| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.6859592711682744, | |
| "eval_steps": 500, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00019931271477663232, | |
| "loss": 2.5587, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0001986254295532646, | |
| "loss": 2.3914, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00019793814432989693, | |
| "loss": 2.4218, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00019725085910652924, | |
| "loss": 2.3414, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001965635738831615, | |
| "loss": 2.2469, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00019587628865979381, | |
| "loss": 2.3241, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00019518900343642613, | |
| "loss": 2.3266, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019450171821305842, | |
| "loss": 2.1856, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019381443298969073, | |
| "loss": 2.3247, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019312714776632305, | |
| "loss": 2.3245, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019243986254295533, | |
| "loss": 2.2591, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019175257731958765, | |
| "loss": 2.1767, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019106529209621996, | |
| "loss": 2.3478, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019037800687285222, | |
| "loss": 2.3339, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00018969072164948454, | |
| "loss": 2.234, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00018900343642611685, | |
| "loss": 2.2651, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00018831615120274914, | |
| "loss": 2.1831, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00018762886597938145, | |
| "loss": 2.216, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00018694158075601377, | |
| "loss": 2.1359, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00018625429553264605, | |
| "loss": 2.1215, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00018556701030927837, | |
| "loss": 2.2179, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00018487972508591068, | |
| "loss": 2.2598, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00018419243986254294, | |
| "loss": 2.1813, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00018350515463917526, | |
| "loss": 2.2006, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00018281786941580757, | |
| "loss": 2.1564, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00018213058419243986, | |
| "loss": 2.2537, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00018144329896907217, | |
| "loss": 2.1975, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0001807560137457045, | |
| "loss": 2.2566, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00018006872852233677, | |
| "loss": 2.1464, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0001793814432989691, | |
| "loss": 2.1421, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0001786941580756014, | |
| "loss": 2.1276, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00017800687285223366, | |
| "loss": 2.0649, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00017731958762886598, | |
| "loss": 2.1835, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0001766323024054983, | |
| "loss": 2.1711, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00017594501718213058, | |
| "loss": 2.2591, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0001752577319587629, | |
| "loss": 2.1471, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0001745704467353952, | |
| "loss": 2.0861, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0001738831615120275, | |
| "loss": 2.0702, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0001731958762886598, | |
| "loss": 2.1096, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00017250859106529212, | |
| "loss": 2.1062, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00017182130584192438, | |
| "loss": 2.2545, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0001711340206185567, | |
| "loss": 2.1572, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.000170446735395189, | |
| "loss": 2.0749, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0001697594501718213, | |
| "loss": 2.1922, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00016907216494845361, | |
| "loss": 2.1915, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00016838487972508593, | |
| "loss": 2.1594, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00016769759450171822, | |
| "loss": 2.176, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00016701030927835053, | |
| "loss": 2.1223, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00016632302405498285, | |
| "loss": 2.1263, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00016563573883161513, | |
| "loss": 2.0481, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00016494845360824742, | |
| "loss": 2.1043, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00016426116838487973, | |
| "loss": 2.1678, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00016357388316151202, | |
| "loss": 2.1602, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00016288659793814434, | |
| "loss": 2.1448, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00016219931271477665, | |
| "loss": 2.1536, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00016151202749140894, | |
| "loss": 2.0339, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00016082474226804125, | |
| "loss": 2.023, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00016013745704467357, | |
| "loss": 2.1407, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00015945017182130585, | |
| "loss": 2.1134, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00015876288659793814, | |
| "loss": 2.1652, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00015807560137457046, | |
| "loss": 2.0051, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00015738831615120274, | |
| "loss": 2.0604, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00015670103092783506, | |
| "loss": 2.1708, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00015601374570446737, | |
| "loss": 2.1106, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00015532646048109966, | |
| "loss": 2.1445, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00015463917525773197, | |
| "loss": 2.0879, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0001539518900343643, | |
| "loss": 2.1498, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00015326460481099657, | |
| "loss": 2.0719, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00015257731958762886, | |
| "loss": 2.2167, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00015189003436426118, | |
| "loss": 2.0811, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00015120274914089346, | |
| "loss": 2.1058, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00015051546391752578, | |
| "loss": 2.0392, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0001498281786941581, | |
| "loss": 2.0957, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00014914089347079038, | |
| "loss": 1.9994, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0001484536082474227, | |
| "loss": 2.0464, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000147766323024055, | |
| "loss": 2.0417, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0001470790378006873, | |
| "loss": 2.105, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00014639175257731958, | |
| "loss": 2.1147, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0001457044673539519, | |
| "loss": 1.9964, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00014501718213058418, | |
| "loss": 1.9723, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0001443298969072165, | |
| "loss": 2.0621, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00014364261168384881, | |
| "loss": 2.2703, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0001429553264604811, | |
| "loss": 2.0815, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00014226804123711342, | |
| "loss": 2.0774, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00014158075601374573, | |
| "loss": 2.066, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00014089347079037802, | |
| "loss": 2.03, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001402061855670103, | |
| "loss": 2.1433, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00013951890034364262, | |
| "loss": 2.0811, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0001388316151202749, | |
| "loss": 1.9791, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00013814432989690722, | |
| "loss": 2.0876, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00013745704467353953, | |
| "loss": 2.0314, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00013676975945017182, | |
| "loss": 1.9485, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00013608247422680414, | |
| "loss": 2.078, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00013539518900343645, | |
| "loss": 2.1251, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00013470790378006874, | |
| "loss": 1.9736, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00013402061855670103, | |
| "loss": 2.0189, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00013333333333333334, | |
| "loss": 2.0061, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00013264604810996563, | |
| "loss": 1.9595, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00013195876288659794, | |
| "loss": 1.9702, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00013127147766323026, | |
| "loss": 2.0322, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00013058419243986254, | |
| "loss": 2.0128, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00012989690721649486, | |
| "loss": 2.1276, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00012920962199312717, | |
| "loss": 2.0177, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00012852233676975946, | |
| "loss": 1.9366, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00012783505154639175, | |
| "loss": 2.0341, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00012714776632302406, | |
| "loss": 2.0251, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00012646048109965635, | |
| "loss": 1.9588, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00012577319587628866, | |
| "loss": 2.0445, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00012508591065292098, | |
| "loss": 1.8917, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00012439862542955326, | |
| "loss": 2.0385, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00012371134020618558, | |
| "loss": 2.0435, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0001230240549828179, | |
| "loss": 2.0666, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00012233676975945018, | |
| "loss": 1.9854, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00012164948453608247, | |
| "loss": 1.9233, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00012096219931271477, | |
| "loss": 1.985, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00012027491408934708, | |
| "loss": 2.0679, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00011958762886597938, | |
| "loss": 1.9717, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00011890034364261168, | |
| "loss": 1.9388, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.000118213058419244, | |
| "loss": 1.9607, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0001175257731958763, | |
| "loss": 1.9543, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0001168384879725086, | |
| "loss": 1.9925, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00011615120274914091, | |
| "loss": 1.9913, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00011546391752577319, | |
| "loss": 1.9857, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00011477663230240549, | |
| "loss": 2.0056, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0001140893470790378, | |
| "loss": 2.0133, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0001134020618556701, | |
| "loss": 1.8104, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0001127147766323024, | |
| "loss": 1.7946, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00011202749140893472, | |
| "loss": 1.8347, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00011134020618556702, | |
| "loss": 2.0342, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00011065292096219932, | |
| "loss": 1.9425, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00010996563573883164, | |
| "loss": 1.9546, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00010927835051546391, | |
| "loss": 1.9694, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00010859106529209621, | |
| "loss": 1.936, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00010790378006872852, | |
| "loss": 2.0004, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00010721649484536083, | |
| "loss": 2.0462, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00010652920962199313, | |
| "loss": 1.9783, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00010584192439862544, | |
| "loss": 1.8628, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00010515463917525774, | |
| "loss": 1.9134, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00010446735395189004, | |
| "loss": 1.8846, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00010378006872852236, | |
| "loss": 1.8119, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00010309278350515463, | |
| "loss": 1.9242, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00010240549828178693, | |
| "loss": 1.9292, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00010171821305841925, | |
| "loss": 1.9763, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00010103092783505155, | |
| "loss": 1.9023, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00010034364261168385, | |
| "loss": 1.8064, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.965635738831616e-05, | |
| "loss": 1.9646, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.896907216494846e-05, | |
| "loss": 1.8234, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.828178694158075e-05, | |
| "loss": 2.0289, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.759450171821306e-05, | |
| "loss": 1.7698, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.690721649484537e-05, | |
| "loss": 1.9497, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.621993127147767e-05, | |
| "loss": 1.9135, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.553264604810998e-05, | |
| "loss": 1.7955, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.484536082474227e-05, | |
| "loss": 1.9452, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.415807560137457e-05, | |
| "loss": 1.9715, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.347079037800688e-05, | |
| "loss": 1.8546, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.278350515463918e-05, | |
| "loss": 1.9756, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.209621993127147e-05, | |
| "loss": 2.0691, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.140893470790379e-05, | |
| "loss": 1.8373, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.072164948453609e-05, | |
| "loss": 1.7061, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.003436426116839e-05, | |
| "loss": 1.9069, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.93470790378007e-05, | |
| "loss": 1.8055, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.865979381443299e-05, | |
| "loss": 1.8873, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.797250859106529e-05, | |
| "loss": 1.8406, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.72852233676976e-05, | |
| "loss": 1.9428, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.65979381443299e-05, | |
| "loss": 1.831, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.591065292096219e-05, | |
| "loss": 1.8802, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.52233676975945e-05, | |
| "loss": 1.8436, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.453608247422681e-05, | |
| "loss": 1.848, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.384879725085911e-05, | |
| "loss": 1.9022, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.316151202749142e-05, | |
| "loss": 1.8015, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.247422680412371e-05, | |
| "loss": 1.8204, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.178694158075601e-05, | |
| "loss": 1.798, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.109965635738833e-05, | |
| "loss": 1.8832, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.041237113402063e-05, | |
| "loss": 1.8176, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.972508591065293e-05, | |
| "loss": 1.9251, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.903780068728523e-05, | |
| "loss": 1.7559, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.835051546391753e-05, | |
| "loss": 1.805, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.766323024054983e-05, | |
| "loss": 1.7995, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.697594501718214e-05, | |
| "loss": 1.9055, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.628865979381443e-05, | |
| "loss": 1.8654, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.560137457044673e-05, | |
| "loss": 1.852, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.491408934707905e-05, | |
| "loss": 1.9377, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.422680412371135e-05, | |
| "loss": 1.8024, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.353951890034365e-05, | |
| "loss": 1.7329, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.285223367697595e-05, | |
| "loss": 1.8203, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.216494845360825e-05, | |
| "loss": 1.8223, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.147766323024055e-05, | |
| "loss": 1.8506, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.079037800687286e-05, | |
| "loss": 1.8169, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.010309278350515e-05, | |
| "loss": 1.796, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 6.941580756013745e-05, | |
| "loss": 1.7975, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.872852233676977e-05, | |
| "loss": 1.8774, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.804123711340207e-05, | |
| "loss": 1.7822, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.735395189003437e-05, | |
| "loss": 1.8542, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 1.8148, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 6.597938144329897e-05, | |
| "loss": 1.7446, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 6.529209621993127e-05, | |
| "loss": 1.7477, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.460481099656359e-05, | |
| "loss": 1.8121, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.391752577319587e-05, | |
| "loss": 1.7289, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.323024054982817e-05, | |
| "loss": 1.8006, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 6.254295532646049e-05, | |
| "loss": 1.7348, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 291, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 1.2148491200262144e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |