, "learning_rate": 2.7298058143458573e-06, "loss": 0.2492, "step": 3660 }, { "epoch": 0.48, "learning_rate": 2.7198794383311405e-06, "loss": 0.2701, "step": 3670 }, { "epoch": 0.48, "learning_rate": 2.7099461323216063e-06, "loss": 0.2509, "step": 3680 }, { "epoch": 0.48, "learning_rate": 2.7000060793334003e-06, "loss": 0.2562, "step": 3690 }, { "epoch": 0.48, "learning_rate": 2.6900594625069746e-06, "loss": 0.2466, "step": 3700 }, { "epoch": 0.48, "learning_rate": 2.6801064651037175e-06, "loss": 0.2495, "step": 3710 }, { "epoch": 0.48, "learning_rate": 2.670147270502576e-06, "loss": 0.2592, "step": 3720 }, { "epoch": 0.48, "learning_rate": 2.6601820621966784e-06, "loss": 0.2578, "step": 3730 }, { "epoch": 0.49, "learning_rate": 2.6502110237899505e-06, "loss": 0.2639, "step": 3740 }, { "epoch": 0.49, "learning_rate": 2.6402343389937352e-06, "loss": 0.2275, "step": 3750 }, { "epoch": 0.49, "learning_rate": 2.63025219162341e-06, "loss": 0.267, "step": 3760 }, { "epoch": 0.49, "learning_rate": 2.6202647655949946e-06, "loss": 0.2408, "step": 3770 }, { "epoch": 0.49, "learning_rate": 2.6102722449217657e-06, "loss": 0.2517, "step": 3780 }, { "epoch": 0.49, "learning_rate": 2.6002748137108672e-06, "loss": 0.2432, "step": 3790 }, { "epoch": 0.49, "learning_rate": 2.5902726561599163e-06, "loss": 0.2687, "step": 3800 }, { "epoch": 0.49, "learning_rate": 2.580265956553612e-06, "loss": 0.2357, "step": 3810 }, { "epoch": 0.5, "learning_rate": 2.5702548992603357e-06, "loss": 0.2627, "step": 3820 }, { "epoch": 0.5, "learning_rate": 2.5602396687287613e-06, "loss": 0.2365, "step": 3830 }, { "epoch": 0.5, "learning_rate": 2.5502204494844483e-06, "loss": 0.259, "step": 3840 }, { "epoch": 0.5, "learning_rate": 2.5401974261264476e-06, "loss": 0.2433, "step": 3850 }, { "epoch": 0.5, "learning_rate": 2.530170783323899e-06, "loss": 0.2546, "step": 3860 }, { "epoch": 0.5, "learning_rate": 2.5201407058126307e-06, "loss": 0.2575, "step": 3870 }, { "epoch": 0.5, "learning_rate": 2.5101073783917504e-06, "loss": 0.2536, "step": 3880 }, { "epoch": 0.5, "learning_rate": 2.5000709859202455e-06, "loss": 0.2444, "step": 3890 }, { "epoch": 0.51, "learning_rate": 2.4900317133135754e-06, "loss": 0.2552, "step": 3900 }, { "epoch": 0.51, "learning_rate": 2.479989745540265e-06, "loss": 0.247, "step": 3910 }, { "epoch": 0.51, "learning_rate": 2.469945267618494e-06, "loss": 0.2461, "step": 3920 }, { "epoch": 0.51, "learning_rate": 2.4598984646126927e-06, "loss": 0.2549, "step": 3930 }, { "epoch": 0.51, "learning_rate": 2.449849521630129e-06, "loss": 0.2479, "step": 3940 }, { "epoch": 0.51, "learning_rate": 2.4397986238175e-06, "loss": 0.2516, "step": 3950 }, { "epoch": 0.51, "learning_rate": 2.4297459563575166e-06, "loss": 0.2527, "step": 3960 }, { "epoch": 0.52, "learning_rate": 2.4196917044654974e-06, "loss": 0.2488, "step": 3970 }, { "epoch": 0.52, "learning_rate": 2.4096360533859527e-06, "loss": 0.2447, "step": 3980 }, { "epoch": 0.52, "learning_rate": 2.399579188389171e-06, "loss": 0.2557, "step": 3990 }, { "epoch": 0.52, "learning_rate": 2.3895212947678075e-06, "loss": 0.2438, "step": 4000 }, { "epoch": 0.52, "learning_rate": 2.3794625578334705e-06, "loss": 0.2504, "step": 4010 }, { "epoch": 0.52, "learning_rate": 2.369403162913303e-06, "loss": 0.2652, "step": 4020 }, { "epoch": 0.52, "learning_rate": 2.359343295346573e-06, "loss": 0.2488, "step": 4030 }, { "epoch": 0.52, "learning_rate": 2.3492831404812565e-06, "loss": 0.2553, "step": 4040 }, { "epoch": 0.53, "learning_rate": 2.3392228836706236e-06, "loss": 0.2425, "step": 4050 }, { "epoch": 0.53, "learning_rate": 2.3291627102698204e-06, "loss": 0.2576, "step": 4060 }, { "epoch": 0.53, "learning_rate": 2.3191028056324595e-06, "loss": 0.2519, "step": 4070 }, { "epoch": 0.53, "learning_rate": 2.309043355107198e-06, "loss": 0.2525, "step": 4080 }, { "epoch": 0.53, "learning_rate": 2.298984544034329e-06, "loss": 0.2633, "step": 4090 }, { "epoch": 0.53, "learning_rate": 2.2889265577423624e-06, "loss": 0.2406, "step": 4100 }, { "epoch": 0.53, "learning_rate": 2.2788695815446142e-06, "loss": 0.2519, "step": 4110 }, { "epoch": 0.53, "learning_rate": 2.2688138007357873e-06, "loss": 0.2453, "step": 4120 }, { "epoch": 0.54, "learning_rate": 2.2587594005885604e-06, "loss": 0.2409, "step": 4130 }, { "epoch": 0.54, "learning_rate": 2.2487065663501758e-06, "loss": 0.2523, "step": 4140 }, { "epoch": 0.54, "learning_rate": 2.238655483239024e-06, "loss": 0.2676, "step": 4150 }, { "epoch": 0.54, "learning_rate": 2.2286063364412324e-06, "loss": 0.2393, "step": 4160 }, { "epoch": 0.54, "learning_rate": 2.2185593111072506e-06, "loss": 0.2501, "step": 4170 }, { "epoch": 0.54, "learning_rate": 2.208514592348443e-06, "loss": 0.2475, "step": 4180 }, { "epoch": 0.54, "learning_rate": 2.1984723652336784e-06, "loss": 0.2686, "step": 4190 }, { "epoch": 0.55, "learning_rate": 2.1884328147859143e-06, "loss": 0.2545, "step": 4200 }, { "epoch": 0.55, "learning_rate": 2.1783961259787957e-06, "loss": 0.2721, "step": 4210 }, { "epoch": 0.55, "learning_rate": 2.1683624837332402e-06, "loss": 0.2429, "step": 4220 }, { "epoch": 0.55, "learning_rate": 2.1583320729140354e-06, "loss": 0.2484, "step": 4230 }, { "epoch": 0.55, "learning_rate": 2.148305078326432e-06, "loss": 0.2528, "step": 4240 }, { "epoch": 0.55, "learning_rate": 2.1382816847127392e-06, "loss": 0.2615, "step": 4250 }, { "epoch": 0.55, "learning_rate": 2.1282620767489166e-06, "loss": 0.256, "step": 4260 }, { "epoch": 0.55, "learning_rate": 2.118246439041178e-06, "loss": 0.2387, "step": 4270 }, { "epoch": 0.56, "learning_rate": 2.1082349561225866e-06, "loss": 0.2553, "step": 4280 }, { "epoch": 0.56, "learning_rate": 2.0982278124496565e-06, "loss": 0.2643, "step": 4290 }, { "epoch": 0.56, "learning_rate": 2.088225192398952e-06, "loss": 0.2431, "step": 4300 }, { "epoch": 0.56, "learning_rate": 2.0782272802636917e-06, "loss": 0.2514, "step": 4310 }, { "epoch": 0.56, "learning_rate": 2.068234260250355e-06, "loss": 0.2439, "step": 4320 }, { "epoch": 0.56, "learning_rate": 2.0582463164752854e-06, "loss": 0.2377, "step": 4330 }, { "epoch": 0.56, "learning_rate": 2.048263632961298e-06, "loss": 0.2535, "step": 4340 }, { "epoch": 0.56, "learning_rate": 2.038286393634294e-06, "loss": 0.2328, "step": 4350 }, { "epoch": 0.57, "learning_rate": 2.028314782319864e-06, "loss": 0.2431, "step": 4360 }, { "epoch": 0.57, "learning_rate": 2.018348982739907e-06, "loss": 0.2439, "step": 4370 }, { "epoch": 0.57, "learning_rate": 2.008389178509244e-06, "loss": 0.2658, "step": 4380 }, { "epoch": 0.57, "learning_rate": 1.998435553132236e-06, "loss": 0.2689, "step": 4390 }, { "epoch": 0.57, "learning_rate": 1.988488289999399e-06, "loss": 0.2602, "step": 4400 }, { "epoch": 0.57, "learning_rate": 1.978547572384028e-06, "loss": 0.2605, "step": 4410 }, { "epoch": 0.57, "learning_rate": 1.968613583438823e-06, "loss": 0.2474, "step": 4420 }, { "epoch": 0.57, "learning_rate": 1.9586865061925098e-06, "loss": 0.2359, "step": 4430 }, { "epoch": 0.58, "learning_rate": 1.9487665235464706e-06, "loss": 0.2455, "step": 4440 }, { "epoch": 0.58, "learning_rate": 1.9388538182713696e-06, "loss": 0.2444, "step": 4450 }, { "epoch": 0.58, "learning_rate": 1.9289485730037945e-06, "loss": 0.2629, "step": 4460 }, { "epoch": 0.58, "learning_rate": 1.919050970242882e-06, "loss": 0.2463, "step": 4470 }, { "epoch": 0.58, "learning_rate": 1.9091611923469618e-06, "loss": 0.2594, "step": 4480 }, { "epoch": 0.58, "learning_rate": 1.899279421530194e-06, "loss": 0.2432, "step": 4490 }, { "epoch": 0.58, "learning_rate": 1.889405839859211e-06, "loss": 0.2421, "step": 4500 }, { "epoch": 0.59, "learning_rate": 1.8795406292497649e-06, "loss": 0.2402, "step": 4510 }, { "epoch": 0.59, "learning_rate": 1.8696839714633764e-06, "loss": 0.2437, "step": 4520 }, { "epoch": 0.59, "learning_rate": 1.8598360481039848e-06, "loss": 0.2442, "step": 4530 }, { "epoch": 0.59, "learning_rate": 1.8499970406145992e-06, "loss": 0.236, "step": 4540 }, { "epoch": 0.59, "learning_rate": 1.8401671302739611e-06, "loss": 0.2423, "step": 4550 }, { "epoch": 0.59, "learning_rate": 1.8303464981932012e-06, "loss": 0.2385, "step": 4560 }, { "epoch": 0.59, "learning_rate": 1.8205353253125017e-06, "loss": 0.2415, "step": 4570 }, { "epoch": 0.59, "learning_rate": 1.8107337923977659e-06, "loss": 0.2364, "step": 4580 }, { "epoch": 0.6, "learning_rate": 1.8009420800372823e-06, "loss": 0.2454, "step": 4590 }, { "epoch": 0.6, "learning_rate": 1.7911603686384046e-06, "loss": 0.2401, "step": 4600 }, { "epoch": 0.6, "learning_rate": 1.7813888384242203e-06, "loss": 0.2437, "step": 4610 }, { "epoch": 0.6, "learning_rate": 1.7716276694302366e-06, "loss": 0.2507, "step": 4620 }, { "epoch": 0.6, "learning_rate": 1.7618770415010592e-06, "loss": 0.2596, "step": 4630 }, { "epoch": 0.6, "learning_rate": 1.7521371342870795e-06, "loss": 0.2611, "step": 4640 }, { "epoch": 0.6, "learning_rate": 1.7424081272411653e-06, "loss": 0.2585, "step": 4650 }, { "epoch": 0.6, "learning_rate": 1.732690199615355e-06, "loss": 0.2426, "step": 4660 }, { "epoch": 0.61, "learning_rate": 1.7229835304575547e-06, "loss": 0.2385, "step": 4670 }, { "epoch": 0.61, "learning_rate": 1.7132882986082373e-06, "loss": 0.2518, "step": 4680 }, { "epoch": 0.61, "learning_rate": 1.7036046826971493e-06, "loss": 0.2417, "step": 4690 }, { "epoch": 0.61, "learning_rate": 1.6939328611400214e-06, "loss": 0.2514, "step": 4700 }, { "epoch": 0.61, "learning_rate": 1.684273012135277e-06, "loss": 0.2449, "step": 4710 }, { "epoch": 0.61, "learning_rate": 1.674625313660753e-06, "loss": 0.2237, "step": 4720 }, { "epoch": 0.61, "learning_rate": 1.664989943470417e-06, "loss": 0.2553, "step": 4730 }, { "epoch": 0.62, "learning_rate": 1.6553670790910962e-06, "loss": 0.2477, "step": 4740 }, { "epoch": 0.62, "learning_rate": 1.645756897819203e-06, "loss": 0.2566, "step": 4750 }, { "epoch": 0.62, "learning_rate": 1.6361595767174719e-06, "loss": 0.2626, "step": 4760 }, { "epoch": 0.62, "learning_rate": 1.6265752926116924e-06, "loss": 0.2409, "step": 4770 }, { "epoch": 0.62, "learning_rate": 1.6170042220874564e-06, "loss": 0.2459, "step": 4780 }, { "epoch": 0.62, "learning_rate": 1.6074465414869008e-06, "loss": 0.2427, "step": 4790 }, { "epoch": 0.62, "learning_rate": 1.597902426905461e-06, "loss": 0.2439, "step": 4800 }, { "epoch": 0.62, "learning_rate": 1.5883720541886253e-06, "loss": 0.2359, "step": 4810 }, { "epoch": 0.63, "learning_rate": 1.5788555989286928e-06, "loss": 0.2444, "step": 4820 }, { "epoch": 0.63, "learning_rate": 1.5693532364615437e-06, "loss": 0.2547, "step": 4830 }, { "epoch": 0.63, "learning_rate": 1.5598651418634044e-06, "loss": 0.2604, "step": 4840 }, { "epoch": 0.63, "learning_rate": 1.5503914899476223e-06, "loss": 0.2745, "step": 4850 }, { "epoch": 0.63, "learning_rate": 1.540932455261447e-06, "loss": 0.2343, "step": 4860 }, { "epoch": 0.63, "learning_rate": 1.5314882120828105e-06, "loss": 0.2498, "step": 4870 }, { "epoch": 0.63, "learning_rate": 1.5220589344171221e-06, "loss": 0.2359, "step": 4880 }, { "epoch": 0.63, "learning_rate": 1.5126447959940568e-06, "loss": 0.2496, "step": 4890 }, { "epoch": 0.64, "learning_rate": 1.5032459702643585e-06, "loss": 0.2543, "step": 4900 }, { "epoch": 0.64, "learning_rate": 1.4938626303966397e-06, "loss": 0.2659, "step": 4910 }, { "epoch": 0.64, "learning_rate": 1.4844949492741964e-06, "loss": 0.2469, "step": 4920 }, { "epoch": 0.64, "learning_rate": 1.4751430994918185e-06, "loss": 0.2415, "step": 4930 }, { "epoch": 0.64, "learning_rate": 1.4658072533526132e-06, "loss": 0.2514, "step": 4940 }, { "epoch": 0.64, "learning_rate": 1.4564875828648266e-06, "loss": 0.2305, "step": 4950 }, { "epoch": 0.64, "learning_rate": 1.4471842597386774e-06, "loss": 0.2433, "step": 4960 }, { "epoch": 0.65, "learning_rate": 1.4378974553831926e-06, "loss": 0.2362, "step": 4970 }, { "epoch": 0.65, "learning_rate": 1.42862734090305e-06, "loss": 0.2437, "step": 4980 }, { "epoch": 0.65, "learning_rate": 1.4193740870954227e-06, "loss": 0.2528, "step": 4990 }, { "epoch": 0.65, "learning_rate": 1.410137864446838e-06, "loss": 0.2347, "step": 5000 }, { "epoch": 0.65, "learning_rate": 1.4009188431300279e-06, "loss": 0.2401, "step": 5010 }, { "epoch": 0.65, "learning_rate": 1.3917171930008023e-06, "loss": 0.2487, "step": 5020 }, { "epoch": 0.65, "learning_rate": 1.3825330835949142e-06, "loss": 0.234, "step": 5030 }, { "epoch": 0.65, "learning_rate": 1.3733666841249394e-06, "loss": 0.2506, "step": 5040 }, { "epoch": 0.66, "learning_rate": 1.3642181634771537e-06, "loss": 0.2361, "step": 5050 }, { "epoch": 0.66, "learning_rate": 1.3550876902084279e-06, "loss": 0.2686, "step": 5060 }, { "epoch": 0.66, "learning_rate": 1.3459754325431172e-06, "loss": 0.2428, "step": 5070 }, { "epoch": 0.66, "learning_rate": 1.3368815583699631e-06, "loss": 0.2452, "step": 5080 }, { "epoch": 0.66, "learning_rate": 1.3278062352390044e-06, "loss": 0.2573, "step": 5090 }, { "epoch": 0.66, "learning_rate": 1.3187496303584809e-06, "loss": 0.2381, "step": 5100 }, { "epoch": 0.66, "learning_rate": 1.3097119105917598e-06, "loss": 0.2545, "step": 5110 }, { "epoch": 0.66, "learning_rate": 1.3006932424542622e-06, "loss": 0.2555, "step": 5120 }, { "epoch": 0.67, "learning_rate": 1.2916937921103908e-06, "loss": 0.2443, "step": 5130 }, { "epoch": 0.67, "learning_rate": 1.2827137253704695e-06, "loss": 0.2493, "step": 5140 }, { "epoch": 0.67, "learning_rate": 1.2737532076876895e-06, "loss": 0.2656, "step": 5150 }, { "epoch": 0.67, "learning_rate": 1.26481240415506e-06, "loss": 0.2538, "step": 5160 }, { "epoch": 0.67, "learning_rate": 1.2558914795023695e-06, "loss": 0.2493, "step": 5170 }, { "epoch": 0.67, "learning_rate": 1.2469905980931466e-06, "loss": 0.2346, "step": 5180 }, { "epoch": 0.67, "learning_rate": 1.2381099239216298e-06, "loss": 0.2405, "step": 5190 }, { "epoch": 0.67, "learning_rate": 1.2292496206097555e-06, "loss": 0.2573, "step": 5200 }, { "epoch": 0.68, "learning_rate": 1.2204098514041339e-06, "loss": 0.2459, "step": 5210 }, { "epoch": 0.68, "learning_rate": 1.211590779173044e-06, "loss": 0.2469, "step": 5220 }, { "epoch": 0.68, "learning_rate": 1.2027925664034376e-06, "loss": 0.2354, "step": 5230 }, { "epoch": 0.68, "learning_rate": 1.1940153751979358e-06, "loss": 0.2448, "step": 5240 }, { "epoch": 0.68, "learning_rate": 1.1852593672718508e-06, "loss": 0.2493, "step": 5250 }, { "epoch": 0.68, "learning_rate": 1.1765247039502054e-06, "loss": 0.2483, "step": 5260 }, { "epoch": 0.68, "learning_rate": 1.1678115461647562e-06, "loss": 0.2513, "step": 5270 }, { "epoch": 0.69, "learning_rate": 1.159120054451032e-06, "loss": 0.2603, "step": 5280 }, { "epoch": 0.69, "learning_rate": 1.1504503889453754e-06, "loss": 0.24, "step": 5290 }, { "epoch": 0.69, "learning_rate": 1.1418027093819907e-06, "loss": 0.2543, "step": 5300 }, { "epoch": 0.69, "learning_rate": 1.1331771750900054e-06, "loss": 0.2311, "step": 5310 }, { "epoch": 0.69, "learning_rate": 1.12457394499053e-06, "loss": 0.2514, "step": 5320 }, { "epoch": 0.69, "learning_rate": 1.1159931775937283e-06, "loss": 0.2437, "step": 5330 }, { "epoch": 0.69, "learning_rate": 1.1074350309959058e-06, "loss": 0.2554, "step": 5340 }, { "epoch": 0.69, "learning_rate": 1.0988996628765874e-06, "loss": 0.2421, "step": 5350 }, { "epoch": 0.7, "learning_rate": 1.0903872304956168e-06, "loss": 0.2521, "step": 5360 }, { "epoch": 0.7, "learning_rate": 1.0818978906902608e-06, "loss": 0.2741, "step": 5370 }, { "epoch": 0.7, "learning_rate": 1.073431799872313e-06, "loss": 0.248, "step": 5380 }, { "epoch": 0.7, "learning_rate": 1.0649891140252177e-06, "loss": 0.2535, "step": 5390 }, { "epoch": 0.7, "learning_rate": 1.0565699887011967e-06, "loss": 0.2426, "step": 5400 }, { "epoch": 0.7, "learning_rate": 1.0481745790183785e-06, "loss": 0.2419, "step": 5410 }, { "epoch": 0.7, "learning_rate": 1.039803039657944e-06, "loss": 0.2487, "step": 5420 }, { "epoch": 0.7, "learning_rate": 1.0314555248612746e-06, "loss": 0.2451, "step": 5430 }, { "epoch": 0.71, "learning_rate": 1.0231321884271116e-06, "loss": 0.2322, "step": 5440 }, { "epoch": 0.71, "learning_rate": 1.0148331837087237e-06, "loss": 0.2418, "step": 5450 }, { "epoch": 0.71, "learning_rate": 1.0065586636110787e-06, "loss": 0.2564, "step": 5460 }, { "epoch": 0.71, "learning_rate": 9.983087805880254e-07, "loss": 0.2492, "step": 5470 }, { "epoch": 0.71, "learning_rate": 9.900836866394918e-07, "loss": 0.2378, "step": 5480 }, { "epoch": 0.71, "learning_rate": 9.818835333086761e-07, "loss": 0.2453, "step": 5490 }, { "epoch": 0.71, "learning_rate": 9.737084716792582e-07, "loss": 0.2557, "step": 5500 }, { "epoch": 0.72, "learning_rate": 9.655586523726204e-07, "loss": 0.2429, "step": 5510 }, { "epoch": 0.72, "learning_rate": 9.574342255450618e-07, "loss": 0.2512, "step": 5520 }, { "epoch": 0.72, "learning_rate": 9.493353408850409e-07, "loss": 0.2825, "step": 5530 }, { "epoch": 0.72, "learning_rate": 9.412621476104161e-07, "loss": 0.2563, "step": 5540 }, { "epoch": 0.72, "learning_rate": 9.332147944656929e-07, "loss": 0.2355, "step": 5550 }, { "epoch": 0.72, "learning_rate": 9.251934297192861e-07, "loss": 0.2505, "step": 5560 }, { "epoch": 0.72, "learning_rate": 9.171982011607873e-07, "loss": 0.2461, "step": 5570 }, { "epoch": 0.72, "learning_rate": 9.092292560982419e-07, "loss": 0.238, "step": 5580 }, { "epoch": 0.73, "learning_rate": 9.01286741355435e-07, "loss": 0.2563, "step": 5590 }, { "epoch": 0.73, "learning_rate": 8.933708032691889e-07, "loss": 0.2606, "step": 5600 }, { "epoch": 0.73, "learning_rate": 8.854815876866593e-07, "loss": 0.2593, "step": 5610 }, { "epoch": 0.73, "learning_rate": 8.776192399626588e-07, "loss": 0.2392, "step": 5620 }, { "epoch": 0.73, "learning_rate": 8.697839049569696e-07, "loss": 0.2286, "step": 5630 }, { "epoch": 0.73, "learning_rate": 8.61975727031679e-07, "loss": 0.2415, "step": 5640 }, { "epoch": 0.73, "learning_rate": 8.541948500485217e-07, "loss": 0.2535, "step": 5650 }, { "epoch": 0.73, "learning_rate": 8.464414173662211e-07, "loss": 0.2472, "step": 5660 }, { "epoch": 0.74, "learning_rate": 8.387155718378564e-07, "loss": 0.232, "step": 5670 }, { "epoch": 0.74, "learning_rate": 8.310174558082282e-07, "loss": 0.2416, "step": 5680 }, { "epoch": 0.74, "learning_rate": 8.233472111112335e-07, "loss": 0.2484, "step": 5690 }, { "epoch": 0.74, "learning_rate": 8.157049790672542e-07, "loss": 0.2325, "step": 5700 }, { "epoch": 0.74, "learning_rate": 8.08090900480554e-07, "loss": 0.2474, "step": 5710 }, { "epoch": 0.74, "learning_rate": 8.005051156366824e-07, "loss": 0.2523, "step": 5720 }, { "epoch": 0.74, "learning_rate": 7.929477642998909e-07, "loss": 0.2592, "step": 5730 }, { "epoch": 0.74, "learning_rate": 7.854189857105611e-07, "loss": 0.2408, "step": 5740 }, { "epoch": 0.75, "learning_rate": 7.779189185826306e-07, "loss": 0.2369, "step": 5750 }, { "epoch": 0.75, "learning_rate": 7.704477011010477e-07, "loss": 0.2438, "step": 5760 }, { "epoch": 0.75, "learning_rate": 7.630054709192186e-07, "loss": 0.2377, "step": 5770 }, { "epoch": 0.75, "learning_rate": 7.555923651564721e-07, "loss": 0.247, "step": 5780 }, { "epoch": 0.75, "learning_rate": 7.482085203955382e-07, "loss": 0.2587, "step": 5790 }, { "epoch": 0.75, "learning_rate": 7.408540726800224e-07, "loss": 0.2483, "step": 5800 }, { "epoch": 0.75, "learning_rate": 7.335291575119075e-07, "loss": 0.2491, "step": 5810 }, { "epoch": 0.76, "learning_rate": 7.262339098490546e-07, "loss": 0.2293, "step": 5820 }, { "epoch": 0.76, "learning_rate": 7.189684641027155e-07, "loss": 0.2452, "step": 5830 }, { "epoch": 0.76, "learning_rate": 7.11732954135054e-07, "loss": 0.2414, "step": 5840 }, { "epoch": 0.76, "learning_rate": 7.045275132566867e-07, "loss": 0.2427, "step": 5850 }, { "epoch": 0.76, "learning_rate": 6.973522742242199e-07, "loss": 0.2431, "step": 5860 }, { "epoch": 0.76, "learning_rate": 6.902073692378057e-07, "loss": 0.2396, "step": 5870 }, { "epoch": 0.76, "learning_rate": 6.830929299387097e-07, "loss": 0.2464, "step": 5880 }, { "epoch": 0.76, "learning_rate": 6.760090874068779e-07, "loss": 0.2605, "step": 5890 }, { "epoch": 0.77, "learning_rate": 6.68955972158531e-07, "loss": 0.2641, "step": 5900 }, { "epoch": 0.77, "learning_rate": 6.619337141437523e-07, "loss": 0.2596, "step": 5910 }, { "epoch": 0.77, "learning_rate": 6.549424427440968e-07, "loss": 0.2519, "step": 5920 }, { "epoch": 0.77, "learning_rate": 6.479822867702067e-07, "loss": 0.2418, "step": 5930 }, { "epoch": 0.77, "learning_rate": 6.41053374459439e-07, "loss": 0.2345, "step": 5940 }, { "epoch": 0.77, "learning_rate": 6.341558334735e-07, "loss": 0.2601, "step": 5950 }, { "epoch": 0.77, "learning_rate": 6.272897908960978e-07, "loss": 0.2479, "step": 5960 }, { "epoch": 0.77, "learning_rate": 6.204553732305978e-07, "loss": 0.2512, "step": 5970 }, { "epoch": 0.78, "learning_rate": 6.136527063976885e-07, "loss": 0.2478, "step": 5980 }, { "epoch": 0.78, "learning_rate": 6.068819157330717e-07, "loss": 0.2542, "step": 5990 }, { "epoch": 0.78, "learning_rate": 6.001431259851415e-07, "loss": 0.2541, "step": 6000 }, { "epoch": 0.78, "learning_rate": 5.934364613126937e-07, "loss": 0.2587, "step": 6010 }, { "epoch": 0.78, "learning_rate": 5.867620452826375e-07, "loss": 0.2406, "step": 6020 }, { "epoch": 0.78, "learning_rate": 5.801200008677121e-07, "loss": 0.2495, "step": 6030 }, { "epoch": 0.78, "learning_rate": 5.735104504442313e-07, "loss": 0.2582, "step": 6040 }, { "epoch": 0.79, "learning_rate": 5.669335157898206e-07, "loss": 0.2451, "step": 6050 }, { "epoch": 0.79, "learning_rate": 5.603893180811778e-07, "loss": 0.247, "step": 6060 }, { "epoch": 0.79, "learning_rate": 5.538779778918377e-07, "loss": 0.248, "step": 6070 }, { "epoch": 0.79, "learning_rate": 5.473996151899528e-07, "loss": 0.2462, "step": 6080 }, { "epoch": 0.79, "learning_rate": 5.409543493360816e-07, "loss": 0.2563, "step": 6090 }, { "epoch": 0.79, "learning_rate": 5.345422990809909e-07, "loss": 0.2565, "step": 6100 }, { "epoch": 0.79, "learning_rate": 5.281635825634669e-07, "loss": 0.2544, "step": 6110 }, { "epoch": 0.79, "learning_rate": 5.218183173081353e-07, "loss": 0.2319, "step": 6120 }, { "epoch": 0.8, "learning_rate": 5.155066202233038e-07, "loss": 0.265, "step": 6130 }, { "epoch": 0.8, "learning_rate": 5.092286075988009e-07, "loss": 0.2586, "step": 6140 }, { "epoch": 0.8, "learning_rate": 5.029843951038359e-07, "loss": 0.2451, "step": 6150 }, { "epoch": 0.8, "learning_rate": 4.9677409778487e-07, "loss": 0.2472, "step": 6160 }, { "epoch": 0.8, "learning_rate": 4.905978300634908e-07, "loss": 0.2452, "step": 6170 }, { "epoch": 0.8, "learning_rate": 4.844557057343104e-07, "loss": 0.2435, "step": 6180 }, { "epoch": 0.8, "learning_rate": 4.783478379628659e-07, "loss": 0.254, "step": 6190 }, { "epoch": 0.8, "learning_rate": 4.722743392835343e-07, "loss": 0.2449, "step": 6200 }, { "epoch": 0.81, "learning_rate": 4.66235321597459e-07, "loss": 0.2534, "step": 6210 }, { "epoch": 0.81, "learning_rate": 4.6023089617048937e-07, "loss": 0.2435, "step": 6220 }, { "epoch": 0.81, "learning_rate": 4.542611736311283e-07, "loss": 0.2325, "step": 6230 }, { "epoch": 0.81, "learning_rate": 4.483262639684977e-07, "loss": 0.2426, "step": 6240 }, { "epoch": 0.81, "learning_rate": 4.4242627653030966e-07, "loss": 0.2465, "step": 6250 }, { "epoch": 0.81, "learning_rate": 4.365613200208484e-07, "loss": 0.2594, "step": 6260 }, { "epoch": 0.81, "learning_rate": 4.307315024989758e-07, "loss": 0.2271, "step": 6270 }, { "epoch": 0.82, "learning_rate": 4.2493693137613284e-07, "loss": 0.2434, "step": 6280 }, { "epoch": 0.82, "learning_rate": 4.191777134143632e-07, "loss": 0.2372, "step": 6290 }, { "epoch": 0.82, "learning_rate": 4.1345395472434947e-07, "loss": 0.2681, "step": 6300 }, { "epoch": 0.82, "learning_rate": 4.077657607634508e-07, "loss": 0.2346, "step": 6310 }, { "epoch": 0.82, "learning_rate": 4.0211323633376616e-07, "loss": 0.2482, "step": 6320 }, { "epoch": 0.82, "learning_rate": 3.96496485580203e-07, "loss": 0.2418, "step": 6330 }, { "epoch": 0.82, "learning_rate": 3.9091561198855486e-07, "loss": 0.2519, "step": 6340 }, { "epoch": 0.82, "learning_rate": 3.85370718383597e-07, "loss": 0.2428, "step": 6350 }, { "epoch": 0.83, "learning_rate": 3.79861906927192e-07, "loss": 0.2482, "step": 6360 }, { "epoch": 0.83, "learning_rate": 3.743892791164063e-07, "loss": 0.2635, "step": 6370 }, { "epoch": 0.83, "learning_rate": 3.6895293578164284e-07, "loss": 0.2432, "step": 6380 }, { "epoch": 0.83, "learning_rate": 3.6355297708477984e-07, "loss": 0.2393, "step": 6390 }, { "epoch": 0.83, "learning_rate": 3.5818950251732617e-07, "loss": 0.2665, "step": 6400 }, { "epoch": 0.83, "learning_rate": 3.5286261089859146e-07, "loss": 0.2331, "step": 6410 }, { "epoch": 0.83, "learning_rate": 3.4757240037386145e-07, "loss": 0.2406, "step": 6420 }, { "epoch": 0.83, "learning_rate": 3.4231896841259104e-07, "loss": 0.2496, "step": 6430 }, { "epoch": 0.84, "learning_rate": 3.3710241180660966e-07, "loss": 0.2435, "step": 6440 }, { "epoch": 0.84, "learning_rate": 3.319228266683365e-07, "loss": 0.2447, "step": 6450 }, { "epoch": 0.84, "learning_rate": 3.267803084290091e-07, "loss": 0.2429, "step": 6460 }, { "epoch": 0.84, "learning_rate": 3.216749518369286e-07, "loss": 0.2487, "step": 6470 }, { "epoch": 0.84, "learning_rate": 3.1660685095570927e-07, "loss": 0.2623, "step": 6480 }, { "epoch": 0.84, "learning_rate": 3.1157609916254893e-07, "loss": 0.2611, "step": 6490 }, { "epoch": 0.84, "learning_rate": 3.0658278914650694e-07, "loss": 0.2578, "step": 6500 }, { "epoch": 0.84, "learning_rate": 3.016270129067964e-07, "loss": 0.2481, "step": 6510 }, { "epoch": 0.85, "learning_rate": 2.9670886175109e-07, "loss": 0.2508, "step": 6520 }, { "epoch": 0.85, "learning_rate": 2.9182842629383864e-07, "loss": 0.2571, "step": 6530 }, { "epoch": 0.85, "learning_rate": 2.869857964545971e-07, "loss": 0.2465, "step": 6540 }, { "epoch": 0.85, "learning_rate": 2.8218106145637425e-07, "loss": 0.2457, "step": 6550 }, { "epoch": 0.85, "learning_rate": 2.77414309823984e-07, "loss": 0.256, "step": 6560 }, { "epoch": 0.85, "learning_rate": 2.7268562938241607e-07, "loss": 0.2428, "step": 6570 }, { "epoch": 0.85, "learning_rate": 2.6799510725521807e-07, "loss": 0.2444, "step": 6580 }, { "epoch": 0.86, "learning_rate": 2.633428298628892e-07, "loss": 0.2619, "step": 6590 }, { "epoch": 0.86, "learning_rate": 2.587288829212894e-07, "loss": 0.2463, "step": 6600 }, { "epoch": 0.86, "learning_rate": 2.541533514400597e-07, "loss": 0.2459, "step": 6610 }, { "epoch": 0.86, "learning_rate": 2.4961631972105535e-07, "loss": 0.2446, "step": 6620 }, { "epoch": 0.86, "learning_rate": 2.451178713567926e-07, "loss": 0.2428, "step": 6630 }, { "epoch": 0.86, "learning_rate": 2.4065808922890927e-07, "loss": 0.2445, "step": 6640 }, { "epoch": 0.86, "learning_rate": 2.3623705550663684e-07, "loss": 0.2479, "step": 6650 }, { "epoch": 0.86, "learning_rate": 2.3185485164528743e-07, "loss": 0.2485, "step": 6660 }, { "epoch": 0.87, "learning_rate": 2.275115583847539e-07, "loss": 0.249, "step": 6670 }, { "epoch": 0.87, "learning_rate": 2.2320725574801746e-07, "loss": 0.2477, "step": 6680 }, { "epoch": 0.87, "learning_rate": 2.189420230396803e-07, "loss": 0.2315, "step": 6690 }, { "epoch": 0.87, "learning_rate": 2.1471593884449914e-07, "loss": 0.2475, "step": 6700 }, { "epoch": 0.87, "learning_rate": 2.1052908102593985e-07, "loss": 0.248, "step": 6710 }, { "epoch": 0.87, "learning_rate": 2.0638152672474132e-07, "loss": 0.2527, "step": 6720 }, { "epoch": 0.87, "learning_rate": 2.0227335235749564e-07, "loss": 0.254, "step": 6730 }, { "epoch": 0.87, "learning_rate": 1.9820463361523924e-07, "loss": 0.2481, "step": 6740 }, { "epoch": 0.88, "learning_rate": 1.9417544546205923e-07, "loss": 0.255, "step": 6750 }, { "epoch": 0.88, "learning_rate": 1.9018586213371086e-07, "loss": 0.2353, "step": 6760 }, { "epoch": 0.88, "learning_rate": 1.8623595713625057e-07, "loss": 0.2518, "step": 6770 }, { "epoch": 0.88, "learning_rate": 1.823258032446819e-07, "loss": 0.263, "step": 6780 }, { "epoch": 0.88, "learning_rate": 1.7845547250161414e-07, "loss": 0.253, "step": 6790 }, { "epoch": 0.88, "learning_rate": 1.7462503621593505e-07, "loss": 0.2388, "step": 6800 }, { "epoch": 0.88, "learning_rate": 1.708345649614984e-07, "loss": 0.251, "step": 6810 }, { "epoch": 0.89, "learning_rate": 1.6708412857581967e-07, "loss": 0.2466, "step": 6820 }, { "epoch": 0.89, "learning_rate": 1.6337379615879572e-07, "loss": 0.2594, "step": 6830 }, { "epoch": 0.89, "learning_rate": 1.5970363607142522e-07, "loss": 0.2303, "step": 6840 }, { "epoch": 0.89, "learning_rate": 1.5607371593455302e-07, "loss": 0.252, "step": 6850 }, { "epoch": 0.89, "learning_rate": 1.5248410262762378e-07, "loss": 0.2456, "step": 6860 }, { "epoch": 0.89, "learning_rate": 1.489348622874479e-07, "loss": 0.2492, "step": 6870 }, { "epoch": 0.89, "learning_rate": 1.4542606030698488e-07, "loss": 0.2412, "step": 6880 }, { "epoch": 0.89, "learning_rate": 1.419577613341387e-07, "loss": 0.2471, "step": 6890 }, { "epoch": 0.9, "learning_rate": 1.385300292705652e-07, "loss": 0.2401, "step": 6900 }, { "epoch": 0.9, "learning_rate": 1.351429272704944e-07, "loss": 0.2416, "step": 6910 }, { "epoch": 0.9, "learning_rate": 1.3179651773956942e-07, "loss": 0.2571, "step": 6920 }, { "epoch": 0.9, "learning_rate": 1.2849086233369491e-07, "loss": 0.2513, "step": 6930 }, { "epoch": 0.9, "learning_rate": 1.2522602195790083e-07, "loss": 0.2482, "step": 6940 }, { "epoch": 0.9, "learning_rate": 1.220020567652213e-07, "loss": 0.2376, "step": 6950 }, { "epoch": 0.9, "learning_rate": 1.1881902615558554e-07, "loss": 0.2629, "step": 6960 }, { "epoch": 0.9, "learning_rate": 1.1567698877472404e-07, "loss": 0.2364, "step": 6970 }, { "epoch": 0.91, "learning_rate": 1.1257600251308756e-07, "loss": 0.2412, "step": 6980 }, { "epoch": 0.91, "learning_rate": 1.0951612450478108e-07, "loss": 0.2451, "step": 6990 }, { "epoch": 0.91, "learning_rate": 1.0649741112651027e-07, "loss": 0.2562, "step": 7000 }, { "epoch": 0.91, "learning_rate": 1.035199179965437e-07, "loss": 0.2238, "step": 7010 }, { "epoch": 0.91, "learning_rate": 1.0058369997368691e-07, "loss": 0.2628, "step": 7020 }, { "epoch": 0.91, "learning_rate": 9.768881115627403e-08, "loss": 0.2454, "step": 7030 }, { "epoch": 0.91, "learning_rate": 9.483530488116837e-08, "loss": 0.2583, "step": 7040 }, { "epoch": 0.91, "learning_rate": 9.202323372278011e-08, "loss": 0.2415, "step": 7050 }, { "epoch": 0.92, "learning_rate": 8.92526494921002e-08, "loss": 0.2389, "step": 7060 }, { "epoch": 0.92, "learning_rate": 8.652360323574267e-08, "loss": 0.2548, "step": 7070 }, { "epoch": 0.92, "learning_rate": 8.383614523500516e-08, "loss": 0.2441, "step": 7080 }, { "epoch": 0.92, "learning_rate": 8.119032500494352e-08, "loss": 0.25, "step": 7090 }, { "epoch": 0.92, "learning_rate": 7.8586191293458e-08, "loss": 0.2306, "step": 7100 }, { "epoch": 0.92, "learning_rate": 7.602379208039681e-08, "loss": 0.2353, "step": 7110 }, { "epoch": 0.92, "learning_rate": 7.350317457667014e-08, "loss": 0.2449, "step": 7120 }, { "epoch": 0.93, "learning_rate": 7.102438522338184e-08, "loss": 0.2367, "step": 7130 }, { "epoch": 0.93, "learning_rate": 6.858746969097272e-08, "loss": 0.2295, "step": 7140 }, { "epoch": 0.93, "learning_rate": 6.619247287838001e-08, "loss": 0.2471, "step": 7150 }, { "epoch": 0.93, "learning_rate": 6.383943891220943e-08, "loss": 0.2594, "step": 7160 }, { "epoch": 0.93, "learning_rate": 6.152841114592296e-08, "loss": 0.2356, "step": 7170 }, { "epoch": 0.93, "learning_rate": 5.9259432159039674e-08, "loss": 0.2438, "step": 7180 }, { "epoch": 0.93, "learning_rate": 5.703254375634967e-08, "loss": 0.2487, "step": 7190 }, { "epoch": 0.93, "learning_rate": 5.484778696714723e-08, "loss": 0.2522, "step": 7200 }, { "epoch": 0.94, "learning_rate": 5.270520204447154e-08, "loss": 0.2515, "step": 7210 }, { "epoch": 0.94, "learning_rate": 5.060482846436743e-08, "loss": 0.2317, "step": 7220 }, { "epoch": 0.94, "learning_rate": 4.854670492515627e-08, "loss": 0.2327, "step": 7230 }, { "epoch": 0.94, "learning_rate": 4.65308693467243e-08, "loss": 0.2473, "step": 7240 }, { "epoch": 0.94, "learning_rate": 4.4557358869823455e-08, "loss": 0.2475, "step": 7250 }, { "epoch": 0.94, "learning_rate": 4.262620985538675e-08, "loss": 0.2386, "step": 7260 }, { "epoch": 0.94, "learning_rate": 4.0737457883858756e-08, "loss": 0.2392, "step": 7270 }, { "epoch": 0.94, "learning_rate": 3.889113775454016e-08, "loss": 0.2411, "step": 7280 }, { "epoch": 0.95, "learning_rate": 3.7087283484946334e-08, "loss": 0.2557, "step": 7290 }, { "epoch": 0.95, "learning_rate": 3.532592831018023e-08, "loss": 0.2487, "step": 7300 }, { "epoch": 0.95, "learning_rate": 3.36071046823214e-08, "loss": 0.2457, "step": 7310 }, { "epoch": 0.95, "learning_rate": 3.193084426982625e-08, "loss": 0.2512, "step": 7320 }, { "epoch": 0.95, "learning_rate": 3.029717795694515e-08, "loss": 0.2586, "step": 7330 }, { "epoch": 0.95, "learning_rate": 2.870613584315517e-08, "loss": 0.2376, "step": 7340 }, { "epoch": 0.95, "learning_rate": 2.7157747242602997e-08, "loss": 0.2464, "step": 7350 }, { "epoch": 0.96, "learning_rate": 2.565204068356577e-08, "loss": 0.2504, "step": 7360 }, { "epoch": 0.96, "learning_rate": 2.4189043907925973e-08, "loss": 0.2613, "step": 7370 }, { "epoch": 0.96, "learning_rate": 2.276878387066014e-08, "loss": 0.2372, "step": 7380 }, { "epoch": 0.96, "learning_rate": 2.139128673934133e-08, "loss": 0.2438, "step": 7390 }, { "epoch": 0.96, "learning_rate": 2.0056577893658253e-08, "loss": 0.2358, "step": 7400 }, { "epoch": 0.96, "learning_rate": 1.876468192494691e-08, "loss": 0.2298, "step": 7410 }, { "epoch": 0.96, "learning_rate": 1.7515622635737304e-08, "loss": 0.245, "step": 7420 }, { "epoch": 0.96, "learning_rate": 1.6309423039316017e-08, "loss": 0.2404, "step": 7430 }, { "epoch": 0.97, "learning_rate": 1.5146105359300794e-08, "loss": 0.2226, "step": 7440 }, { "epoch": 0.97, "learning_rate": 1.4025691029232247e-08, "loss": 0.2644, "step": 7450 }, { "epoch": 0.97, "learning_rate": 1.2948200692177845e-08, "loss": 0.2508, "step": 7460 }, { "epoch": 0.97, "learning_rate": 1.191365420035225e-08, "loss": 0.2522, "step": 7470 }, { "epoch": 0.97, "learning_rate": 1.0922070614752514e-08, "loss": 0.2363, "step": 7480 }, { "epoch": 0.97, "learning_rate": 9.973468204804713e-09, "loss": 0.2651, "step": 7490 }, { "epoch": 0.97, "learning_rate": 9.0678644480288e-09, "loss": 0.2611, "step": 7500 }, { "epoch": 0.97, "learning_rate": 8.205276029716462e-09, "loss": 0.2615, "step": 7510 }, { "epoch": 0.98, "learning_rate": 7.385718842623043e-09, "loss": 0.254, "step": 7520 }, { "epoch": 0.98, "learning_rate": 6.60920798667584e-09, "loss": 0.2486, "step": 7530 }, { "epoch": 0.98, "learning_rate": 5.875757768694904e-09, "loss": 0.2527, "step": 7540 }, { "epoch": 0.98, "learning_rate": 5.185381702129971e-09, "loss": 0.2544, "step": 7550 }, { "epoch": 0.98, "learning_rate": 4.538092506810916e-09, "loss": 0.2395, "step": 7560 }, { "epoch": 0.98, "learning_rate": 3.933902108714092e-09, "loss": 0.233, "step": 7570 }, { "epoch": 0.98, "learning_rate": 3.372821639742191e-09, "loss": 0.2373, "step": 7580 }, { "epoch": 0.99, "learning_rate": 2.8548614375194582e-09, "loss": 0.2417, "step": 7590 }, { "epoch": 0.99, "learning_rate": 2.380031045200962e-09, "loss": 0.235, "step": 7600 }, { "epoch": 0.99, "learning_rate": 1.948339211296432e-09, "loss": 0.2374, "step": 7610 }, { "epoch": 0.99, "learning_rate": 1.5597938895097092e-09, "loss": 0.2549, "step": 7620 }, { "epoch": 0.99, "learning_rate": 1.2144022385922498e-09, "loss": 0.2466, "step": 7630 }, { "epoch": 0.99, "learning_rate": 9.121706222106791e-10, "loss": 0.2675, "step": 7640 }, { "epoch": 0.99, "learning_rate": 6.531046088294363e-10, "loss": 0.2547, "step": 7650 }, { "epoch": 0.99, "learning_rate": 4.372089716092944e-10, "loss": 0.2441, "step": 7660 }, { "epoch": 1.0, "learning_rate": 2.644876883181081e-10, "loss": 0.2326, "step": 7670 }, { "epoch": 1.0, "learning_rate": 1.3494394125821567e-10, "loss": 0.2472, "step": 7680 }, { "epoch": 1.0, "learning_rate": 4.858011720763194e-11, "loss": 0.2293, "step": 7690 }, { "epoch": 1.0, "learning_rate": 5.397807376072982e-12, "loss": 0.2611, "step": 7700 }, { "epoch": 1.0, "eval_loss": 0.24734805524349213, "eval_runtime": 55.3916, "eval_samples_per_second": 196.365, "eval_steps_per_second": 24.552, "step": 7705 }, { "epoch": 1.0, "step": 7705, "total_flos": 6.229344692427817e+18, "train_loss": 0.284148887006746, "train_runtime": 946.9107, "train_samples_per_second": 65.088, "train_steps_per_second": 8.137 } ], "max_steps": 7705, "num_train_epochs": 1, "total_flos": 6.229344692427817e+18, "trial_name": null, "trial_params": null }