| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.21733678843099558, |
| "eval_steps": 200, |
| "global_step": 2600, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001, |
| "loss": 5.6047, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0002, |
| "loss": 5.7127, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001999944269512637, |
| "loss": 4.9696, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001999888539025274, |
| "loss": 4.786, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019998328085379108, |
| "loss": 4.4821, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019997770780505475, |
| "loss": 3.913, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019997213475631847, |
| "loss": 3.9184, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019996656170758216, |
| "loss": 4.0557, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019996098865884582, |
| "loss": 4.0889, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019995541561010954, |
| "loss": 3.6161, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001999498425613732, |
| "loss": 3.5843, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001999442695126369, |
| "loss": 3.5603, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019993869646390059, |
| "loss": 3.5113, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019993312341516428, |
| "loss": 3.5638, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019992755036642797, |
| "loss": 3.5064, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019992197731769163, |
| "loss": 3.4272, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019991640426895535, |
| "loss": 3.1662, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019991083122021904, |
| "loss": 3.2655, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001999052581714827, |
| "loss": 3.3496, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019989968512274642, |
| "loss": 3.2338, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001998941120740101, |
| "loss": 3.18, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019988853902527378, |
| "loss": 3.1195, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001998829659765375, |
| "loss": 2.8658, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019987739292780116, |
| "loss": 2.9657, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019987181987906485, |
| "loss": 2.5891, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019986624683032854, |
| "loss": 2.8152, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019986067378159223, |
| "loss": 2.7372, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019985510073285592, |
| "loss": 2.5434, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001998495276841196, |
| "loss": 2.485, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001998439546353833, |
| "loss": 2.2765, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019983838158664697, |
| "loss": 2.1537, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019983280853791066, |
| "loss": 2.5342, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019982723548917438, |
| "loss": 2.089, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019982166244043804, |
| "loss": 2.6249, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019981608939170174, |
| "loss": 2.0061, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019981051634296543, |
| "loss": 1.7938, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019980494329422912, |
| "loss": 2.127, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001997993702454928, |
| "loss": 2.1229, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001997937971967565, |
| "loss": 2.1513, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001997882241480202, |
| "loss": 2.3028, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019978265109928386, |
| "loss": 2.0165, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019977707805054757, |
| "loss": 2.396, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019977150500181126, |
| "loss": 1.578, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019976593195307493, |
| "loss": 1.7637, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019976035890433862, |
| "loss": 2.3954, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001997547858556023, |
| "loss": 1.9072, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.000199749212806866, |
| "loss": 1.6484, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001997436397581297, |
| "loss": 2.2662, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019973806670939338, |
| "loss": 2.1409, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019973249366065708, |
| "loss": 1.5637, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019972692061192077, |
| "loss": 1.8624, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019972134756318446, |
| "loss": 1.7322, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019971577451444815, |
| "loss": 2.0485, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001997102014657118, |
| "loss": 1.7355, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019970462841697553, |
| "loss": 2.3112, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001996990553682392, |
| "loss": 2.0858, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001996934823195029, |
| "loss": 1.8583, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.0001996879092707666, |
| "loss": 1.8217, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 0.00019968233622203027, |
| "loss": 1.9657, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019967676317329396, |
| "loss": 1.8796, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019967119012455765, |
| "loss": 1.4888, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019966561707582134, |
| "loss": 1.4468, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019966004402708503, |
| "loss": 2.0091, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001996544709783487, |
| "loss": 1.9454, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019964889792961242, |
| "loss": 1.8409, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019964332488087608, |
| "loss": 1.8389, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019963775183213977, |
| "loss": 2.058, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001996321787834035, |
| "loss": 1.8841, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019962660573466715, |
| "loss": 1.83, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019962103268593084, |
| "loss": 1.9224, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019961545963719454, |
| "loss": 1.4684, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019960988658845823, |
| "loss": 1.9383, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019960431353972192, |
| "loss": 1.6136, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001995987404909856, |
| "loss": 1.9532, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001995931674422493, |
| "loss": 1.7999, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.000199587594393513, |
| "loss": 2.0945, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019958202134477666, |
| "loss": 1.8812, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019957644829604037, |
| "loss": 2.061, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019957087524730404, |
| "loss": 1.8466, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019956530219856773, |
| "loss": 2.1327, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019955972914983142, |
| "loss": 1.8487, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001995541561010951, |
| "loss": 1.5964, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001995485830523588, |
| "loss": 1.8794, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001995430100036225, |
| "loss": 1.7655, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019953743695488618, |
| "loss": 1.83, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019953186390614988, |
| "loss": 1.6116, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019952629085741357, |
| "loss": 1.5026, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019952071780867726, |
| "loss": 1.7605, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019951514475994092, |
| "loss": 2.1237, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019950957171120464, |
| "loss": 1.7356, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001995039986624683, |
| "loss": 1.3219, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.000199498425613732, |
| "loss": 1.7981, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019949285256499569, |
| "loss": 1.7495, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019948727951625938, |
| "loss": 1.5518, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019948170646752307, |
| "loss": 1.7605, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019947613341878676, |
| "loss": 1.9295, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019947056037005045, |
| "loss": 1.7838, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019946498732131414, |
| "loss": 1.5007, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001994594142725778, |
| "loss": 1.5487, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019945384122384152, |
| "loss": 1.5372, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001994482681751052, |
| "loss": 1.6703, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019944269512636888, |
| "loss": 1.7879, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001994371220776326, |
| "loss": 2.292, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019943154902889626, |
| "loss": 1.7342, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019942597598015995, |
| "loss": 1.9287, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019942040293142364, |
| "loss": 1.6518, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019941482988268734, |
| "loss": 1.7454, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019940925683395103, |
| "loss": 1.5082, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001994036837852147, |
| "loss": 1.6209, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001993981107364784, |
| "loss": 1.5589, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001993925376877421, |
| "loss": 1.6977, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019938696463900576, |
| "loss": 1.461, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019938139159026948, |
| "loss": 1.8184, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019937581854153315, |
| "loss": 1.839, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019937024549279684, |
| "loss": 1.519, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019936467244406053, |
| "loss": 1.6532, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019935909939532422, |
| "loss": 1.7062, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001993535263465879, |
| "loss": 2.1245, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001993479532978516, |
| "loss": 1.5458, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001993423802491153, |
| "loss": 1.6948, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019933680720037898, |
| "loss": 1.2724, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019933123415164268, |
| "loss": 1.6057, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019932566110290637, |
| "loss": 1.5584, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019932008805417003, |
| "loss": 1.4616, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019931451500543372, |
| "loss": 1.368, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001993089419566974, |
| "loss": 1.9614, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001993033689079611, |
| "loss": 1.5037, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001992977958592248, |
| "loss": 1.8847, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019929222281048849, |
| "loss": 1.8973, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019928664976175218, |
| "loss": 1.7571, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019928107671301587, |
| "loss": 1.8246, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019927550366427956, |
| "loss": 2.0371, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019926993061554325, |
| "loss": 1.592, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019926435756680691, |
| "loss": 1.4302, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019925878451807063, |
| "loss": 1.8449, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019925321146933432, |
| "loss": 1.7325, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.000199247638420598, |
| "loss": 1.5165, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001992420653718617, |
| "loss": 1.3814, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019923649232312537, |
| "loss": 1.8283, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019923091927438906, |
| "loss": 1.9724, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019922534622565275, |
| "loss": 1.7483, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019921977317691644, |
| "loss": 1.4138, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019921420012818013, |
| "loss": 1.4558, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001992086270794438, |
| "loss": 1.6143, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019920305403070752, |
| "loss": 1.5576, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001991974809819712, |
| "loss": 1.4894, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019919190793323487, |
| "loss": 1.8957, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001991863348844986, |
| "loss": 1.4005, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019918076183576225, |
| "loss": 1.6747, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019917518878702595, |
| "loss": 1.3135, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019916961573828964, |
| "loss": 1.8817, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019916404268955333, |
| "loss": 1.7336, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019915846964081702, |
| "loss": 1.8838, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001991528965920807, |
| "loss": 1.4749, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001991473235433444, |
| "loss": 1.2198, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001991417504946081, |
| "loss": 1.7477, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019913617744587176, |
| "loss": 1.6846, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019913060439713547, |
| "loss": 1.7421, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019912503134839914, |
| "loss": 1.7707, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019911945829966283, |
| "loss": 1.5612, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019911388525092655, |
| "loss": 1.6013, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001991083122021902, |
| "loss": 1.727, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001991027391534539, |
| "loss": 1.2887, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001990971661047176, |
| "loss": 1.4303, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019909159305598129, |
| "loss": 1.1922, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019908602000724498, |
| "loss": 1.6117, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019908044695850867, |
| "loss": 1.8894, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019907487390977236, |
| "loss": 1.9297, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019906930086103602, |
| "loss": 1.4803, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019906372781229974, |
| "loss": 1.8682, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019905815476356343, |
| "loss": 1.4978, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001990525817148271, |
| "loss": 1.8577, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001990470086660908, |
| "loss": 1.3308, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019904143561735448, |
| "loss": 1.5809, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019903586256861817, |
| "loss": 1.6966, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019903028951988186, |
| "loss": 1.647, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019902471647114555, |
| "loss": 1.4956, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.00019901914342240924, |
| "loss": 1.5894, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 0.0001990135703736729, |
| "loss": 1.6021, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019900799732493663, |
| "loss": 1.3499, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019900242427620032, |
| "loss": 1.3679, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019899685122746398, |
| "loss": 1.909, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001989912781787277, |
| "loss": 1.5095, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019898570512999136, |
| "loss": 1.7182, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019898013208125505, |
| "loss": 1.5059, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019897455903251877, |
| "loss": 1.6163, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019896898598378244, |
| "loss": 1.7454, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019896341293504613, |
| "loss": 1.7269, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001989578398863098, |
| "loss": 1.5514, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001989522668375735, |
| "loss": 1.3184, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001989466937888372, |
| "loss": 1.8007, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019894112074010087, |
| "loss": 1.9123, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019893554769136458, |
| "loss": 1.4551, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019892997464262825, |
| "loss": 1.5712, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019892440159389194, |
| "loss": 1.6837, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019891882854515566, |
| "loss": 1.471, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019891325549641932, |
| "loss": 1.5618, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.000198907682447683, |
| "loss": 1.9567, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001989021093989467, |
| "loss": 1.2549, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001988965363502104, |
| "loss": 1.8665, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.02, |
| "eval_loss": 1.6084641218185425, |
| "eval_runtime": 332.2056, |
| "eval_samples_per_second": 7.525, |
| "eval_steps_per_second": 0.942, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019889096330147409, |
| "loss": 1.3196, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019888539025273778, |
| "loss": 1.6354, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019887981720400147, |
| "loss": 1.5879, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019887424415526513, |
| "loss": 1.5668, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019886867110652882, |
| "loss": 1.5922, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019886309805779254, |
| "loss": 1.2698, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001988575250090562, |
| "loss": 1.7446, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001988519519603199, |
| "loss": 1.6848, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001988463789115836, |
| "loss": 1.6244, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019884080586284728, |
| "loss": 1.4094, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019883523281411097, |
| "loss": 1.423, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019882965976537466, |
| "loss": 1.7363, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019882408671663835, |
| "loss": 1.6854, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019881851366790202, |
| "loss": 1.6603, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019881294061916573, |
| "loss": 1.079, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019880736757042943, |
| "loss": 1.8023, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001988017945216931, |
| "loss": 1.8508, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001987962214729568, |
| "loss": 1.4372, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019879064842422047, |
| "loss": 1.4804, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019878507537548416, |
| "loss": 1.3178, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019877950232674785, |
| "loss": 1.6053, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019877392927801155, |
| "loss": 1.6545, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019876835622927524, |
| "loss": 1.7882, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019876278318053893, |
| "loss": 1.5764, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019875721013180262, |
| "loss": 1.1906, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001987516370830663, |
| "loss": 1.6198, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019874606403432997, |
| "loss": 1.2939, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001987404909855937, |
| "loss": 1.2911, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019873491793685736, |
| "loss": 1.7041, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019872934488812105, |
| "loss": 1.5262, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019872377183938477, |
| "loss": 1.4176, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019871819879064843, |
| "loss": 1.5251, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019871262574191212, |
| "loss": 1.5523, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001987070526931758, |
| "loss": 1.136, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001987014796444395, |
| "loss": 1.4139, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001986959065957032, |
| "loss": 1.3022, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019869033354696686, |
| "loss": 1.7039, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019868476049823058, |
| "loss": 1.4306, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019867918744949424, |
| "loss": 1.5096, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019867361440075793, |
| "loss": 1.3083, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019866804135202165, |
| "loss": 1.3421, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019866246830328531, |
| "loss": 1.4146, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.000198656895254549, |
| "loss": 1.4521, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001986513222058127, |
| "loss": 1.4978, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001986457491570764, |
| "loss": 1.716, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019864017610834008, |
| "loss": 1.8349, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019863460305960377, |
| "loss": 1.8288, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019862903001086746, |
| "loss": 1.4265, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019862345696213115, |
| "loss": 1.3402, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019861788391339484, |
| "loss": 1.886, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019861231086465853, |
| "loss": 1.6905, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001986067378159222, |
| "loss": 1.7907, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001986011647671859, |
| "loss": 1.5227, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019859559171844958, |
| "loss": 1.8401, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019859001866971327, |
| "loss": 1.6064, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019858444562097696, |
| "loss": 1.2978, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019857887257224065, |
| "loss": 1.3655, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019857329952350434, |
| "loss": 1.6027, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019856772647476804, |
| "loss": 1.586, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019856215342603173, |
| "loss": 1.1341, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019855658037729542, |
| "loss": 1.4883, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019855100732855908, |
| "loss": 1.3435, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001985454342798228, |
| "loss": 1.4128, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019853986123108646, |
| "loss": 1.6067, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019853428818235016, |
| "loss": 1.1114, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019852871513361387, |
| "loss": 1.6321, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019852314208487754, |
| "loss": 1.5946, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019851756903614123, |
| "loss": 1.5754, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019851199598740492, |
| "loss": 1.5315, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001985064229386686, |
| "loss": 1.4563, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001985008498899323, |
| "loss": 1.4059, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019849527684119597, |
| "loss": 1.5746, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019848970379245968, |
| "loss": 1.3102, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019848413074372335, |
| "loss": 1.8039, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019847855769498704, |
| "loss": 1.6097, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019847298464625076, |
| "loss": 1.621, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019846741159751442, |
| "loss": 1.3808, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001984618385487781, |
| "loss": 1.8673, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001984562655000418, |
| "loss": 1.0692, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001984506924513055, |
| "loss": 1.5123, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001984451194025692, |
| "loss": 1.4968, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019843954635383288, |
| "loss": 1.7477, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019843397330509657, |
| "loss": 1.3281, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019842840025636026, |
| "loss": 1.2689, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019842282720762392, |
| "loss": 1.424, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019841725415888764, |
| "loss": 1.8043, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001984116811101513, |
| "loss": 1.7377, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.000198406108061415, |
| "loss": 1.3095, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001984005350126787, |
| "loss": 1.8551, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019839496196394238, |
| "loss": 1.397, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019838938891520607, |
| "loss": 1.392, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019838381586646976, |
| "loss": 1.5771, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019837824281773345, |
| "loss": 1.3431, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019837266976899714, |
| "loss": 1.4507, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019836709672026084, |
| "loss": 1.4567, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019836152367152453, |
| "loss": 1.4657, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001983559506227882, |
| "loss": 1.8878, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.0001983503775740519, |
| "loss": 1.5648, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 0.00019834480452531557, |
| "loss": 1.6085, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019833923147657926, |
| "loss": 1.3087, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019833365842784296, |
| "loss": 1.6899, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019832808537910665, |
| "loss": 1.5068, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019832251233037034, |
| "loss": 1.6126, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019831693928163403, |
| "loss": 1.4062, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019831136623289772, |
| "loss": 1.3456, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001983057931841614, |
| "loss": 1.6299, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019830022013542507, |
| "loss": 1.8296, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001982946470866888, |
| "loss": 1.4487, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019828907403795248, |
| "loss": 1.3323, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019828350098921615, |
| "loss": 1.7534, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019827792794047987, |
| "loss": 1.3204, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019827235489174353, |
| "loss": 1.4958, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019826678184300722, |
| "loss": 1.5474, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001982612087942709, |
| "loss": 1.5403, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001982556357455346, |
| "loss": 1.5899, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001982500626967983, |
| "loss": 1.5831, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019824448964806199, |
| "loss": 1.4816, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019823891659932568, |
| "loss": 1.4043, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019823334355058937, |
| "loss": 1.4102, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019822777050185303, |
| "loss": 1.6007, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019822219745311675, |
| "loss": 1.1657, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019821662440438041, |
| "loss": 1.3578, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001982110513556441, |
| "loss": 1.49, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001982054783069078, |
| "loss": 1.3736, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001981999052581715, |
| "loss": 1.1815, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019819433220943518, |
| "loss": 1.4721, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019818875916069887, |
| "loss": 1.5377, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019818318611196256, |
| "loss": 1.3485, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019817761306322625, |
| "loss": 1.3084, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019817204001448994, |
| "loss": 1.5412, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019816646696575364, |
| "loss": 1.4822, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001981608939170173, |
| "loss": 1.3214, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019815532086828102, |
| "loss": 1.2575, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001981497478195447, |
| "loss": 1.4562, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019814417477080837, |
| "loss": 1.59, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019813860172207206, |
| "loss": 1.5205, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019813302867333575, |
| "loss": 1.6017, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019812745562459945, |
| "loss": 1.4708, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019812188257586314, |
| "loss": 1.4072, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019811630952712683, |
| "loss": 1.345, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019811073647839052, |
| "loss": 1.4589, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019810516342965418, |
| "loss": 1.4654, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001980995903809179, |
| "loss": 1.4234, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001980940173321816, |
| "loss": 1.4393, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019808844428344526, |
| "loss": 1.5868, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019808287123470898, |
| "loss": 1.3939, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019807729818597264, |
| "loss": 1.3384, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019807172513723633, |
| "loss": 1.4185, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019806615208850002, |
| "loss": 1.4124, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001980605790397637, |
| "loss": 0.9699, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001980550059910274, |
| "loss": 1.0186, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019804943294229107, |
| "loss": 1.7373, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019804385989355479, |
| "loss": 1.0341, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019803828684481848, |
| "loss": 1.2764, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019803271379608214, |
| "loss": 1.8265, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019802714074734586, |
| "loss": 0.9158, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019802156769860952, |
| "loss": 1.3845, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019801599464987321, |
| "loss": 1.409, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019801042160113693, |
| "loss": 1.5294, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001980048485524006, |
| "loss": 1.5666, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001979992755036643, |
| "loss": 1.7121, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019799370245492798, |
| "loss": 1.1658, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019798812940619167, |
| "loss": 1.4492, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019798255635745536, |
| "loss": 1.1314, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019797698330871905, |
| "loss": 1.5509, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019797141025998274, |
| "loss": 1.5616, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001979658372112464, |
| "loss": 1.5735, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001979602641625101, |
| "loss": 1.4693, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019795469111377382, |
| "loss": 1.5813, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019794911806503748, |
| "loss": 1.1636, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019794354501630117, |
| "loss": 1.4733, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019793797196756486, |
| "loss": 1.1779, |
| "step": 372 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019793239891882855, |
| "loss": 1.3228, |
| "step": 373 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019792682587009225, |
| "loss": 1.7091, |
| "step": 374 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019792125282135594, |
| "loss": 1.4055, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019791567977261963, |
| "loss": 1.2592, |
| "step": 376 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001979101067238833, |
| "loss": 1.5345, |
| "step": 377 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.000197904533675147, |
| "loss": 1.4606, |
| "step": 378 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001978989606264107, |
| "loss": 1.4161, |
| "step": 379 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019789338757767437, |
| "loss": 1.535, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019788781452893808, |
| "loss": 1.169, |
| "step": 381 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019788224148020175, |
| "loss": 1.6242, |
| "step": 382 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019787666843146544, |
| "loss": 1.3576, |
| "step": 383 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019787109538272913, |
| "loss": 1.5814, |
| "step": 384 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019786552233399282, |
| "loss": 0.987, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001978599492852565, |
| "loss": 1.6049, |
| "step": 386 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019785437623652018, |
| "loss": 1.4358, |
| "step": 387 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001978488031877839, |
| "loss": 1.191, |
| "step": 388 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019784323013904759, |
| "loss": 1.3212, |
| "step": 389 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019783765709031125, |
| "loss": 1.4077, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019783208404157497, |
| "loss": 1.3443, |
| "step": 391 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019782651099283863, |
| "loss": 1.0604, |
| "step": 392 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019782093794410232, |
| "loss": 1.3385, |
| "step": 393 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019781536489536604, |
| "loss": 1.3244, |
| "step": 394 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001978097918466297, |
| "loss": 1.4297, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001978042187978934, |
| "loss": 1.5026, |
| "step": 396 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001977986457491571, |
| "loss": 1.3039, |
| "step": 397 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019779307270042078, |
| "loss": 1.2345, |
| "step": 398 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019778749965168447, |
| "loss": 1.3409, |
| "step": 399 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019778192660294813, |
| "loss": 1.6601, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.03, |
| "eval_loss": 1.4233789443969727, |
| "eval_runtime": 335.1392, |
| "eval_samples_per_second": 7.46, |
| "eval_steps_per_second": 0.934, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019777635355421185, |
| "loss": 1.5624, |
| "step": 401 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019777078050547552, |
| "loss": 1.5794, |
| "step": 402 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001977652074567392, |
| "loss": 1.4353, |
| "step": 403 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019775963440800293, |
| "loss": 1.4252, |
| "step": 404 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001977540613592666, |
| "loss": 1.334, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019774848831053028, |
| "loss": 1.0804, |
| "step": 406 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019774291526179397, |
| "loss": 1.7049, |
| "step": 407 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019773734221305766, |
| "loss": 1.1258, |
| "step": 408 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019773176916432135, |
| "loss": 1.2113, |
| "step": 409 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019772619611558505, |
| "loss": 1.376, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019772062306684874, |
| "loss": 1.3902, |
| "step": 411 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001977150500181124, |
| "loss": 1.3029, |
| "step": 412 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019770947696937612, |
| "loss": 1.5506, |
| "step": 413 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.0001977039039206398, |
| "loss": 1.3724, |
| "step": 414 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019769833087190347, |
| "loss": 1.0413, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019769275782316717, |
| "loss": 1.6466, |
| "step": 416 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019768718477443086, |
| "loss": 1.6243, |
| "step": 417 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 0.00019768161172569455, |
| "loss": 1.269, |
| "step": 418 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019767603867695824, |
| "loss": 0.9952, |
| "step": 419 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019767046562822193, |
| "loss": 1.48, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019766489257948562, |
| "loss": 1.3463, |
| "step": 421 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001976593195307493, |
| "loss": 1.2149, |
| "step": 422 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.000197653746482013, |
| "loss": 1.071, |
| "step": 423 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001976481734332767, |
| "loss": 1.4573, |
| "step": 424 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019764260038454036, |
| "loss": 1.4514, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019763702733580408, |
| "loss": 1.1117, |
| "step": 426 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019763145428706774, |
| "loss": 1.2302, |
| "step": 427 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019762588123833143, |
| "loss": 1.254, |
| "step": 428 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019762030818959515, |
| "loss": 1.331, |
| "step": 429 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019761473514085881, |
| "loss": 1.3154, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001976091620921225, |
| "loss": 1.5243, |
| "step": 431 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001976035890433862, |
| "loss": 1.7667, |
| "step": 432 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001975980159946499, |
| "loss": 0.9618, |
| "step": 433 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019759244294591358, |
| "loss": 1.0904, |
| "step": 434 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019758686989717724, |
| "loss": 1.5432, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019758129684844096, |
| "loss": 1.1655, |
| "step": 436 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019757572379970462, |
| "loss": 1.0637, |
| "step": 437 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019757015075096832, |
| "loss": 1.5018, |
| "step": 438 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019756457770223203, |
| "loss": 1.4103, |
| "step": 439 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001975590046534957, |
| "loss": 1.1777, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001975534316047594, |
| "loss": 1.6201, |
| "step": 441 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019754785855602308, |
| "loss": 1.6166, |
| "step": 442 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019754228550728677, |
| "loss": 1.4456, |
| "step": 443 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019753671245855046, |
| "loss": 1.4931, |
| "step": 444 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019753113940981415, |
| "loss": 1.3381, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019752556636107785, |
| "loss": 1.6941, |
| "step": 446 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001975199933123415, |
| "loss": 1.3825, |
| "step": 447 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001975144202636052, |
| "loss": 1.1368, |
| "step": 448 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019750884721486892, |
| "loss": 1.1474, |
| "step": 449 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019750327416613258, |
| "loss": 1.1068, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019749770111739627, |
| "loss": 1.2148, |
| "step": 451 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019749212806865996, |
| "loss": 1.9149, |
| "step": 452 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019748655501992366, |
| "loss": 1.4608, |
| "step": 453 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019748098197118735, |
| "loss": 1.2321, |
| "step": 454 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019747540892245104, |
| "loss": 1.4942, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019746983587371473, |
| "loss": 1.2238, |
| "step": 456 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019746426282497842, |
| "loss": 1.1944, |
| "step": 457 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001974586897762421, |
| "loss": 1.3321, |
| "step": 458 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001974531167275058, |
| "loss": 1.1132, |
| "step": 459 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019744754367876947, |
| "loss": 1.5997, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019744197063003318, |
| "loss": 1.2013, |
| "step": 461 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019743639758129685, |
| "loss": 1.7279, |
| "step": 462 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019743082453256054, |
| "loss": 1.3702, |
| "step": 463 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019742525148382423, |
| "loss": 1.3479, |
| "step": 464 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019741967843508792, |
| "loss": 1.5324, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019741410538635161, |
| "loss": 1.3117, |
| "step": 466 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001974085323376153, |
| "loss": 1.4931, |
| "step": 467 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.000197402959288879, |
| "loss": 1.2159, |
| "step": 468 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001973973862401427, |
| "loss": 1.2253, |
| "step": 469 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019739181319140635, |
| "loss": 1.5539, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019738624014267007, |
| "loss": 1.1525, |
| "step": 471 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019738066709393373, |
| "loss": 1.8453, |
| "step": 472 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019737509404519742, |
| "loss": 1.302, |
| "step": 473 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019736952099646114, |
| "loss": 1.3239, |
| "step": 474 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001973639479477248, |
| "loss": 1.519, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001973583748989885, |
| "loss": 1.1394, |
| "step": 476 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001973528018502522, |
| "loss": 1.2456, |
| "step": 477 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019734722880151588, |
| "loss": 0.9569, |
| "step": 478 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019734165575277957, |
| "loss": 1.5852, |
| "step": 479 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019733608270404324, |
| "loss": 1.4276, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019733050965530695, |
| "loss": 1.6277, |
| "step": 481 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019732493660657064, |
| "loss": 1.3751, |
| "step": 482 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001973193635578343, |
| "loss": 1.4353, |
| "step": 483 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019731379050909803, |
| "loss": 1.2159, |
| "step": 484 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001973082174603617, |
| "loss": 1.1527, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019730264441162538, |
| "loss": 1.1742, |
| "step": 486 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019729707136288907, |
| "loss": 1.5021, |
| "step": 487 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019729149831415276, |
| "loss": 1.1783, |
| "step": 488 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019728592526541646, |
| "loss": 1.2113, |
| "step": 489 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019728035221668015, |
| "loss": 1.3851, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019727477916794384, |
| "loss": 1.0802, |
| "step": 491 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019726920611920753, |
| "loss": 1.3358, |
| "step": 492 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019726363307047122, |
| "loss": 1.4349, |
| "step": 493 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001972580600217349, |
| "loss": 1.517, |
| "step": 494 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019725248697299858, |
| "loss": 1.2039, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019724691392426227, |
| "loss": 1.7408, |
| "step": 496 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019724134087552596, |
| "loss": 1.4491, |
| "step": 497 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019723576782678965, |
| "loss": 1.5237, |
| "step": 498 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019723019477805334, |
| "loss": 1.6063, |
| "step": 499 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019722462172931703, |
| "loss": 1.514, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019721904868058072, |
| "loss": 1.2975, |
| "step": 501 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001972134756318444, |
| "loss": 1.2387, |
| "step": 502 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001972079025831081, |
| "loss": 1.4889, |
| "step": 503 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001972023295343718, |
| "loss": 1.2638, |
| "step": 504 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019719675648563546, |
| "loss": 1.4609, |
| "step": 505 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019719118343689918, |
| "loss": 1.3076, |
| "step": 506 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019718561038816287, |
| "loss": 1.2533, |
| "step": 507 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019718003733942653, |
| "loss": 1.4696, |
| "step": 508 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019717446429069025, |
| "loss": 1.6291, |
| "step": 509 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019716889124195392, |
| "loss": 1.107, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001971633181932176, |
| "loss": 1.0402, |
| "step": 511 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001971577451444813, |
| "loss": 1.3351, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.000197152172095745, |
| "loss": 1.4979, |
| "step": 513 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019714659904700868, |
| "loss": 1.3694, |
| "step": 514 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019714102599827234, |
| "loss": 1.3115, |
| "step": 515 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019713545294953606, |
| "loss": 1.3879, |
| "step": 516 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019712987990079975, |
| "loss": 1.7215, |
| "step": 517 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019712430685206342, |
| "loss": 1.1419, |
| "step": 518 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019711873380332714, |
| "loss": 1.127, |
| "step": 519 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001971131607545908, |
| "loss": 1.2805, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001971075877058545, |
| "loss": 1.6912, |
| "step": 521 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019710201465711818, |
| "loss": 1.3006, |
| "step": 522 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019709644160838187, |
| "loss": 1.5531, |
| "step": 523 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019709086855964556, |
| "loss": 1.1554, |
| "step": 524 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019708529551090926, |
| "loss": 1.0171, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019707972246217295, |
| "loss": 1.3881, |
| "step": 526 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019707414941343664, |
| "loss": 1.0376, |
| "step": 527 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001970685763647003, |
| "loss": 1.1094, |
| "step": 528 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019706300331596402, |
| "loss": 1.1364, |
| "step": 529 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019705743026722768, |
| "loss": 1.7949, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019705185721849137, |
| "loss": 1.4688, |
| "step": 531 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001970462841697551, |
| "loss": 1.1975, |
| "step": 532 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019704071112101876, |
| "loss": 1.1809, |
| "step": 533 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019703513807228245, |
| "loss": 1.1835, |
| "step": 534 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019702956502354614, |
| "loss": 1.471, |
| "step": 535 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019702399197480983, |
| "loss": 1.5101, |
| "step": 536 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.00019701841892607352, |
| "loss": 1.108, |
| "step": 537 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 0.0001970128458773372, |
| "loss": 1.1264, |
| "step": 538 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001970072728286009, |
| "loss": 1.2666, |
| "step": 539 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019700169977986457, |
| "loss": 1.2887, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019699612673112829, |
| "loss": 1.1921, |
| "step": 541 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019699055368239198, |
| "loss": 1.1259, |
| "step": 542 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019698498063365564, |
| "loss": 1.0627, |
| "step": 543 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019697940758491933, |
| "loss": 1.5108, |
| "step": 544 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019697383453618302, |
| "loss": 1.2729, |
| "step": 545 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019696826148744671, |
| "loss": 1.2217, |
| "step": 546 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001969626884387104, |
| "loss": 1.43, |
| "step": 547 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001969571153899741, |
| "loss": 1.6275, |
| "step": 548 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001969515423412378, |
| "loss": 1.5857, |
| "step": 549 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019694596929250145, |
| "loss": 1.3591, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019694039624376517, |
| "loss": 1.0603, |
| "step": 551 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019693482319502886, |
| "loss": 1.1923, |
| "step": 552 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019692925014629253, |
| "loss": 1.2334, |
| "step": 553 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019692367709755624, |
| "loss": 1.3894, |
| "step": 554 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001969181040488199, |
| "loss": 1.6009, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001969125310000836, |
| "loss": 1.1167, |
| "step": 556 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019690695795134732, |
| "loss": 1.4777, |
| "step": 557 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019690138490261098, |
| "loss": 1.565, |
| "step": 558 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019689581185387467, |
| "loss": 1.3719, |
| "step": 559 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019689023880513834, |
| "loss": 1.3698, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019688466575640205, |
| "loss": 1.1463, |
| "step": 561 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019687909270766575, |
| "loss": 1.1895, |
| "step": 562 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001968735196589294, |
| "loss": 1.5997, |
| "step": 563 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019686794661019313, |
| "loss": 1.3521, |
| "step": 564 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001968623735614568, |
| "loss": 1.267, |
| "step": 565 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019685680051272048, |
| "loss": 1.548, |
| "step": 566 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001968512274639842, |
| "loss": 1.2816, |
| "step": 567 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019684565441524787, |
| "loss": 1.3505, |
| "step": 568 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019684008136651156, |
| "loss": 1.2535, |
| "step": 569 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019683450831777525, |
| "loss": 1.1666, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019682893526903894, |
| "loss": 1.1876, |
| "step": 571 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019682336222030263, |
| "loss": 1.3072, |
| "step": 572 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019681778917156632, |
| "loss": 1.5452, |
| "step": 573 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019681221612283, |
| "loss": 1.4827, |
| "step": 574 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019680664307409368, |
| "loss": 1.4328, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019680107002535737, |
| "loss": 1.119, |
| "step": 576 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019679549697662109, |
| "loss": 1.0863, |
| "step": 577 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019678992392788475, |
| "loss": 1.5337, |
| "step": 578 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019678435087914844, |
| "loss": 0.8858, |
| "step": 579 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019677877783041213, |
| "loss": 1.4041, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019677320478167582, |
| "loss": 1.1418, |
| "step": 581 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019676763173293951, |
| "loss": 1.6521, |
| "step": 582 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001967620586842032, |
| "loss": 1.1429, |
| "step": 583 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001967564856354669, |
| "loss": 1.2655, |
| "step": 584 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019675091258673056, |
| "loss": 1.2651, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019674533953799428, |
| "loss": 1.3223, |
| "step": 586 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019673976648925797, |
| "loss": 1.5126, |
| "step": 587 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019673419344052163, |
| "loss": 1.2796, |
| "step": 588 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019672862039178535, |
| "loss": 1.0685, |
| "step": 589 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019672304734304902, |
| "loss": 1.3054, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001967174742943127, |
| "loss": 1.2795, |
| "step": 591 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001967119012455764, |
| "loss": 1.5516, |
| "step": 592 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001967063281968401, |
| "loss": 1.3156, |
| "step": 593 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019670075514810378, |
| "loss": 1.1541, |
| "step": 594 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019669518209936747, |
| "loss": 1.2086, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019668960905063116, |
| "loss": 1.3997, |
| "step": 596 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019668403600189485, |
| "loss": 0.9795, |
| "step": 597 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019667846295315852, |
| "loss": 1.4819, |
| "step": 598 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019667288990442224, |
| "loss": 1.4381, |
| "step": 599 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001966673168556859, |
| "loss": 1.5546, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.05, |
| "eval_loss": 1.324468731880188, |
| "eval_runtime": 338.4128, |
| "eval_samples_per_second": 7.387, |
| "eval_steps_per_second": 0.925, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001966617438069496, |
| "loss": 1.3287, |
| "step": 601 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001966561707582133, |
| "loss": 1.1576, |
| "step": 602 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019665059770947697, |
| "loss": 1.28, |
| "step": 603 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019664502466074067, |
| "loss": 0.9028, |
| "step": 604 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019663945161200436, |
| "loss": 1.6244, |
| "step": 605 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019663387856326805, |
| "loss": 0.9109, |
| "step": 606 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019662830551453174, |
| "loss": 1.4525, |
| "step": 607 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001966227324657954, |
| "loss": 1.15, |
| "step": 608 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019661715941705912, |
| "loss": 1.1845, |
| "step": 609 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019661158636832279, |
| "loss": 1.6454, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019660601331958648, |
| "loss": 1.3668, |
| "step": 611 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001966004402708502, |
| "loss": 1.3132, |
| "step": 612 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019659486722211386, |
| "loss": 1.1158, |
| "step": 613 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019658929417337755, |
| "loss": 1.3215, |
| "step": 614 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019658372112464124, |
| "loss": 1.392, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019657814807590493, |
| "loss": 1.3152, |
| "step": 616 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019657257502716862, |
| "loss": 1.025, |
| "step": 617 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019656700197843231, |
| "loss": 1.4891, |
| "step": 618 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.000196561428929696, |
| "loss": 1.4011, |
| "step": 619 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019655585588095967, |
| "loss": 1.3183, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001965502828322234, |
| "loss": 1.3777, |
| "step": 621 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019654470978348708, |
| "loss": 1.3587, |
| "step": 622 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019653913673475074, |
| "loss": 1.2479, |
| "step": 623 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019653356368601446, |
| "loss": 1.7257, |
| "step": 624 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019652799063727813, |
| "loss": 1.361, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019652241758854182, |
| "loss": 1.4671, |
| "step": 626 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001965168445398055, |
| "loss": 0.8526, |
| "step": 627 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001965112714910692, |
| "loss": 1.4492, |
| "step": 628 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001965056984423329, |
| "loss": 1.5607, |
| "step": 629 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019650012539359658, |
| "loss": 1.5547, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019649455234486027, |
| "loss": 1.2846, |
| "step": 631 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019648897929612396, |
| "loss": 1.1433, |
| "step": 632 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019648340624738763, |
| "loss": 1.1029, |
| "step": 633 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019647783319865135, |
| "loss": 1.121, |
| "step": 634 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.000196472260149915, |
| "loss": 1.1928, |
| "step": 635 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001964666871011787, |
| "loss": 1.54, |
| "step": 636 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019646111405244242, |
| "loss": 1.3633, |
| "step": 637 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019645554100370608, |
| "loss": 1.7346, |
| "step": 638 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019644996795496977, |
| "loss": 1.3023, |
| "step": 639 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019644439490623347, |
| "loss": 1.0183, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019643882185749716, |
| "loss": 1.5982, |
| "step": 641 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019643324880876085, |
| "loss": 0.9071, |
| "step": 642 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001964276757600245, |
| "loss": 1.3193, |
| "step": 643 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019642210271128823, |
| "loss": 1.2086, |
| "step": 644 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001964165296625519, |
| "loss": 1.0997, |
| "step": 645 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019641095661381558, |
| "loss": 1.0801, |
| "step": 646 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001964053835650793, |
| "loss": 1.479, |
| "step": 647 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019639981051634297, |
| "loss": 1.2891, |
| "step": 648 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019639423746760666, |
| "loss": 1.1838, |
| "step": 649 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019638866441887035, |
| "loss": 1.3726, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019638309137013404, |
| "loss": 1.0871, |
| "step": 651 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019637751832139773, |
| "loss": 1.5271, |
| "step": 652 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019637194527266142, |
| "loss": 1.3749, |
| "step": 653 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.00019636637222392511, |
| "loss": 1.1373, |
| "step": 654 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001963607991751888, |
| "loss": 1.1627, |
| "step": 655 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001963552261264525, |
| "loss": 1.063, |
| "step": 656 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 0.0001963496530777162, |
| "loss": 1.5199, |
| "step": 657 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019634408002897985, |
| "loss": 1.3137, |
| "step": 658 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019633850698024354, |
| "loss": 0.992, |
| "step": 659 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019633293393150723, |
| "loss": 1.1835, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019632736088277092, |
| "loss": 1.2469, |
| "step": 661 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019632178783403462, |
| "loss": 1.3454, |
| "step": 662 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001963162147852983, |
| "loss": 0.8585, |
| "step": 663 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.000196310641736562, |
| "loss": 1.4115, |
| "step": 664 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001963050686878257, |
| "loss": 1.3726, |
| "step": 665 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019629949563908938, |
| "loss": 1.1821, |
| "step": 666 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019629392259035307, |
| "loss": 1.659, |
| "step": 667 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019628834954161674, |
| "loss": 1.4643, |
| "step": 668 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019628277649288045, |
| "loss": 1.3801, |
| "step": 669 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019627720344414412, |
| "loss": 1.1955, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001962716303954078, |
| "loss": 1.0883, |
| "step": 671 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019626605734667153, |
| "loss": 1.4792, |
| "step": 672 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001962604842979352, |
| "loss": 1.3863, |
| "step": 673 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019625491124919888, |
| "loss": 1.5576, |
| "step": 674 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019624933820046257, |
| "loss": 1.6188, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019624376515172626, |
| "loss": 1.5054, |
| "step": 676 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019623819210298996, |
| "loss": 0.9798, |
| "step": 677 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019623261905425362, |
| "loss": 1.4019, |
| "step": 678 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019622704600551734, |
| "loss": 1.4353, |
| "step": 679 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019622147295678103, |
| "loss": 1.517, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001962158999080447, |
| "loss": 1.2821, |
| "step": 681 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001962103268593084, |
| "loss": 1.62, |
| "step": 682 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019620475381057208, |
| "loss": 1.2364, |
| "step": 683 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019619918076183577, |
| "loss": 1.6882, |
| "step": 684 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019619360771309946, |
| "loss": 1.4248, |
| "step": 685 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019618803466436315, |
| "loss": 1.1829, |
| "step": 686 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019618246161562684, |
| "loss": 1.3408, |
| "step": 687 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019617688856689053, |
| "loss": 1.6336, |
| "step": 688 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019617131551815422, |
| "loss": 1.152, |
| "step": 689 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001961657424694179, |
| "loss": 1.1494, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019616016942068158, |
| "loss": 1.374, |
| "step": 691 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001961545963719453, |
| "loss": 1.1504, |
| "step": 692 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019614902332320896, |
| "loss": 1.0261, |
| "step": 693 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019614345027447265, |
| "loss": 1.2125, |
| "step": 694 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019613787722573634, |
| "loss": 1.0525, |
| "step": 695 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019613230417700003, |
| "loss": 1.3588, |
| "step": 696 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019612673112826372, |
| "loss": 1.316, |
| "step": 697 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019612115807952742, |
| "loss": 1.3402, |
| "step": 698 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001961155850307911, |
| "loss": 1.4421, |
| "step": 699 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001961100119820548, |
| "loss": 1.1457, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001961044389333185, |
| "loss": 1.552, |
| "step": 701 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019609886588458218, |
| "loss": 1.3951, |
| "step": 702 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019609329283584584, |
| "loss": 1.2811, |
| "step": 703 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019608771978710956, |
| "loss": 1.2365, |
| "step": 704 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019608214673837325, |
| "loss": 1.3474, |
| "step": 705 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019607657368963692, |
| "loss": 1.2968, |
| "step": 706 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001960710006409006, |
| "loss": 1.6594, |
| "step": 707 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001960654275921643, |
| "loss": 1.5059, |
| "step": 708 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.000196059854543428, |
| "loss": 0.9962, |
| "step": 709 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019605428149469168, |
| "loss": 1.0291, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019604870844595537, |
| "loss": 1.5544, |
| "step": 711 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019604313539721906, |
| "loss": 1.3657, |
| "step": 712 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019603756234848273, |
| "loss": 1.2451, |
| "step": 713 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019603198929974645, |
| "loss": 1.2719, |
| "step": 714 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019602641625101014, |
| "loss": 1.2008, |
| "step": 715 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001960208432022738, |
| "loss": 1.6069, |
| "step": 716 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019601527015353752, |
| "loss": 1.1346, |
| "step": 717 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019600969710480118, |
| "loss": 1.266, |
| "step": 718 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019600412405606488, |
| "loss": 1.1099, |
| "step": 719 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019599855100732857, |
| "loss": 1.5211, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019599297795859226, |
| "loss": 1.3112, |
| "step": 721 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019598740490985595, |
| "loss": 1.5219, |
| "step": 722 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001959818318611196, |
| "loss": 1.2916, |
| "step": 723 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019597625881238333, |
| "loss": 1.4544, |
| "step": 724 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019597068576364702, |
| "loss": 0.9491, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019596511271491069, |
| "loss": 1.3305, |
| "step": 726 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001959595396661744, |
| "loss": 1.26, |
| "step": 727 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019595396661743807, |
| "loss": 1.0675, |
| "step": 728 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019594839356870176, |
| "loss": 1.4793, |
| "step": 729 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019594282051996548, |
| "loss": 1.3387, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019593724747122914, |
| "loss": 1.1784, |
| "step": 731 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019593167442249283, |
| "loss": 1.0901, |
| "step": 732 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019592610137375652, |
| "loss": 0.9419, |
| "step": 733 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019592052832502022, |
| "loss": 1.381, |
| "step": 734 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001959149552762839, |
| "loss": 0.9874, |
| "step": 735 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001959093822275476, |
| "loss": 1.3332, |
| "step": 736 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001959038091788113, |
| "loss": 1.2267, |
| "step": 737 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019589823613007495, |
| "loss": 0.9698, |
| "step": 738 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019589266308133864, |
| "loss": 1.2921, |
| "step": 739 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019588709003260236, |
| "loss": 1.2611, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019588151698386603, |
| "loss": 1.4582, |
| "step": 741 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019587594393512972, |
| "loss": 1.1253, |
| "step": 742 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001958703708863934, |
| "loss": 1.8419, |
| "step": 743 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001958647978376571, |
| "loss": 1.1018, |
| "step": 744 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001958592247889208, |
| "loss": 0.9794, |
| "step": 745 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019585365174018448, |
| "loss": 1.2863, |
| "step": 746 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019584807869144817, |
| "loss": 1.1328, |
| "step": 747 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019584250564271184, |
| "loss": 1.1193, |
| "step": 748 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019583693259397556, |
| "loss": 1.5978, |
| "step": 749 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019583135954523925, |
| "loss": 1.1674, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001958257864965029, |
| "loss": 1.2662, |
| "step": 751 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019582021344776663, |
| "loss": 1.6532, |
| "step": 752 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001958146403990303, |
| "loss": 1.545, |
| "step": 753 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019580906735029398, |
| "loss": 1.2984, |
| "step": 754 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019580349430155767, |
| "loss": 1.6682, |
| "step": 755 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019579792125282137, |
| "loss": 1.1616, |
| "step": 756 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019579234820408506, |
| "loss": 1.7423, |
| "step": 757 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019578677515534872, |
| "loss": 1.2747, |
| "step": 758 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019578120210661244, |
| "loss": 1.3213, |
| "step": 759 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019577562905787613, |
| "loss": 0.7832, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001957700560091398, |
| "loss": 1.4308, |
| "step": 761 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001957644829604035, |
| "loss": 1.6004, |
| "step": 762 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019575890991166718, |
| "loss": 1.4012, |
| "step": 763 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019575333686293087, |
| "loss": 1.4956, |
| "step": 764 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019574776381419459, |
| "loss": 1.4045, |
| "step": 765 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019574219076545825, |
| "loss": 1.2474, |
| "step": 766 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019573661771672194, |
| "loss": 1.5545, |
| "step": 767 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019573104466798563, |
| "loss": 1.2795, |
| "step": 768 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019572547161924932, |
| "loss": 0.9551, |
| "step": 769 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019571989857051301, |
| "loss": 1.4014, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019571432552177668, |
| "loss": 0.8991, |
| "step": 771 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.0001957087524730404, |
| "loss": 1.281, |
| "step": 772 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019570317942430406, |
| "loss": 1.8978, |
| "step": 773 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019569760637556775, |
| "loss": 1.1675, |
| "step": 774 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019569203332683147, |
| "loss": 1.1951, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019568646027809513, |
| "loss": 1.2632, |
| "step": 776 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 0.00019568088722935883, |
| "loss": 1.269, |
| "step": 777 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019567531418062252, |
| "loss": 1.2243, |
| "step": 778 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001956697411318862, |
| "loss": 1.3442, |
| "step": 779 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001956641680831499, |
| "loss": 1.187, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001956585950344136, |
| "loss": 1.1981, |
| "step": 781 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019565302198567728, |
| "loss": 1.3294, |
| "step": 782 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019564744893694095, |
| "loss": 1.2843, |
| "step": 783 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019564187588820466, |
| "loss": 1.1876, |
| "step": 784 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019563630283946835, |
| "loss": 1.0514, |
| "step": 785 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019563072979073202, |
| "loss": 1.2881, |
| "step": 786 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001956251567419957, |
| "loss": 1.4303, |
| "step": 787 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001956195836932594, |
| "loss": 1.2964, |
| "step": 788 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001956140106445231, |
| "loss": 1.3653, |
| "step": 789 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019560843759578678, |
| "loss": 1.3172, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019560286454705047, |
| "loss": 1.5568, |
| "step": 791 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019559729149831417, |
| "loss": 1.2487, |
| "step": 792 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019559171844957783, |
| "loss": 1.0581, |
| "step": 793 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019558614540084155, |
| "loss": 1.4896, |
| "step": 794 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019558057235210524, |
| "loss": 1.4263, |
| "step": 795 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001955749993033689, |
| "loss": 1.6564, |
| "step": 796 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019556942625463262, |
| "loss": 1.1535, |
| "step": 797 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019556385320589629, |
| "loss": 1.2686, |
| "step": 798 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019555828015715998, |
| "loss": 1.1949, |
| "step": 799 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001955527071084237, |
| "loss": 1.4209, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.07, |
| "eval_loss": 1.266950011253357, |
| "eval_runtime": 335.7957, |
| "eval_samples_per_second": 7.445, |
| "eval_steps_per_second": 0.932, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019554713405968736, |
| "loss": 1.2128, |
| "step": 801 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019554156101095105, |
| "loss": 1.3104, |
| "step": 802 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019553598796221474, |
| "loss": 1.03, |
| "step": 803 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019553041491347843, |
| "loss": 1.1629, |
| "step": 804 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019552484186474212, |
| "loss": 1.3553, |
| "step": 805 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001955192688160058, |
| "loss": 0.9448, |
| "step": 806 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001955136957672695, |
| "loss": 1.0055, |
| "step": 807 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019550812271853317, |
| "loss": 1.3357, |
| "step": 808 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019550254966979686, |
| "loss": 0.9765, |
| "step": 809 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019549697662106058, |
| "loss": 1.1532, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019549140357232424, |
| "loss": 1.2812, |
| "step": 811 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019548583052358793, |
| "loss": 1.2928, |
| "step": 812 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019548025747485163, |
| "loss": 1.4332, |
| "step": 813 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019547468442611532, |
| "loss": 1.3307, |
| "step": 814 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.000195469111377379, |
| "loss": 1.105, |
| "step": 815 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001954635383286427, |
| "loss": 1.2064, |
| "step": 816 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001954579652799064, |
| "loss": 1.5388, |
| "step": 817 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019545239223117005, |
| "loss": 0.9905, |
| "step": 818 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019544681918243375, |
| "loss": 1.2416, |
| "step": 819 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019544124613369746, |
| "loss": 1.0135, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019543567308496113, |
| "loss": 1.1938, |
| "step": 821 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019543010003622482, |
| "loss": 0.9052, |
| "step": 822 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001954245269874885, |
| "loss": 1.1654, |
| "step": 823 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001954189539387522, |
| "loss": 1.3552, |
| "step": 824 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001954133808900159, |
| "loss": 1.2389, |
| "step": 825 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019540780784127958, |
| "loss": 1.3809, |
| "step": 826 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019540223479254327, |
| "loss": 1.3791, |
| "step": 827 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019539666174380697, |
| "loss": 1.4006, |
| "step": 828 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019539108869507066, |
| "loss": 1.406, |
| "step": 829 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019538551564633435, |
| "loss": 1.3309, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.000195379942597598, |
| "loss": 1.3305, |
| "step": 831 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019537436954886173, |
| "loss": 1.1471, |
| "step": 832 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001953687965001254, |
| "loss": 1.1592, |
| "step": 833 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019536322345138909, |
| "loss": 1.3974, |
| "step": 834 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019535765040265278, |
| "loss": 1.5061, |
| "step": 835 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019535207735391647, |
| "loss": 0.9674, |
| "step": 836 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019534650430518016, |
| "loss": 1.3929, |
| "step": 837 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019534093125644385, |
| "loss": 1.4159, |
| "step": 838 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019533535820770754, |
| "loss": 1.0006, |
| "step": 839 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019532978515897123, |
| "loss": 1.3088, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001953242121102349, |
| "loss": 0.9453, |
| "step": 841 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019531863906149861, |
| "loss": 1.296, |
| "step": 842 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019531306601276228, |
| "loss": 0.9098, |
| "step": 843 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019530749296402597, |
| "loss": 1.1839, |
| "step": 844 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001953019199152897, |
| "loss": 1.1279, |
| "step": 845 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019529634686655335, |
| "loss": 1.2579, |
| "step": 846 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019529077381781704, |
| "loss": 1.2611, |
| "step": 847 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019528520076908073, |
| "loss": 1.7317, |
| "step": 848 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019527962772034443, |
| "loss": 1.2227, |
| "step": 849 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019527405467160812, |
| "loss": 1.5092, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019526848162287178, |
| "loss": 1.1227, |
| "step": 851 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001952629085741355, |
| "loss": 1.1195, |
| "step": 852 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001952573355253992, |
| "loss": 1.4941, |
| "step": 853 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019525176247666285, |
| "loss": 1.1511, |
| "step": 854 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019524618942792657, |
| "loss": 1.2617, |
| "step": 855 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019524061637919024, |
| "loss": 1.4552, |
| "step": 856 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019523504333045393, |
| "loss": 1.3651, |
| "step": 857 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019522947028171762, |
| "loss": 1.3737, |
| "step": 858 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001952238972329813, |
| "loss": 1.2444, |
| "step": 859 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.000195218324184245, |
| "loss": 1.1565, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001952127511355087, |
| "loss": 1.2848, |
| "step": 861 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019520717808677238, |
| "loss": 1.3267, |
| "step": 862 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019520160503803607, |
| "loss": 1.2781, |
| "step": 863 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019519603198929977, |
| "loss": 1.2919, |
| "step": 864 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019519045894056346, |
| "loss": 1.6833, |
| "step": 865 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019518488589182712, |
| "loss": 1.3214, |
| "step": 866 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001951793128430908, |
| "loss": 1.2948, |
| "step": 867 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001951737397943545, |
| "loss": 1.32, |
| "step": 868 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001951681667456182, |
| "loss": 1.402, |
| "step": 869 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019516259369688188, |
| "loss": 0.9637, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019515702064814558, |
| "loss": 1.2191, |
| "step": 871 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019515144759940927, |
| "loss": 1.1487, |
| "step": 872 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019514587455067296, |
| "loss": 1.4219, |
| "step": 873 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019514030150193665, |
| "loss": 1.2563, |
| "step": 874 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019513472845320034, |
| "loss": 1.2979, |
| "step": 875 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.000195129155404464, |
| "loss": 1.0446, |
| "step": 876 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019512358235572772, |
| "loss": 0.9354, |
| "step": 877 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019511800930699141, |
| "loss": 1.0332, |
| "step": 878 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019511243625825508, |
| "loss": 1.397, |
| "step": 879 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001951068632095188, |
| "loss": 1.5956, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019510129016078246, |
| "loss": 1.5201, |
| "step": 881 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019509571711204615, |
| "loss": 0.7995, |
| "step": 882 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019509014406330984, |
| "loss": 1.3428, |
| "step": 883 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019508457101457353, |
| "loss": 1.2337, |
| "step": 884 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019507899796583722, |
| "loss": 1.1943, |
| "step": 885 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001950734249171009, |
| "loss": 1.3515, |
| "step": 886 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001950678518683646, |
| "loss": 1.3067, |
| "step": 887 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001950622788196283, |
| "loss": 1.33, |
| "step": 888 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019505670577089196, |
| "loss": 1.4342, |
| "step": 889 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019505113272215568, |
| "loss": 0.9805, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019504555967341934, |
| "loss": 1.344, |
| "step": 891 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019503998662468304, |
| "loss": 1.284, |
| "step": 892 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019503441357594673, |
| "loss": 1.3546, |
| "step": 893 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.00019502884052721042, |
| "loss": 1.5881, |
| "step": 894 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001950232674784741, |
| "loss": 1.1926, |
| "step": 895 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001950176944297378, |
| "loss": 1.2381, |
| "step": 896 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 0.0001950121213810015, |
| "loss": 1.5537, |
| "step": 897 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019500654833226518, |
| "loss": 1.2881, |
| "step": 898 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019500097528352885, |
| "loss": 1.092, |
| "step": 899 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019499540223479256, |
| "loss": 1.1725, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019498982918605623, |
| "loss": 1.2175, |
| "step": 901 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019498425613731992, |
| "loss": 1.126, |
| "step": 902 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019497868308858364, |
| "loss": 1.0826, |
| "step": 903 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001949731100398473, |
| "loss": 1.2947, |
| "step": 904 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.000194967536991111, |
| "loss": 1.6054, |
| "step": 905 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019496196394237468, |
| "loss": 1.4027, |
| "step": 906 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019495639089363838, |
| "loss": 0.8667, |
| "step": 907 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019495081784490207, |
| "loss": 1.4129, |
| "step": 908 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019494524479616576, |
| "loss": 1.06, |
| "step": 909 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019493967174742945, |
| "loss": 1.3846, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001949340986986931, |
| "loss": 1.2854, |
| "step": 911 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019492852564995683, |
| "loss": 1.4006, |
| "step": 912 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019492295260122052, |
| "loss": 1.1776, |
| "step": 913 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019491737955248419, |
| "loss": 1.3006, |
| "step": 914 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019491180650374788, |
| "loss": 1.1347, |
| "step": 915 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019490623345501157, |
| "loss": 1.4366, |
| "step": 916 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019490066040627526, |
| "loss": 1.415, |
| "step": 917 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019489508735753895, |
| "loss": 1.1139, |
| "step": 918 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019488951430880264, |
| "loss": 1.3203, |
| "step": 919 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019488394126006633, |
| "loss": 1.3933, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019487836821133, |
| "loss": 1.3785, |
| "step": 921 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019487279516259372, |
| "loss": 1.0783, |
| "step": 922 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001948672221138574, |
| "loss": 1.5086, |
| "step": 923 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019486164906512107, |
| "loss": 1.157, |
| "step": 924 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001948560760163848, |
| "loss": 1.1194, |
| "step": 925 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019485050296764845, |
| "loss": 1.3122, |
| "step": 926 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019484492991891214, |
| "loss": 1.4338, |
| "step": 927 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019483935687017584, |
| "loss": 1.0879, |
| "step": 928 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019483378382143953, |
| "loss": 1.0764, |
| "step": 929 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019482821077270322, |
| "loss": 1.2771, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001948226377239669, |
| "loss": 1.2639, |
| "step": 931 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001948170646752306, |
| "loss": 1.1461, |
| "step": 932 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001948114916264943, |
| "loss": 1.3512, |
| "step": 933 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019480591857775795, |
| "loss": 1.5393, |
| "step": 934 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019480034552902167, |
| "loss": 1.042, |
| "step": 935 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019479477248028534, |
| "loss": 1.3993, |
| "step": 936 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019478919943154903, |
| "loss": 1.2221, |
| "step": 937 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019478362638281275, |
| "loss": 0.9009, |
| "step": 938 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001947780533340764, |
| "loss": 1.1361, |
| "step": 939 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001947724802853401, |
| "loss": 0.9519, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001947669072366038, |
| "loss": 1.1502, |
| "step": 941 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019476133418786748, |
| "loss": 1.3629, |
| "step": 942 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019475576113913118, |
| "loss": 1.3921, |
| "step": 943 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019475018809039487, |
| "loss": 1.2214, |
| "step": 944 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019474461504165856, |
| "loss": 1.2931, |
| "step": 945 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019473904199292222, |
| "loss": 1.4022, |
| "step": 946 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019473346894418594, |
| "loss": 1.2765, |
| "step": 947 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019472789589544963, |
| "loss": 1.0564, |
| "step": 948 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001947223228467133, |
| "loss": 1.3838, |
| "step": 949 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019471674979797699, |
| "loss": 1.2021, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019471117674924068, |
| "loss": 1.3176, |
| "step": 951 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019470560370050437, |
| "loss": 1.1771, |
| "step": 952 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019470003065176806, |
| "loss": 1.3712, |
| "step": 953 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019469445760303175, |
| "loss": 1.1437, |
| "step": 954 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019468888455429544, |
| "loss": 1.3199, |
| "step": 955 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001946833115055591, |
| "loss": 1.1929, |
| "step": 956 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019467773845682282, |
| "loss": 1.239, |
| "step": 957 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019467216540808652, |
| "loss": 1.1912, |
| "step": 958 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019466659235935018, |
| "loss": 1.2826, |
| "step": 959 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001946610193106139, |
| "loss": 1.3047, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019465544626187756, |
| "loss": 1.3338, |
| "step": 961 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019464987321314125, |
| "loss": 1.3601, |
| "step": 962 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019464430016440497, |
| "loss": 1.6185, |
| "step": 963 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019463872711566863, |
| "loss": 1.1702, |
| "step": 964 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019463315406693233, |
| "loss": 1.3376, |
| "step": 965 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.000194627581018196, |
| "loss": 1.4091, |
| "step": 966 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001946220079694597, |
| "loss": 1.5644, |
| "step": 967 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001946164349207234, |
| "loss": 1.2682, |
| "step": 968 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019461086187198706, |
| "loss": 1.2137, |
| "step": 969 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019460528882325078, |
| "loss": 1.5127, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019459971577451445, |
| "loss": 1.1806, |
| "step": 971 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019459414272577814, |
| "loss": 1.1052, |
| "step": 972 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019458856967704186, |
| "loss": 1.194, |
| "step": 973 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019458299662830552, |
| "loss": 1.1145, |
| "step": 974 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001945774235795692, |
| "loss": 1.2175, |
| "step": 975 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001945718505308329, |
| "loss": 0.9813, |
| "step": 976 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001945662774820966, |
| "loss": 1.2283, |
| "step": 977 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019456070443336028, |
| "loss": 0.9486, |
| "step": 978 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019455513138462397, |
| "loss": 1.3503, |
| "step": 979 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019454955833588767, |
| "loss": 1.4512, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019454398528715133, |
| "loss": 1.151, |
| "step": 981 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019453841223841502, |
| "loss": 1.4991, |
| "step": 982 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019453283918967874, |
| "loss": 1.2788, |
| "step": 983 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001945272661409424, |
| "loss": 1.1673, |
| "step": 984 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001945216930922061, |
| "loss": 1.071, |
| "step": 985 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019451612004346979, |
| "loss": 1.262, |
| "step": 986 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019451054699473348, |
| "loss": 0.8417, |
| "step": 987 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019450497394599717, |
| "loss": 1.0144, |
| "step": 988 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019449940089726086, |
| "loss": 1.178, |
| "step": 989 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019449382784852455, |
| "loss": 1.1975, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019448825479978821, |
| "loss": 1.2633, |
| "step": 991 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019448268175105193, |
| "loss": 1.0848, |
| "step": 992 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019447710870231562, |
| "loss": 1.3465, |
| "step": 993 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001944715356535793, |
| "loss": 1.2955, |
| "step": 994 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.000194465962604843, |
| "loss": 1.2062, |
| "step": 995 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019446038955610667, |
| "loss": 1.1467, |
| "step": 996 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019445481650737036, |
| "loss": 1.6877, |
| "step": 997 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019444924345863405, |
| "loss": 0.927, |
| "step": 998 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019444367040989774, |
| "loss": 1.163, |
| "step": 999 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019443809736116143, |
| "loss": 0.9808, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.08, |
| "eval_loss": 1.2298390865325928, |
| "eval_runtime": 335.3164, |
| "eval_samples_per_second": 7.456, |
| "eval_steps_per_second": 0.933, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019443252431242513, |
| "loss": 1.5363, |
| "step": 1001 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019442695126368882, |
| "loss": 1.0627, |
| "step": 1002 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001944213782149525, |
| "loss": 1.2339, |
| "step": 1003 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019441580516621617, |
| "loss": 1.598, |
| "step": 1004 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001944102321174799, |
| "loss": 1.0745, |
| "step": 1005 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019440465906874355, |
| "loss": 1.1209, |
| "step": 1006 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019439908602000725, |
| "loss": 1.5394, |
| "step": 1007 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019439351297127096, |
| "loss": 1.2985, |
| "step": 1008 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019438793992253463, |
| "loss": 1.1478, |
| "step": 1009 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019438236687379832, |
| "loss": 1.3328, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.000194376793825062, |
| "loss": 1.5156, |
| "step": 1011 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001943712207763257, |
| "loss": 1.1239, |
| "step": 1012 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.0001943656477275894, |
| "loss": 1.2798, |
| "step": 1013 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019436007467885306, |
| "loss": 1.3353, |
| "step": 1014 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019435450163011677, |
| "loss": 1.2072, |
| "step": 1015 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00019434892858138044, |
| "loss": 1.1777, |
| "step": 1016 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019434335553264413, |
| "loss": 1.1929, |
| "step": 1017 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019433778248390785, |
| "loss": 0.8905, |
| "step": 1018 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001943322094351715, |
| "loss": 1.1023, |
| "step": 1019 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001943266363864352, |
| "loss": 1.4121, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001943210633376989, |
| "loss": 0.8888, |
| "step": 1021 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019431549028896259, |
| "loss": 1.4086, |
| "step": 1022 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019430991724022628, |
| "loss": 1.3479, |
| "step": 1023 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019430434419148997, |
| "loss": 1.1927, |
| "step": 1024 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019429877114275366, |
| "loss": 1.2784, |
| "step": 1025 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019429319809401735, |
| "loss": 1.137, |
| "step": 1026 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019428762504528104, |
| "loss": 1.254, |
| "step": 1027 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019428205199654473, |
| "loss": 1.2929, |
| "step": 1028 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001942764789478084, |
| "loss": 1.426, |
| "step": 1029 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001942709058990721, |
| "loss": 1.1741, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019426533285033578, |
| "loss": 1.2022, |
| "step": 1031 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019425975980159947, |
| "loss": 1.0017, |
| "step": 1032 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019425418675286316, |
| "loss": 0.9802, |
| "step": 1033 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019424861370412685, |
| "loss": 1.092, |
| "step": 1034 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019424304065539054, |
| "loss": 1.0989, |
| "step": 1035 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019423746760665423, |
| "loss": 0.8877, |
| "step": 1036 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019423189455791793, |
| "loss": 1.6528, |
| "step": 1037 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019422632150918162, |
| "loss": 1.2604, |
| "step": 1038 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019422074846044528, |
| "loss": 1.4248, |
| "step": 1039 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.000194215175411709, |
| "loss": 1.2264, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019420960236297266, |
| "loss": 1.2461, |
| "step": 1041 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019420402931423635, |
| "loss": 1.2471, |
| "step": 1042 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019419845626550007, |
| "loss": 1.2429, |
| "step": 1043 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019419288321676374, |
| "loss": 1.2154, |
| "step": 1044 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019418731016802743, |
| "loss": 0.9198, |
| "step": 1045 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019418173711929112, |
| "loss": 1.3046, |
| "step": 1046 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001941761640705548, |
| "loss": 1.2664, |
| "step": 1047 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001941705910218185, |
| "loss": 1.0377, |
| "step": 1048 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019416501797308216, |
| "loss": 0.9935, |
| "step": 1049 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019415944492434588, |
| "loss": 1.3128, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019415387187560957, |
| "loss": 1.162, |
| "step": 1051 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019414829882687324, |
| "loss": 1.0107, |
| "step": 1052 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019414272577813696, |
| "loss": 1.1744, |
| "step": 1053 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019413715272940062, |
| "loss": 0.9117, |
| "step": 1054 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001941315796806643, |
| "loss": 1.1199, |
| "step": 1055 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.000194126006631928, |
| "loss": 0.9564, |
| "step": 1056 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001941204335831917, |
| "loss": 0.9818, |
| "step": 1057 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019411486053445539, |
| "loss": 1.2819, |
| "step": 1058 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019410928748571908, |
| "loss": 1.0244, |
| "step": 1059 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019410371443698277, |
| "loss": 1.1171, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019409814138824646, |
| "loss": 1.4476, |
| "step": 1061 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019409256833951012, |
| "loss": 1.3266, |
| "step": 1062 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019408699529077384, |
| "loss": 1.0029, |
| "step": 1063 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001940814222420375, |
| "loss": 1.0301, |
| "step": 1064 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001940758491933012, |
| "loss": 1.0985, |
| "step": 1065 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001940702761445649, |
| "loss": 1.1136, |
| "step": 1066 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019406470309582858, |
| "loss": 1.133, |
| "step": 1067 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019405913004709227, |
| "loss": 1.18, |
| "step": 1068 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019405355699835596, |
| "loss": 1.182, |
| "step": 1069 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019404798394961965, |
| "loss": 1.2494, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019404241090088334, |
| "loss": 1.0142, |
| "step": 1071 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019403683785214703, |
| "loss": 1.0812, |
| "step": 1072 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019403126480341073, |
| "loss": 1.2911, |
| "step": 1073 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001940256917546744, |
| "loss": 0.8737, |
| "step": 1074 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001940201187059381, |
| "loss": 1.093, |
| "step": 1075 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001940145456572018, |
| "loss": 0.9175, |
| "step": 1076 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019400897260846546, |
| "loss": 1.1662, |
| "step": 1077 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019400339955972915, |
| "loss": 1.1228, |
| "step": 1078 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019399782651099284, |
| "loss": 1.3625, |
| "step": 1079 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019399225346225654, |
| "loss": 1.1024, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019398668041352023, |
| "loss": 1.3526, |
| "step": 1081 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019398110736478392, |
| "loss": 1.3015, |
| "step": 1082 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001939755343160476, |
| "loss": 1.269, |
| "step": 1083 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019396996126731127, |
| "loss": 1.2318, |
| "step": 1084 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.000193964388218575, |
| "loss": 1.3367, |
| "step": 1085 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019395881516983868, |
| "loss": 1.1352, |
| "step": 1086 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019395324212110235, |
| "loss": 1.1896, |
| "step": 1087 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019394766907236607, |
| "loss": 1.4012, |
| "step": 1088 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019394209602362973, |
| "loss": 1.3227, |
| "step": 1089 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019393652297489342, |
| "loss": 0.8919, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001939309499261571, |
| "loss": 1.2599, |
| "step": 1091 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001939253768774208, |
| "loss": 1.2739, |
| "step": 1092 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001939198038286845, |
| "loss": 1.2172, |
| "step": 1093 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019391423077994816, |
| "loss": 1.123, |
| "step": 1094 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019390865773121188, |
| "loss": 1.1185, |
| "step": 1095 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019390308468247557, |
| "loss": 1.2129, |
| "step": 1096 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019389751163373923, |
| "loss": 1.338, |
| "step": 1097 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019389193858500295, |
| "loss": 1.0814, |
| "step": 1098 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001938863655362666, |
| "loss": 1.3794, |
| "step": 1099 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001938807924875303, |
| "loss": 1.4278, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.000193875219438794, |
| "loss": 1.3475, |
| "step": 1101 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001938696463900577, |
| "loss": 1.2439, |
| "step": 1102 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019386407334132138, |
| "loss": 1.213, |
| "step": 1103 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019385850029258507, |
| "loss": 1.3873, |
| "step": 1104 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019385292724384876, |
| "loss": 0.9215, |
| "step": 1105 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019384735419511245, |
| "loss": 0.9843, |
| "step": 1106 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019384178114637614, |
| "loss": 1.2136, |
| "step": 1107 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019383620809763983, |
| "loss": 1.3081, |
| "step": 1108 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001938306350489035, |
| "loss": 1.1979, |
| "step": 1109 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001938250620001672, |
| "loss": 1.2707, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001938194889514309, |
| "loss": 1.4537, |
| "step": 1111 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019381391590269457, |
| "loss": 1.1772, |
| "step": 1112 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019380834285395826, |
| "loss": 0.9286, |
| "step": 1113 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019380276980522195, |
| "loss": 0.9999, |
| "step": 1114 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019379719675648564, |
| "loss": 1.2043, |
| "step": 1115 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019379162370774934, |
| "loss": 1.0697, |
| "step": 1116 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019378605065901303, |
| "loss": 0.9873, |
| "step": 1117 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019378047761027672, |
| "loss": 1.3224, |
| "step": 1118 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019377490456154038, |
| "loss": 1.1977, |
| "step": 1119 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001937693315128041, |
| "loss": 1.0302, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001937637584640678, |
| "loss": 1.3423, |
| "step": 1121 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019375818541533146, |
| "loss": 1.0534, |
| "step": 1122 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019375261236659517, |
| "loss": 1.1376, |
| "step": 1123 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019374703931785884, |
| "loss": 1.4215, |
| "step": 1124 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019374146626912253, |
| "loss": 1.0303, |
| "step": 1125 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019373589322038622, |
| "loss": 1.0293, |
| "step": 1126 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001937303201716499, |
| "loss": 1.4243, |
| "step": 1127 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001937247471229136, |
| "loss": 1.1761, |
| "step": 1128 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019371917407417727, |
| "loss": 0.9815, |
| "step": 1129 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019371360102544098, |
| "loss": 1.3422, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019370802797670468, |
| "loss": 1.2999, |
| "step": 1131 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019370245492796834, |
| "loss": 1.1452, |
| "step": 1132 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019369688187923206, |
| "loss": 1.0217, |
| "step": 1133 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019369130883049572, |
| "loss": 1.3833, |
| "step": 1134 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.0001936857357817594, |
| "loss": 1.4013, |
| "step": 1135 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 0.00019368016273302313, |
| "loss": 1.3016, |
| "step": 1136 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001936745896842868, |
| "loss": 1.406, |
| "step": 1137 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019366901663555049, |
| "loss": 0.9434, |
| "step": 1138 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019366344358681418, |
| "loss": 1.4208, |
| "step": 1139 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019365787053807787, |
| "loss": 1.2194, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019365229748934156, |
| "loss": 1.0899, |
| "step": 1141 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019364672444060522, |
| "loss": 1.2113, |
| "step": 1142 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019364115139186894, |
| "loss": 1.151, |
| "step": 1143 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001936355783431326, |
| "loss": 1.0539, |
| "step": 1144 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001936300052943963, |
| "loss": 1.1184, |
| "step": 1145 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019362443224566002, |
| "loss": 1.1578, |
| "step": 1146 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019361885919692368, |
| "loss": 1.4283, |
| "step": 1147 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019361328614818737, |
| "loss": 1.073, |
| "step": 1148 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019360771309945106, |
| "loss": 1.1779, |
| "step": 1149 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019360214005071475, |
| "loss": 1.254, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019359656700197844, |
| "loss": 1.5673, |
| "step": 1151 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019359099395324214, |
| "loss": 1.1627, |
| "step": 1152 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019358542090450583, |
| "loss": 1.2797, |
| "step": 1153 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001935798478557695, |
| "loss": 1.1232, |
| "step": 1154 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001935742748070332, |
| "loss": 1.3957, |
| "step": 1155 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001935687017582969, |
| "loss": 1.349, |
| "step": 1156 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019356312870956056, |
| "loss": 1.3759, |
| "step": 1157 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019355755566082425, |
| "loss": 1.274, |
| "step": 1158 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019355198261208795, |
| "loss": 1.3563, |
| "step": 1159 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019354640956335164, |
| "loss": 1.2315, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019354083651461533, |
| "loss": 1.0633, |
| "step": 1161 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019353526346587902, |
| "loss": 1.077, |
| "step": 1162 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001935296904171427, |
| "loss": 1.167, |
| "step": 1163 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019352411736840637, |
| "loss": 1.5249, |
| "step": 1164 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001935185443196701, |
| "loss": 1.3342, |
| "step": 1165 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019351297127093378, |
| "loss": 1.0158, |
| "step": 1166 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019350739822219745, |
| "loss": 0.8628, |
| "step": 1167 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019350182517346117, |
| "loss": 1.4473, |
| "step": 1168 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019349625212472483, |
| "loss": 1.0282, |
| "step": 1169 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019349067907598852, |
| "loss": 0.8979, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019348510602725224, |
| "loss": 1.2513, |
| "step": 1171 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001934795329785159, |
| "loss": 1.0077, |
| "step": 1172 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001934739599297796, |
| "loss": 1.0327, |
| "step": 1173 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019346838688104329, |
| "loss": 1.1176, |
| "step": 1174 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019346281383230698, |
| "loss": 1.3309, |
| "step": 1175 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019345724078357067, |
| "loss": 1.4316, |
| "step": 1176 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019345166773483433, |
| "loss": 0.8597, |
| "step": 1177 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019344609468609805, |
| "loss": 1.3275, |
| "step": 1178 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019344052163736171, |
| "loss": 1.3712, |
| "step": 1179 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001934349485886254, |
| "loss": 1.3807, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019342937553988912, |
| "loss": 1.2281, |
| "step": 1181 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001934238024911528, |
| "loss": 1.2941, |
| "step": 1182 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019341822944241648, |
| "loss": 1.042, |
| "step": 1183 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019341265639368017, |
| "loss": 1.1746, |
| "step": 1184 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019340708334494386, |
| "loss": 1.1887, |
| "step": 1185 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019340151029620755, |
| "loss": 1.2484, |
| "step": 1186 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019339593724747124, |
| "loss": 1.2106, |
| "step": 1187 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019339036419873493, |
| "loss": 1.3317, |
| "step": 1188 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001933847911499986, |
| "loss": 1.1706, |
| "step": 1189 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001933792181012623, |
| "loss": 1.4321, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.000193373645052526, |
| "loss": 1.1906, |
| "step": 1191 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019336807200378967, |
| "loss": 1.0152, |
| "step": 1192 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019336249895505336, |
| "loss": 1.421, |
| "step": 1193 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019335692590631705, |
| "loss": 1.0722, |
| "step": 1194 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019335135285758075, |
| "loss": 1.4288, |
| "step": 1195 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019334577980884444, |
| "loss": 0.9111, |
| "step": 1196 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019334020676010813, |
| "loss": 1.3859, |
| "step": 1197 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019333463371137182, |
| "loss": 1.0314, |
| "step": 1198 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001933290606626355, |
| "loss": 1.1992, |
| "step": 1199 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001933234876138992, |
| "loss": 0.8524, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.1, |
| "eval_loss": 1.2000236511230469, |
| "eval_runtime": 340.3539, |
| "eval_samples_per_second": 7.345, |
| "eval_steps_per_second": 0.92, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001933179145651629, |
| "loss": 1.2325, |
| "step": 1201 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019331234151642656, |
| "loss": 1.2821, |
| "step": 1202 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019330676846769027, |
| "loss": 1.3642, |
| "step": 1203 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019330119541895394, |
| "loss": 1.362, |
| "step": 1204 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019329562237021763, |
| "loss": 1.3452, |
| "step": 1205 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019329004932148132, |
| "loss": 1.0043, |
| "step": 1206 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.000193284476272745, |
| "loss": 1.5709, |
| "step": 1207 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001932789032240087, |
| "loss": 1.0189, |
| "step": 1208 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001932733301752724, |
| "loss": 0.9555, |
| "step": 1209 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019326775712653609, |
| "loss": 1.0179, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019326218407779978, |
| "loss": 0.9294, |
| "step": 1211 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019325661102906344, |
| "loss": 1.3548, |
| "step": 1212 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019325103798032716, |
| "loss": 1.21, |
| "step": 1213 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019324546493159082, |
| "loss": 0.9961, |
| "step": 1214 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019323989188285451, |
| "loss": 0.8625, |
| "step": 1215 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019323431883411823, |
| "loss": 1.305, |
| "step": 1216 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001932287457853819, |
| "loss": 1.2502, |
| "step": 1217 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001932231727366456, |
| "loss": 1.181, |
| "step": 1218 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019321759968790928, |
| "loss": 1.1019, |
| "step": 1219 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019321202663917297, |
| "loss": 1.1409, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019320645359043666, |
| "loss": 1.3311, |
| "step": 1221 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019320088054170033, |
| "loss": 1.1954, |
| "step": 1222 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019319530749296404, |
| "loss": 1.2213, |
| "step": 1223 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019318973444422773, |
| "loss": 0.9975, |
| "step": 1224 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001931841613954914, |
| "loss": 1.1659, |
| "step": 1225 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019317858834675512, |
| "loss": 1.4524, |
| "step": 1226 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019317301529801878, |
| "loss": 1.071, |
| "step": 1227 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019316744224928247, |
| "loss": 1.0248, |
| "step": 1228 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019316186920054616, |
| "loss": 1.4124, |
| "step": 1229 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019315629615180985, |
| "loss": 1.3157, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019315072310307355, |
| "loss": 1.0985, |
| "step": 1231 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019314515005433724, |
| "loss": 1.4938, |
| "step": 1232 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019313957700560093, |
| "loss": 1.4369, |
| "step": 1233 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019313400395686462, |
| "loss": 1.1318, |
| "step": 1234 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001931284309081283, |
| "loss": 1.1227, |
| "step": 1235 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.000193122857859392, |
| "loss": 1.4437, |
| "step": 1236 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019311728481065567, |
| "loss": 1.1824, |
| "step": 1237 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019311171176191938, |
| "loss": 0.8434, |
| "step": 1238 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019310613871318305, |
| "loss": 1.2331, |
| "step": 1239 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019310056566444674, |
| "loss": 1.2534, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019309499261571043, |
| "loss": 1.1355, |
| "step": 1241 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019308941956697412, |
| "loss": 1.0885, |
| "step": 1242 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001930838465182378, |
| "loss": 1.0709, |
| "step": 1243 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001930782734695015, |
| "loss": 1.182, |
| "step": 1244 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001930727004207652, |
| "loss": 1.0661, |
| "step": 1245 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019306712737202889, |
| "loss": 1.1287, |
| "step": 1246 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019306155432329255, |
| "loss": 1.6649, |
| "step": 1247 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019305598127455627, |
| "loss": 1.081, |
| "step": 1248 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019305040822581996, |
| "loss": 1.1536, |
| "step": 1249 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019304483517708362, |
| "loss": 1.1749, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019303926212834734, |
| "loss": 1.0295, |
| "step": 1251 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.000193033689079611, |
| "loss": 1.0391, |
| "step": 1252 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001930281160308747, |
| "loss": 1.5651, |
| "step": 1253 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.0001930225429821384, |
| "loss": 1.3661, |
| "step": 1254 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019301696993340208, |
| "loss": 1.2866, |
| "step": 1255 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 0.00019301139688466577, |
| "loss": 1.4991, |
| "step": 1256 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019300582383592943, |
| "loss": 1.1126, |
| "step": 1257 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019300025078719315, |
| "loss": 1.3542, |
| "step": 1258 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019299467773845684, |
| "loss": 0.8879, |
| "step": 1259 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001929891046897205, |
| "loss": 1.0657, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019298353164098423, |
| "loss": 1.0622, |
| "step": 1261 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001929779585922479, |
| "loss": 1.1014, |
| "step": 1262 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019297238554351158, |
| "loss": 1.2585, |
| "step": 1263 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019296681249477527, |
| "loss": 1.0764, |
| "step": 1264 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019296123944603896, |
| "loss": 1.3296, |
| "step": 1265 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019295566639730265, |
| "loss": 1.0764, |
| "step": 1266 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019295009334856635, |
| "loss": 1.2223, |
| "step": 1267 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019294452029983004, |
| "loss": 1.2085, |
| "step": 1268 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019293894725109373, |
| "loss": 1.2538, |
| "step": 1269 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019293337420235742, |
| "loss": 1.099, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001929278011536211, |
| "loss": 1.4428, |
| "step": 1271 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019292222810488477, |
| "loss": 1.1416, |
| "step": 1272 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019291665505614846, |
| "loss": 0.863, |
| "step": 1273 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019291108200741216, |
| "loss": 1.1065, |
| "step": 1274 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019290550895867585, |
| "loss": 1.0851, |
| "step": 1275 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019289993590993954, |
| "loss": 1.1004, |
| "step": 1276 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019289436286120323, |
| "loss": 1.4009, |
| "step": 1277 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019288878981246692, |
| "loss": 1.0918, |
| "step": 1278 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001928832167637306, |
| "loss": 1.2646, |
| "step": 1279 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001928776437149943, |
| "loss": 0.9876, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.000192872070666258, |
| "loss": 1.0403, |
| "step": 1281 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019286649761752166, |
| "loss": 1.365, |
| "step": 1282 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019286092456878538, |
| "loss": 1.1263, |
| "step": 1283 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019285535152004907, |
| "loss": 0.9112, |
| "step": 1284 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019284977847131273, |
| "loss": 0.915, |
| "step": 1285 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019284420542257645, |
| "loss": 1.3051, |
| "step": 1286 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019283863237384011, |
| "loss": 1.0453, |
| "step": 1287 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001928330593251038, |
| "loss": 1.1712, |
| "step": 1288 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001928274862763675, |
| "loss": 1.3691, |
| "step": 1289 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001928219132276312, |
| "loss": 1.4234, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019281634017889488, |
| "loss": 1.1152, |
| "step": 1291 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019281076713015854, |
| "loss": 1.3003, |
| "step": 1292 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019280519408142226, |
| "loss": 1.2676, |
| "step": 1293 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019279962103268595, |
| "loss": 1.1128, |
| "step": 1294 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019279404798394962, |
| "loss": 1.1777, |
| "step": 1295 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019278847493521333, |
| "loss": 1.1711, |
| "step": 1296 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.000192782901886477, |
| "loss": 1.0425, |
| "step": 1297 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001927773288377407, |
| "loss": 1.1897, |
| "step": 1298 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019277175578900438, |
| "loss": 1.0969, |
| "step": 1299 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019276618274026807, |
| "loss": 1.3076, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019276060969153176, |
| "loss": 1.4467, |
| "step": 1301 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019275503664279545, |
| "loss": 1.164, |
| "step": 1302 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019274946359405914, |
| "loss": 1.1374, |
| "step": 1303 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019274389054532284, |
| "loss": 1.193, |
| "step": 1304 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001927383174965865, |
| "loss": 1.0974, |
| "step": 1305 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019273274444785022, |
| "loss": 1.1949, |
| "step": 1306 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019272717139911388, |
| "loss": 1.3475, |
| "step": 1307 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019272159835037757, |
| "loss": 1.45, |
| "step": 1308 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001927160253016413, |
| "loss": 1.2412, |
| "step": 1309 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019271045225290496, |
| "loss": 1.182, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019270487920416865, |
| "loss": 1.164, |
| "step": 1311 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019269930615543234, |
| "loss": 0.9939, |
| "step": 1312 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019269373310669603, |
| "loss": 1.3454, |
| "step": 1313 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019268816005795972, |
| "loss": 1.2915, |
| "step": 1314 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001926825870092234, |
| "loss": 0.9285, |
| "step": 1315 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001926770139604871, |
| "loss": 1.0952, |
| "step": 1316 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019267144091175077, |
| "loss": 1.5117, |
| "step": 1317 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019266586786301448, |
| "loss": 1.0206, |
| "step": 1318 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019266029481427818, |
| "loss": 1.0674, |
| "step": 1319 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019265472176554184, |
| "loss": 0.9538, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019264914871680553, |
| "loss": 1.1128, |
| "step": 1321 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019264357566806922, |
| "loss": 1.1485, |
| "step": 1322 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001926380026193329, |
| "loss": 0.7513, |
| "step": 1323 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001926324295705966, |
| "loss": 1.3245, |
| "step": 1324 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001926268565218603, |
| "loss": 1.3742, |
| "step": 1325 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.000192621283473124, |
| "loss": 1.4126, |
| "step": 1326 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019261571042438765, |
| "loss": 1.1545, |
| "step": 1327 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019261013737565137, |
| "loss": 0.8756, |
| "step": 1328 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019260456432691506, |
| "loss": 1.2918, |
| "step": 1329 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019259899127817872, |
| "loss": 1.47, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019259341822944244, |
| "loss": 1.4201, |
| "step": 1331 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001925878451807061, |
| "loss": 1.0393, |
| "step": 1332 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001925822721319698, |
| "loss": 1.2965, |
| "step": 1333 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019257669908323352, |
| "loss": 1.5396, |
| "step": 1334 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019257112603449718, |
| "loss": 1.2073, |
| "step": 1335 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019256555298576087, |
| "loss": 1.1631, |
| "step": 1336 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019255997993702454, |
| "loss": 1.038, |
| "step": 1337 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019255440688828825, |
| "loss": 1.2831, |
| "step": 1338 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019254883383955194, |
| "loss": 1.4127, |
| "step": 1339 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001925432607908156, |
| "loss": 1.0839, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019253768774207933, |
| "loss": 1.3711, |
| "step": 1341 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.000192532114693343, |
| "loss": 1.105, |
| "step": 1342 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019252654164460668, |
| "loss": 1.1521, |
| "step": 1343 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001925209685958704, |
| "loss": 1.2919, |
| "step": 1344 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019251539554713406, |
| "loss": 1.0802, |
| "step": 1345 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019250982249839776, |
| "loss": 1.0488, |
| "step": 1346 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019250424944966145, |
| "loss": 0.9023, |
| "step": 1347 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019249867640092514, |
| "loss": 1.3583, |
| "step": 1348 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019249310335218883, |
| "loss": 1.1062, |
| "step": 1349 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019248753030345252, |
| "loss": 1.0262, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001924819572547162, |
| "loss": 1.1491, |
| "step": 1351 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019247638420597988, |
| "loss": 1.4069, |
| "step": 1352 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019247081115724357, |
| "loss": 1.0424, |
| "step": 1353 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019246523810850728, |
| "loss": 1.4738, |
| "step": 1354 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019245966505977095, |
| "loss": 1.2528, |
| "step": 1355 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019245409201103464, |
| "loss": 1.2726, |
| "step": 1356 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019244851896229833, |
| "loss": 1.359, |
| "step": 1357 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019244294591356202, |
| "loss": 1.349, |
| "step": 1358 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001924373728648257, |
| "loss": 1.1386, |
| "step": 1359 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001924317998160894, |
| "loss": 1.0529, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001924262267673531, |
| "loss": 0.9043, |
| "step": 1361 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019242065371861676, |
| "loss": 1.4212, |
| "step": 1362 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019241508066988048, |
| "loss": 1.236, |
| "step": 1363 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019240950762114417, |
| "loss": 1.2852, |
| "step": 1364 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019240393457240783, |
| "loss": 1.4515, |
| "step": 1365 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019239836152367155, |
| "loss": 0.9616, |
| "step": 1366 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019239278847493521, |
| "loss": 1.2158, |
| "step": 1367 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001923872154261989, |
| "loss": 1.1901, |
| "step": 1368 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001923816423774626, |
| "loss": 1.4584, |
| "step": 1369 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.0001923760693287263, |
| "loss": 0.9969, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019237049627998998, |
| "loss": 1.3601, |
| "step": 1371 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019236492323125367, |
| "loss": 1.5132, |
| "step": 1372 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019235935018251736, |
| "loss": 1.1854, |
| "step": 1373 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019235377713378105, |
| "loss": 1.3411, |
| "step": 1374 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 0.00019234820408504472, |
| "loss": 1.1116, |
| "step": 1375 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019234263103630844, |
| "loss": 1.0935, |
| "step": 1376 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001923370579875721, |
| "loss": 1.1641, |
| "step": 1377 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001923314849388358, |
| "loss": 1.2387, |
| "step": 1378 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001923259118900995, |
| "loss": 1.0264, |
| "step": 1379 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019232033884136317, |
| "loss": 1.0818, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019231476579262686, |
| "loss": 1.1352, |
| "step": 1381 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019230919274389055, |
| "loss": 1.4058, |
| "step": 1382 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019230361969515425, |
| "loss": 1.1458, |
| "step": 1383 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019229804664641794, |
| "loss": 1.1671, |
| "step": 1384 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001922924735976816, |
| "loss": 1.1466, |
| "step": 1385 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019228690054894532, |
| "loss": 1.4361, |
| "step": 1386 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019228132750020898, |
| "loss": 1.177, |
| "step": 1387 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019227575445147267, |
| "loss": 0.7267, |
| "step": 1388 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001922701814027364, |
| "loss": 1.0102, |
| "step": 1389 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019226460835400006, |
| "loss": 1.4188, |
| "step": 1390 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019225903530526375, |
| "loss": 0.862, |
| "step": 1391 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019225346225652744, |
| "loss": 1.0191, |
| "step": 1392 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019224788920779113, |
| "loss": 1.2016, |
| "step": 1393 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019224231615905482, |
| "loss": 1.3822, |
| "step": 1394 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001922367431103185, |
| "loss": 1.2253, |
| "step": 1395 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001922311700615822, |
| "loss": 1.1714, |
| "step": 1396 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001922255970128459, |
| "loss": 1.2149, |
| "step": 1397 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019222002396410959, |
| "loss": 1.0215, |
| "step": 1398 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019221445091537328, |
| "loss": 1.0405, |
| "step": 1399 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019220887786663694, |
| "loss": 1.4017, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.12, |
| "eval_loss": 1.1665562391281128, |
| "eval_runtime": 327.8779, |
| "eval_samples_per_second": 7.625, |
| "eval_steps_per_second": 0.955, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019220330481790063, |
| "loss": 1.1964, |
| "step": 1401 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019219773176916432, |
| "loss": 1.1125, |
| "step": 1402 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019219215872042801, |
| "loss": 1.3824, |
| "step": 1403 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001921865856716917, |
| "loss": 0.9097, |
| "step": 1404 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001921810126229554, |
| "loss": 1.1429, |
| "step": 1405 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001921754395742191, |
| "loss": 1.4914, |
| "step": 1406 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019216986652548278, |
| "loss": 1.068, |
| "step": 1407 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019216429347674647, |
| "loss": 0.8961, |
| "step": 1408 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019215872042801016, |
| "loss": 1.1012, |
| "step": 1409 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019215314737927383, |
| "loss": 1.2527, |
| "step": 1410 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019214757433053754, |
| "loss": 1.0999, |
| "step": 1411 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001921420012818012, |
| "loss": 1.1481, |
| "step": 1412 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001921364282330649, |
| "loss": 1.3547, |
| "step": 1413 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019213085518432862, |
| "loss": 0.9436, |
| "step": 1414 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019212528213559228, |
| "loss": 0.8395, |
| "step": 1415 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019211970908685597, |
| "loss": 1.1498, |
| "step": 1416 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019211413603811966, |
| "loss": 1.5606, |
| "step": 1417 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019210856298938335, |
| "loss": 1.1923, |
| "step": 1418 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019210298994064705, |
| "loss": 1.353, |
| "step": 1419 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001920974168919107, |
| "loss": 1.5093, |
| "step": 1420 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019209184384317443, |
| "loss": 1.1402, |
| "step": 1421 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019208627079443812, |
| "loss": 1.3652, |
| "step": 1422 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019208069774570178, |
| "loss": 0.9717, |
| "step": 1423 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001920751246969655, |
| "loss": 1.5994, |
| "step": 1424 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019206955164822917, |
| "loss": 1.0433, |
| "step": 1425 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019206397859949286, |
| "loss": 1.3135, |
| "step": 1426 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019205840555075655, |
| "loss": 1.1506, |
| "step": 1427 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019205283250202024, |
| "loss": 1.1936, |
| "step": 1428 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019204725945328393, |
| "loss": 0.9366, |
| "step": 1429 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019204168640454762, |
| "loss": 1.0535, |
| "step": 1430 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001920361133558113, |
| "loss": 1.4523, |
| "step": 1431 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.000192030540307075, |
| "loss": 1.1486, |
| "step": 1432 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019202496725833867, |
| "loss": 0.9461, |
| "step": 1433 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019201939420960239, |
| "loss": 1.2933, |
| "step": 1434 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019201382116086605, |
| "loss": 1.316, |
| "step": 1435 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019200824811212974, |
| "loss": 1.1872, |
| "step": 1436 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019200267506339343, |
| "loss": 1.2905, |
| "step": 1437 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019199710201465712, |
| "loss": 1.056, |
| "step": 1438 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019199152896592081, |
| "loss": 1.3167, |
| "step": 1439 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001919859559171845, |
| "loss": 1.182, |
| "step": 1440 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001919803828684482, |
| "loss": 1.228, |
| "step": 1441 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001919748098197119, |
| "loss": 0.892, |
| "step": 1442 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019196923677097558, |
| "loss": 1.2763, |
| "step": 1443 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019196366372223927, |
| "loss": 1.4267, |
| "step": 1444 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019195809067350293, |
| "loss": 1.0776, |
| "step": 1445 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019195251762476665, |
| "loss": 1.3386, |
| "step": 1446 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019194694457603032, |
| "loss": 0.9512, |
| "step": 1447 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.000191941371527294, |
| "loss": 1.3044, |
| "step": 1448 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001919357984785577, |
| "loss": 1.3314, |
| "step": 1449 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001919302254298214, |
| "loss": 1.1549, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019192465238108508, |
| "loss": 1.5689, |
| "step": 1451 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019191907933234877, |
| "loss": 1.1171, |
| "step": 1452 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019191350628361246, |
| "loss": 1.151, |
| "step": 1453 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019190793323487615, |
| "loss": 1.0682, |
| "step": 1454 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019190236018613982, |
| "loss": 1.129, |
| "step": 1455 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019189678713740354, |
| "loss": 1.019, |
| "step": 1456 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019189121408866723, |
| "loss": 1.2154, |
| "step": 1457 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001918856410399309, |
| "loss": 1.1298, |
| "step": 1458 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001918800679911946, |
| "loss": 1.2665, |
| "step": 1459 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019187449494245827, |
| "loss": 1.0305, |
| "step": 1460 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019186892189372197, |
| "loss": 1.183, |
| "step": 1461 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019186334884498566, |
| "loss": 1.1417, |
| "step": 1462 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019185777579624935, |
| "loss": 1.399, |
| "step": 1463 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019185220274751304, |
| "loss": 1.1837, |
| "step": 1464 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001918466296987767, |
| "loss": 1.1726, |
| "step": 1465 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019184105665004042, |
| "loss": 0.8539, |
| "step": 1466 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001918354836013041, |
| "loss": 0.9877, |
| "step": 1467 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019182991055256778, |
| "loss": 1.1403, |
| "step": 1468 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001918243375038315, |
| "loss": 1.1251, |
| "step": 1469 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019181876445509516, |
| "loss": 0.8338, |
| "step": 1470 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019181319140635885, |
| "loss": 1.0004, |
| "step": 1471 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019180761835762254, |
| "loss": 1.2612, |
| "step": 1472 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019180204530888623, |
| "loss": 1.2266, |
| "step": 1473 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019179647226014992, |
| "loss": 0.9146, |
| "step": 1474 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019179089921141361, |
| "loss": 1.1519, |
| "step": 1475 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001917853261626773, |
| "loss": 1.4246, |
| "step": 1476 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.000191779753113941, |
| "loss": 0.9761, |
| "step": 1477 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001917741800652047, |
| "loss": 1.0315, |
| "step": 1478 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019176860701646838, |
| "loss": 0.985, |
| "step": 1479 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019176303396773204, |
| "loss": 1.4143, |
| "step": 1480 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019175746091899573, |
| "loss": 1.0042, |
| "step": 1481 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019175188787025945, |
| "loss": 1.2024, |
| "step": 1482 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019174631482152312, |
| "loss": 0.9474, |
| "step": 1483 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001917407417727868, |
| "loss": 1.1631, |
| "step": 1484 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001917351687240505, |
| "loss": 1.3142, |
| "step": 1485 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.0001917295956753142, |
| "loss": 1.0917, |
| "step": 1486 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019172402262657788, |
| "loss": 1.1581, |
| "step": 1487 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019171844957784157, |
| "loss": 1.226, |
| "step": 1488 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019171287652910526, |
| "loss": 1.0155, |
| "step": 1489 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019170730348036893, |
| "loss": 0.9809, |
| "step": 1490 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019170173043163265, |
| "loss": 1.176, |
| "step": 1491 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019169615738289634, |
| "loss": 1.0372, |
| "step": 1492 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019169058433416, |
| "loss": 1.1557, |
| "step": 1493 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019168501128542372, |
| "loss": 1.2501, |
| "step": 1494 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 0.00019167943823668738, |
| "loss": 1.2716, |
| "step": 1495 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019167386518795107, |
| "loss": 1.2829, |
| "step": 1496 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019166829213921476, |
| "loss": 1.328, |
| "step": 1497 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019166271909047846, |
| "loss": 1.1182, |
| "step": 1498 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019165714604174215, |
| "loss": 1.0426, |
| "step": 1499 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001916515729930058, |
| "loss": 1.0957, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019164599994426953, |
| "loss": 1.1941, |
| "step": 1501 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019164042689553322, |
| "loss": 0.9335, |
| "step": 1502 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019163485384679688, |
| "loss": 1.2799, |
| "step": 1503 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001916292807980606, |
| "loss": 0.8841, |
| "step": 1504 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019162370774932427, |
| "loss": 1.2144, |
| "step": 1505 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019161813470058796, |
| "loss": 1.1778, |
| "step": 1506 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019161256165185168, |
| "loss": 0.7735, |
| "step": 1507 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019160698860311534, |
| "loss": 1.2204, |
| "step": 1508 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019160141555437903, |
| "loss": 1.1002, |
| "step": 1509 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019159584250564272, |
| "loss": 1.0857, |
| "step": 1510 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019159026945690641, |
| "loss": 0.9145, |
| "step": 1511 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001915846964081701, |
| "loss": 1.1851, |
| "step": 1512 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019157912335943377, |
| "loss": 1.122, |
| "step": 1513 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001915735503106975, |
| "loss": 1.1806, |
| "step": 1514 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019156797726196115, |
| "loss": 1.1607, |
| "step": 1515 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019156240421322484, |
| "loss": 1.1977, |
| "step": 1516 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019155683116448856, |
| "loss": 1.2688, |
| "step": 1517 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019155125811575222, |
| "loss": 0.6619, |
| "step": 1518 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019154568506701592, |
| "loss": 0.9936, |
| "step": 1519 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001915401120182796, |
| "loss": 1.1469, |
| "step": 1520 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001915345389695433, |
| "loss": 1.2209, |
| "step": 1521 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.000191528965920807, |
| "loss": 1.0166, |
| "step": 1522 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019152339287207068, |
| "loss": 0.783, |
| "step": 1523 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019151781982333437, |
| "loss": 1.3299, |
| "step": 1524 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019151224677459804, |
| "loss": 1.3132, |
| "step": 1525 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019150667372586175, |
| "loss": 1.0869, |
| "step": 1526 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019150110067712544, |
| "loss": 0.976, |
| "step": 1527 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001914955276283891, |
| "loss": 1.3061, |
| "step": 1528 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001914899545796528, |
| "loss": 0.8581, |
| "step": 1529 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001914843815309165, |
| "loss": 1.2351, |
| "step": 1530 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019147880848218018, |
| "loss": 1.1405, |
| "step": 1531 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019147323543344387, |
| "loss": 0.9639, |
| "step": 1532 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019146766238470756, |
| "loss": 1.0442, |
| "step": 1533 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019146208933597126, |
| "loss": 1.0799, |
| "step": 1534 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019145651628723492, |
| "loss": 1.199, |
| "step": 1535 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019145094323849864, |
| "loss": 1.2353, |
| "step": 1536 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019144537018976233, |
| "loss": 1.1126, |
| "step": 1537 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.000191439797141026, |
| "loss": 1.0697, |
| "step": 1538 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001914342240922897, |
| "loss": 1.419, |
| "step": 1539 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019142865104355338, |
| "loss": 1.1019, |
| "step": 1540 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019142307799481707, |
| "loss": 1.1041, |
| "step": 1541 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019141750494608078, |
| "loss": 1.4483, |
| "step": 1542 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019141193189734445, |
| "loss": 1.2548, |
| "step": 1543 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019140635884860814, |
| "loss": 1.0546, |
| "step": 1544 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019140078579987183, |
| "loss": 1.0322, |
| "step": 1545 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019139521275113552, |
| "loss": 1.0886, |
| "step": 1546 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001913896397023992, |
| "loss": 0.8544, |
| "step": 1547 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019138406665366288, |
| "loss": 1.2837, |
| "step": 1548 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001913784936049266, |
| "loss": 1.1647, |
| "step": 1549 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019137292055619026, |
| "loss": 1.0729, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019136734750745395, |
| "loss": 0.8369, |
| "step": 1551 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019136177445871767, |
| "loss": 1.1524, |
| "step": 1552 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019135620140998133, |
| "loss": 1.1289, |
| "step": 1553 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019135062836124502, |
| "loss": 1.175, |
| "step": 1554 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019134505531250872, |
| "loss": 0.9882, |
| "step": 1555 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001913394822637724, |
| "loss": 1.1599, |
| "step": 1556 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001913339092150361, |
| "loss": 1.2442, |
| "step": 1557 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001913283361662998, |
| "loss": 0.8461, |
| "step": 1558 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019132276311756348, |
| "loss": 1.2808, |
| "step": 1559 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019131719006882714, |
| "loss": 1.396, |
| "step": 1560 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019131161702009086, |
| "loss": 1.2583, |
| "step": 1561 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019130604397135455, |
| "loss": 1.3001, |
| "step": 1562 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019130047092261822, |
| "loss": 1.1768, |
| "step": 1563 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001912948978738819, |
| "loss": 1.3313, |
| "step": 1564 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001912893248251456, |
| "loss": 1.2876, |
| "step": 1565 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001912837517764093, |
| "loss": 0.9956, |
| "step": 1566 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019127817872767298, |
| "loss": 1.0823, |
| "step": 1567 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019127260567893667, |
| "loss": 1.156, |
| "step": 1568 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019126703263020036, |
| "loss": 1.2475, |
| "step": 1569 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019126145958146406, |
| "loss": 1.3739, |
| "step": 1570 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019125588653272775, |
| "loss": 1.1111, |
| "step": 1571 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019125031348399144, |
| "loss": 1.1276, |
| "step": 1572 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001912447404352551, |
| "loss": 1.1698, |
| "step": 1573 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019123916738651882, |
| "loss": 1.0037, |
| "step": 1574 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019123359433778248, |
| "loss": 0.9158, |
| "step": 1575 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019122802128904617, |
| "loss": 1.4917, |
| "step": 1576 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001912224482403099, |
| "loss": 0.9887, |
| "step": 1577 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019121687519157356, |
| "loss": 1.1122, |
| "step": 1578 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019121130214283725, |
| "loss": 1.0683, |
| "step": 1579 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019120572909410094, |
| "loss": 1.3064, |
| "step": 1580 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019120015604536463, |
| "loss": 1.4323, |
| "step": 1581 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019119458299662832, |
| "loss": 1.1678, |
| "step": 1582 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019118900994789199, |
| "loss": 1.0395, |
| "step": 1583 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001911834368991557, |
| "loss": 1.1243, |
| "step": 1584 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019117786385041937, |
| "loss": 1.1653, |
| "step": 1585 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019117229080168306, |
| "loss": 1.2014, |
| "step": 1586 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019116671775294678, |
| "loss": 0.9698, |
| "step": 1587 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019116114470421044, |
| "loss": 1.0687, |
| "step": 1588 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019115557165547413, |
| "loss": 1.2428, |
| "step": 1589 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019114999860673782, |
| "loss": 1.465, |
| "step": 1590 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019114442555800151, |
| "loss": 1.3103, |
| "step": 1591 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001911388525092652, |
| "loss": 1.228, |
| "step": 1592 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001911332794605289, |
| "loss": 0.8827, |
| "step": 1593 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001911277064117926, |
| "loss": 1.2108, |
| "step": 1594 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019112213336305628, |
| "loss": 1.1728, |
| "step": 1595 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019111656031431994, |
| "loss": 1.2073, |
| "step": 1596 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019111098726558366, |
| "loss": 1.412, |
| "step": 1597 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019110541421684733, |
| "loss": 1.1045, |
| "step": 1598 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019109984116811102, |
| "loss": 1.0349, |
| "step": 1599 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001910942681193747, |
| "loss": 1.1782, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.13, |
| "eval_loss": 1.1423087120056152, |
| "eval_runtime": 334.9697, |
| "eval_samples_per_second": 7.463, |
| "eval_steps_per_second": 0.934, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001910886950706384, |
| "loss": 1.3783, |
| "step": 1601 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001910831220219021, |
| "loss": 1.0844, |
| "step": 1602 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019107754897316578, |
| "loss": 1.1508, |
| "step": 1603 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019107197592442947, |
| "loss": 0.8586, |
| "step": 1604 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019106640287569316, |
| "loss": 0.9407, |
| "step": 1605 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019106082982695685, |
| "loss": 1.3276, |
| "step": 1606 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019105525677822055, |
| "loss": 1.1746, |
| "step": 1607 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001910496837294842, |
| "loss": 1.0579, |
| "step": 1608 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019104411068074793, |
| "loss": 1.0106, |
| "step": 1609 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.0001910385376320116, |
| "loss": 0.9762, |
| "step": 1610 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019103296458327528, |
| "loss": 1.1978, |
| "step": 1611 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019102739153453897, |
| "loss": 1.1678, |
| "step": 1612 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019102181848580267, |
| "loss": 1.0779, |
| "step": 1613 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019101624543706636, |
| "loss": 0.9737, |
| "step": 1614 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 0.00019101067238833005, |
| "loss": 1.327, |
| "step": 1615 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019100509933959374, |
| "loss": 0.9171, |
| "step": 1616 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019099952629085743, |
| "loss": 0.8715, |
| "step": 1617 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001909939532421211, |
| "loss": 1.1349, |
| "step": 1618 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001909883801933848, |
| "loss": 1.2493, |
| "step": 1619 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019098280714464848, |
| "loss": 1.3801, |
| "step": 1620 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019097723409591217, |
| "loss": 0.919, |
| "step": 1621 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019097166104717589, |
| "loss": 1.2133, |
| "step": 1622 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019096608799843955, |
| "loss": 0.7352, |
| "step": 1623 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019096051494970324, |
| "loss": 0.9488, |
| "step": 1624 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019095494190096693, |
| "loss": 0.9864, |
| "step": 1625 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019094936885223062, |
| "loss": 0.8861, |
| "step": 1626 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019094379580349431, |
| "loss": 0.9744, |
| "step": 1627 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019093822275475798, |
| "loss": 1.2037, |
| "step": 1628 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001909326497060217, |
| "loss": 0.9278, |
| "step": 1629 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001909270766572854, |
| "loss": 0.8685, |
| "step": 1630 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019092150360854905, |
| "loss": 0.951, |
| "step": 1631 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019091593055981277, |
| "loss": 0.916, |
| "step": 1632 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019091035751107643, |
| "loss": 1.3175, |
| "step": 1633 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019090478446234013, |
| "loss": 1.0757, |
| "step": 1634 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019089921141360382, |
| "loss": 1.0403, |
| "step": 1635 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001908936383648675, |
| "loss": 1.0756, |
| "step": 1636 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001908880653161312, |
| "loss": 1.4667, |
| "step": 1637 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001908824922673949, |
| "loss": 1.1918, |
| "step": 1638 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019087691921865858, |
| "loss": 0.8648, |
| "step": 1639 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019087134616992227, |
| "loss": 1.3553, |
| "step": 1640 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019086577312118596, |
| "loss": 0.9318, |
| "step": 1641 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019086020007244965, |
| "loss": 0.9082, |
| "step": 1642 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019085462702371332, |
| "loss": 1.1612, |
| "step": 1643 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.000190849053974977, |
| "loss": 1.2735, |
| "step": 1644 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001908434809262407, |
| "loss": 0.8829, |
| "step": 1645 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001908379078775044, |
| "loss": 1.1543, |
| "step": 1646 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019083233482876808, |
| "loss": 1.346, |
| "step": 1647 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019082676178003177, |
| "loss": 0.8767, |
| "step": 1648 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019082118873129547, |
| "loss": 1.0879, |
| "step": 1649 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019081561568255916, |
| "loss": 1.0069, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019081004263382285, |
| "loss": 1.2423, |
| "step": 1651 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019080446958508654, |
| "loss": 1.0051, |
| "step": 1652 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001907988965363502, |
| "loss": 1.1549, |
| "step": 1653 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019079332348761392, |
| "loss": 1.3945, |
| "step": 1654 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001907877504388776, |
| "loss": 1.1635, |
| "step": 1655 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019078217739014128, |
| "loss": 1.3012, |
| "step": 1656 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.000190776604341405, |
| "loss": 1.3518, |
| "step": 1657 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019077103129266866, |
| "loss": 1.2531, |
| "step": 1658 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019076545824393235, |
| "loss": 1.1135, |
| "step": 1659 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019075988519519604, |
| "loss": 1.2988, |
| "step": 1660 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019075431214645973, |
| "loss": 0.9057, |
| "step": 1661 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019074873909772342, |
| "loss": 0.9388, |
| "step": 1662 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001907431660489871, |
| "loss": 1.1183, |
| "step": 1663 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001907375930002508, |
| "loss": 1.1098, |
| "step": 1664 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001907320199515145, |
| "loss": 1.2464, |
| "step": 1665 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019072644690277816, |
| "loss": 1.1258, |
| "step": 1666 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019072087385404188, |
| "loss": 1.1922, |
| "step": 1667 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019071530080530554, |
| "loss": 1.1996, |
| "step": 1668 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019070972775656923, |
| "loss": 1.1848, |
| "step": 1669 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019070415470783293, |
| "loss": 1.0215, |
| "step": 1670 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019069858165909662, |
| "loss": 0.9944, |
| "step": 1671 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001906930086103603, |
| "loss": 1.0028, |
| "step": 1672 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.000190687435561624, |
| "loss": 1.0286, |
| "step": 1673 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001906818625128877, |
| "loss": 1.2025, |
| "step": 1674 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019067628946415138, |
| "loss": 1.1569, |
| "step": 1675 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019067071641541504, |
| "loss": 0.975, |
| "step": 1676 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019066514336667876, |
| "loss": 0.9151, |
| "step": 1677 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019065957031794243, |
| "loss": 1.1593, |
| "step": 1678 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019065399726920612, |
| "loss": 1.5012, |
| "step": 1679 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019064842422046984, |
| "loss": 1.0337, |
| "step": 1680 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001906428511717335, |
| "loss": 1.4288, |
| "step": 1681 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001906372781229972, |
| "loss": 1.6042, |
| "step": 1682 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019063170507426088, |
| "loss": 0.9243, |
| "step": 1683 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019062613202552457, |
| "loss": 0.9943, |
| "step": 1684 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019062055897678827, |
| "loss": 1.362, |
| "step": 1685 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019061498592805196, |
| "loss": 1.1406, |
| "step": 1686 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019060941287931565, |
| "loss": 1.1759, |
| "step": 1687 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001906038398305793, |
| "loss": 0.9714, |
| "step": 1688 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019059826678184303, |
| "loss": 0.9132, |
| "step": 1689 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019059269373310672, |
| "loss": 1.2707, |
| "step": 1690 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019058712068437038, |
| "loss": 1.2254, |
| "step": 1691 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019058154763563408, |
| "loss": 1.127, |
| "step": 1692 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019057597458689777, |
| "loss": 1.193, |
| "step": 1693 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019057040153816146, |
| "loss": 1.2829, |
| "step": 1694 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019056482848942515, |
| "loss": 1.1205, |
| "step": 1695 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019055925544068884, |
| "loss": 1.1202, |
| "step": 1696 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019055368239195253, |
| "loss": 0.8762, |
| "step": 1697 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001905481093432162, |
| "loss": 0.8542, |
| "step": 1698 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019054253629447991, |
| "loss": 1.0446, |
| "step": 1699 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001905369632457436, |
| "loss": 1.3055, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019053139019700727, |
| "loss": 0.9441, |
| "step": 1701 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.000190525817148271, |
| "loss": 1.3184, |
| "step": 1702 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019052024409953465, |
| "loss": 1.1562, |
| "step": 1703 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019051467105079834, |
| "loss": 0.9685, |
| "step": 1704 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019050909800206206, |
| "loss": 1.1422, |
| "step": 1705 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019050352495332572, |
| "loss": 1.2628, |
| "step": 1706 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019049795190458942, |
| "loss": 0.888, |
| "step": 1707 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019049237885585308, |
| "loss": 0.9997, |
| "step": 1708 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001904868058071168, |
| "loss": 1.1946, |
| "step": 1709 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001904812327583805, |
| "loss": 1.0221, |
| "step": 1710 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019047565970964415, |
| "loss": 1.1762, |
| "step": 1711 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019047008666090787, |
| "loss": 0.9687, |
| "step": 1712 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019046451361217154, |
| "loss": 1.2558, |
| "step": 1713 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019045894056343523, |
| "loss": 1.1072, |
| "step": 1714 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019045336751469895, |
| "loss": 1.2227, |
| "step": 1715 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001904477944659626, |
| "loss": 1.2399, |
| "step": 1716 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001904422214172263, |
| "loss": 0.9483, |
| "step": 1717 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019043664836849, |
| "loss": 1.1822, |
| "step": 1718 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019043107531975368, |
| "loss": 1.0953, |
| "step": 1719 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019042550227101737, |
| "loss": 1.2457, |
| "step": 1720 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019041992922228106, |
| "loss": 1.2141, |
| "step": 1721 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019041435617354476, |
| "loss": 0.9902, |
| "step": 1722 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019040878312480842, |
| "loss": 1.3767, |
| "step": 1723 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001904032100760721, |
| "loss": 1.0079, |
| "step": 1724 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019039763702733583, |
| "loss": 1.042, |
| "step": 1725 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001903920639785995, |
| "loss": 1.239, |
| "step": 1726 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019038649092986318, |
| "loss": 1.071, |
| "step": 1727 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019038091788112688, |
| "loss": 0.9213, |
| "step": 1728 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019037534483239057, |
| "loss": 0.9808, |
| "step": 1729 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019036977178365426, |
| "loss": 1.2428, |
| "step": 1730 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019036419873491795, |
| "loss": 1.2112, |
| "step": 1731 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019035862568618164, |
| "loss": 1.0475, |
| "step": 1732 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.0001903530526374453, |
| "loss": 0.8244, |
| "step": 1733 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 0.00019034747958870902, |
| "loss": 1.2624, |
| "step": 1734 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019034190653997271, |
| "loss": 0.8536, |
| "step": 1735 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019033633349123638, |
| "loss": 1.0855, |
| "step": 1736 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001903307604425001, |
| "loss": 1.1901, |
| "step": 1737 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019032518739376376, |
| "loss": 1.0323, |
| "step": 1738 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019031961434502745, |
| "loss": 1.3115, |
| "step": 1739 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019031404129629114, |
| "loss": 1.313, |
| "step": 1740 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019030846824755483, |
| "loss": 0.9892, |
| "step": 1741 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019030289519881852, |
| "loss": 1.1109, |
| "step": 1742 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019029732215008222, |
| "loss": 1.0249, |
| "step": 1743 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001902917491013459, |
| "loss": 1.3509, |
| "step": 1744 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001902861760526096, |
| "loss": 0.8917, |
| "step": 1745 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019028060300387326, |
| "loss": 1.2007, |
| "step": 1746 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019027502995513698, |
| "loss": 1.15, |
| "step": 1747 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019026945690640064, |
| "loss": 1.1333, |
| "step": 1748 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019026388385766434, |
| "loss": 1.2644, |
| "step": 1749 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019025831080892805, |
| "loss": 1.5704, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019025273776019172, |
| "loss": 1.2305, |
| "step": 1751 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001902471647114554, |
| "loss": 0.9593, |
| "step": 1752 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001902415916627191, |
| "loss": 1.1749, |
| "step": 1753 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001902360186139828, |
| "loss": 1.315, |
| "step": 1754 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019023044556524648, |
| "loss": 1.1195, |
| "step": 1755 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019022487251651015, |
| "loss": 1.1073, |
| "step": 1756 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019021929946777386, |
| "loss": 1.2327, |
| "step": 1757 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019021372641903753, |
| "loss": 1.1626, |
| "step": 1758 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019020815337030122, |
| "loss": 1.1473, |
| "step": 1759 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019020258032156494, |
| "loss": 1.3613, |
| "step": 1760 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001901970072728286, |
| "loss": 1.0124, |
| "step": 1761 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001901914342240923, |
| "loss": 1.2782, |
| "step": 1762 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019018586117535598, |
| "loss": 1.0324, |
| "step": 1763 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019018028812661968, |
| "loss": 0.9708, |
| "step": 1764 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019017471507788337, |
| "loss": 1.0052, |
| "step": 1765 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019016914202914706, |
| "loss": 1.1678, |
| "step": 1766 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019016356898041075, |
| "loss": 1.0013, |
| "step": 1767 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019015799593167444, |
| "loss": 1.331, |
| "step": 1768 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019015242288293813, |
| "loss": 1.0465, |
| "step": 1769 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019014684983420182, |
| "loss": 1.0917, |
| "step": 1770 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019014127678546549, |
| "loss": 1.2606, |
| "step": 1771 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019013570373672918, |
| "loss": 0.8727, |
| "step": 1772 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019013013068799287, |
| "loss": 1.0389, |
| "step": 1773 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019012455763925656, |
| "loss": 1.0539, |
| "step": 1774 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019011898459052025, |
| "loss": 1.0717, |
| "step": 1775 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019011341154178394, |
| "loss": 1.0984, |
| "step": 1776 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019010783849304763, |
| "loss": 1.0744, |
| "step": 1777 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019010226544431132, |
| "loss": 1.1332, |
| "step": 1778 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019009669239557502, |
| "loss": 0.9963, |
| "step": 1779 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001900911193468387, |
| "loss": 1.1677, |
| "step": 1780 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019008554629810237, |
| "loss": 1.0692, |
| "step": 1781 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001900799732493661, |
| "loss": 0.8546, |
| "step": 1782 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019007440020062975, |
| "loss": 1.1858, |
| "step": 1783 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019006882715189344, |
| "loss": 1.5141, |
| "step": 1784 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019006325410315716, |
| "loss": 0.9926, |
| "step": 1785 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019005768105442083, |
| "loss": 1.4305, |
| "step": 1786 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019005210800568452, |
| "loss": 1.2816, |
| "step": 1787 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001900465349569482, |
| "loss": 1.0206, |
| "step": 1788 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001900409619082119, |
| "loss": 1.2905, |
| "step": 1789 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001900353888594756, |
| "loss": 1.0182, |
| "step": 1790 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019002981581073925, |
| "loss": 1.2296, |
| "step": 1791 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019002424276200297, |
| "loss": 1.0762, |
| "step": 1792 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019001866971326664, |
| "loss": 0.9531, |
| "step": 1793 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019001309666453033, |
| "loss": 0.8755, |
| "step": 1794 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00019000752361579405, |
| "loss": 1.0594, |
| "step": 1795 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001900019505670577, |
| "loss": 1.1801, |
| "step": 1796 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001899963775183214, |
| "loss": 1.0239, |
| "step": 1797 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001899908044695851, |
| "loss": 1.1198, |
| "step": 1798 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018998523142084878, |
| "loss": 1.0988, |
| "step": 1799 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018997965837211247, |
| "loss": 1.0162, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.15, |
| "eval_loss": 1.120098352432251, |
| "eval_runtime": 329.0237, |
| "eval_samples_per_second": 7.598, |
| "eval_steps_per_second": 0.951, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018997408532337617, |
| "loss": 0.9643, |
| "step": 1801 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018996851227463986, |
| "loss": 1.0468, |
| "step": 1802 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018996293922590355, |
| "loss": 1.2109, |
| "step": 1803 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001899573661771672, |
| "loss": 1.1712, |
| "step": 1804 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018995179312843093, |
| "loss": 1.1016, |
| "step": 1805 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001899462200796946, |
| "loss": 1.1458, |
| "step": 1806 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018994064703095829, |
| "loss": 1.1772, |
| "step": 1807 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018993507398222198, |
| "loss": 1.4679, |
| "step": 1808 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018992950093348567, |
| "loss": 1.1333, |
| "step": 1809 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018992392788474936, |
| "loss": 1.3236, |
| "step": 1810 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018991835483601305, |
| "loss": 1.2844, |
| "step": 1811 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018991278178727674, |
| "loss": 1.031, |
| "step": 1812 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018990720873854043, |
| "loss": 1.0468, |
| "step": 1813 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018990163568980412, |
| "loss": 0.9621, |
| "step": 1814 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018989606264106781, |
| "loss": 1.3954, |
| "step": 1815 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018989048959233148, |
| "loss": 1.2072, |
| "step": 1816 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001898849165435952, |
| "loss": 0.9698, |
| "step": 1817 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018987934349485886, |
| "loss": 1.1436, |
| "step": 1818 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018987377044612255, |
| "loss": 1.2116, |
| "step": 1819 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018986819739738624, |
| "loss": 0.9974, |
| "step": 1820 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018986262434864993, |
| "loss": 1.1314, |
| "step": 1821 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018985705129991363, |
| "loss": 0.9028, |
| "step": 1822 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018985147825117732, |
| "loss": 1.4585, |
| "step": 1823 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.000189845905202441, |
| "loss": 1.3539, |
| "step": 1824 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001898403321537047, |
| "loss": 1.1488, |
| "step": 1825 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018983475910496836, |
| "loss": 0.988, |
| "step": 1826 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018982918605623208, |
| "loss": 1.3645, |
| "step": 1827 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018982361300749577, |
| "loss": 0.9478, |
| "step": 1828 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018981803995875944, |
| "loss": 1.3235, |
| "step": 1829 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018981246691002315, |
| "loss": 1.1297, |
| "step": 1830 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018980689386128682, |
| "loss": 1.2604, |
| "step": 1831 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001898013208125505, |
| "loss": 0.8137, |
| "step": 1832 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001897957477638142, |
| "loss": 1.5389, |
| "step": 1833 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001897901747150779, |
| "loss": 1.1322, |
| "step": 1834 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018978460166634158, |
| "loss": 1.1027, |
| "step": 1835 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018977902861760525, |
| "loss": 1.0587, |
| "step": 1836 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018977345556886897, |
| "loss": 1.4686, |
| "step": 1837 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018976788252013266, |
| "loss": 1.1603, |
| "step": 1838 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018976230947139632, |
| "loss": 1.0585, |
| "step": 1839 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018975673642266004, |
| "loss": 1.2631, |
| "step": 1840 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001897511633739237, |
| "loss": 1.1495, |
| "step": 1841 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001897455903251874, |
| "loss": 1.0983, |
| "step": 1842 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018974001727645109, |
| "loss": 1.1111, |
| "step": 1843 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018973444422771478, |
| "loss": 1.0397, |
| "step": 1844 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018972887117897847, |
| "loss": 0.8885, |
| "step": 1845 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018972329813024216, |
| "loss": 1.3004, |
| "step": 1846 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018971772508150585, |
| "loss": 0.9847, |
| "step": 1847 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018971215203276954, |
| "loss": 0.9886, |
| "step": 1848 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018970657898403323, |
| "loss": 1.1435, |
| "step": 1849 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018970100593529692, |
| "loss": 1.105, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001896954328865606, |
| "loss": 0.7416, |
| "step": 1851 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.0001896898598378243, |
| "loss": 1.1095, |
| "step": 1852 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.000189684286789088, |
| "loss": 1.3497, |
| "step": 1853 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 0.00018967871374035166, |
| "loss": 1.3613, |
| "step": 1854 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018967314069161535, |
| "loss": 1.1519, |
| "step": 1855 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018966756764287904, |
| "loss": 1.0983, |
| "step": 1856 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018966199459414273, |
| "loss": 1.0854, |
| "step": 1857 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018965642154540643, |
| "loss": 1.2412, |
| "step": 1858 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018965084849667012, |
| "loss": 1.4569, |
| "step": 1859 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001896452754479338, |
| "loss": 1.142, |
| "step": 1860 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018963970239919747, |
| "loss": 1.1713, |
| "step": 1861 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001896341293504612, |
| "loss": 1.3198, |
| "step": 1862 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018962855630172488, |
| "loss": 1.2895, |
| "step": 1863 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018962298325298855, |
| "loss": 1.1603, |
| "step": 1864 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018961741020425226, |
| "loss": 1.3046, |
| "step": 1865 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018961183715551593, |
| "loss": 1.1925, |
| "step": 1866 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018960626410677962, |
| "loss": 0.9829, |
| "step": 1867 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001896006910580433, |
| "loss": 1.0992, |
| "step": 1868 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.000189595118009307, |
| "loss": 0.9472, |
| "step": 1869 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001895895449605707, |
| "loss": 1.1177, |
| "step": 1870 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018958397191183436, |
| "loss": 0.8849, |
| "step": 1871 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018957839886309807, |
| "loss": 1.1186, |
| "step": 1872 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018957282581436177, |
| "loss": 1.0297, |
| "step": 1873 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018956725276562543, |
| "loss": 1.1363, |
| "step": 1874 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018956167971688915, |
| "loss": 1.1442, |
| "step": 1875 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001895561066681528, |
| "loss": 1.1572, |
| "step": 1876 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001895505336194165, |
| "loss": 1.1647, |
| "step": 1877 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018954496057068022, |
| "loss": 1.2869, |
| "step": 1878 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018953938752194389, |
| "loss": 1.0744, |
| "step": 1879 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018953381447320758, |
| "loss": 1.1294, |
| "step": 1880 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018952824142447127, |
| "loss": 1.1241, |
| "step": 1881 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018952266837573496, |
| "loss": 0.9782, |
| "step": 1882 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018951709532699865, |
| "loss": 0.9557, |
| "step": 1883 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018951152227826234, |
| "loss": 1.0299, |
| "step": 1884 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018950594922952603, |
| "loss": 0.9555, |
| "step": 1885 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001895003761807897, |
| "loss": 1.0213, |
| "step": 1886 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001894948031320534, |
| "loss": 1.0628, |
| "step": 1887 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001894892300833171, |
| "loss": 1.0816, |
| "step": 1888 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018948365703458077, |
| "loss": 1.1352, |
| "step": 1889 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018947808398584446, |
| "loss": 1.0898, |
| "step": 1890 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018947251093710815, |
| "loss": 1.1472, |
| "step": 1891 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018946693788837184, |
| "loss": 1.1328, |
| "step": 1892 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018946136483963553, |
| "loss": 0.8944, |
| "step": 1893 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018945579179089923, |
| "loss": 1.2956, |
| "step": 1894 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018945021874216292, |
| "loss": 1.1021, |
| "step": 1895 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018944464569342658, |
| "loss": 0.9959, |
| "step": 1896 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001894390726446903, |
| "loss": 0.8886, |
| "step": 1897 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.000189433499595954, |
| "loss": 1.2343, |
| "step": 1898 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018942792654721765, |
| "loss": 1.2677, |
| "step": 1899 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018942235349848137, |
| "loss": 1.0745, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018941678044974504, |
| "loss": 1.0638, |
| "step": 1901 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018941120740100873, |
| "loss": 1.0753, |
| "step": 1902 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018940563435227242, |
| "loss": 0.9902, |
| "step": 1903 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001894000613035361, |
| "loss": 1.1666, |
| "step": 1904 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001893944882547998, |
| "loss": 0.8571, |
| "step": 1905 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018938891520606346, |
| "loss": 1.083, |
| "step": 1906 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018938334215732718, |
| "loss": 1.0729, |
| "step": 1907 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018937776910859087, |
| "loss": 1.209, |
| "step": 1908 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018937219605985454, |
| "loss": 1.236, |
| "step": 1909 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018936662301111826, |
| "loss": 0.9337, |
| "step": 1910 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018936104996238192, |
| "loss": 1.3111, |
| "step": 1911 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001893554769136456, |
| "loss": 1.0945, |
| "step": 1912 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018934990386490933, |
| "loss": 1.5764, |
| "step": 1913 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.000189344330816173, |
| "loss": 0.9615, |
| "step": 1914 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018933875776743668, |
| "loss": 0.9532, |
| "step": 1915 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018933318471870038, |
| "loss": 1.0962, |
| "step": 1916 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018932761166996407, |
| "loss": 1.1104, |
| "step": 1917 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018932203862122776, |
| "loss": 1.0819, |
| "step": 1918 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018931646557249142, |
| "loss": 1.0329, |
| "step": 1919 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018931089252375514, |
| "loss": 1.2034, |
| "step": 1920 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001893053194750188, |
| "loss": 1.1555, |
| "step": 1921 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001892997464262825, |
| "loss": 1.4493, |
| "step": 1922 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018929417337754621, |
| "loss": 1.1017, |
| "step": 1923 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018928860032880988, |
| "loss": 1.0149, |
| "step": 1924 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018928302728007357, |
| "loss": 1.2676, |
| "step": 1925 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018927745423133726, |
| "loss": 0.8906, |
| "step": 1926 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018927188118260095, |
| "loss": 1.4849, |
| "step": 1927 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018926630813386464, |
| "loss": 1.177, |
| "step": 1928 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018926073508512833, |
| "loss": 1.0367, |
| "step": 1929 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018925516203639202, |
| "loss": 1.0633, |
| "step": 1930 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001892495889876557, |
| "loss": 1.2042, |
| "step": 1931 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001892440159389194, |
| "loss": 1.0476, |
| "step": 1932 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001892384428901831, |
| "loss": 1.283, |
| "step": 1933 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018923286984144676, |
| "loss": 1.0968, |
| "step": 1934 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018922729679271045, |
| "loss": 1.0734, |
| "step": 1935 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018922172374397414, |
| "loss": 1.1556, |
| "step": 1936 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018921615069523784, |
| "loss": 0.6505, |
| "step": 1937 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018921057764650153, |
| "loss": 1.1878, |
| "step": 1938 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018920500459776522, |
| "loss": 1.1115, |
| "step": 1939 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001891994315490289, |
| "loss": 1.0704, |
| "step": 1940 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001891938585002926, |
| "loss": 1.091, |
| "step": 1941 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001891882854515563, |
| "loss": 0.9347, |
| "step": 1942 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018918271240281998, |
| "loss": 0.9724, |
| "step": 1943 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018917713935408365, |
| "loss": 1.1398, |
| "step": 1944 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018917156630534736, |
| "loss": 1.1776, |
| "step": 1945 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018916599325661103, |
| "loss": 1.0045, |
| "step": 1946 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018916042020787472, |
| "loss": 1.1009, |
| "step": 1947 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018915484715913844, |
| "loss": 1.5092, |
| "step": 1948 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001891492741104021, |
| "loss": 1.1293, |
| "step": 1949 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001891437010616658, |
| "loss": 0.9871, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018913812801292948, |
| "loss": 1.2316, |
| "step": 1951 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018913255496419318, |
| "loss": 1.2568, |
| "step": 1952 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018912698191545687, |
| "loss": 1.0275, |
| "step": 1953 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018912140886672053, |
| "loss": 1.1176, |
| "step": 1954 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018911583581798425, |
| "loss": 0.8576, |
| "step": 1955 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001891102627692479, |
| "loss": 0.7834, |
| "step": 1956 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001891046897205116, |
| "loss": 1.1402, |
| "step": 1957 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018909911667177532, |
| "loss": 1.1685, |
| "step": 1958 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.000189093543623039, |
| "loss": 0.9923, |
| "step": 1959 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018908797057430268, |
| "loss": 1.2892, |
| "step": 1960 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018908239752556637, |
| "loss": 0.9548, |
| "step": 1961 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018907682447683006, |
| "loss": 0.9242, |
| "step": 1962 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018907125142809375, |
| "loss": 1.3084, |
| "step": 1963 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018906567837935744, |
| "loss": 0.6558, |
| "step": 1964 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018906010533062113, |
| "loss": 1.0064, |
| "step": 1965 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001890545322818848, |
| "loss": 0.9914, |
| "step": 1966 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001890489592331485, |
| "loss": 0.9741, |
| "step": 1967 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.0001890433861844122, |
| "loss": 0.9185, |
| "step": 1968 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018903781313567587, |
| "loss": 1.3875, |
| "step": 1969 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018903224008693956, |
| "loss": 1.1394, |
| "step": 1970 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018902666703820325, |
| "loss": 1.088, |
| "step": 1971 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018902109398946694, |
| "loss": 1.1448, |
| "step": 1972 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00018901552094073064, |
| "loss": 0.9273, |
| "step": 1973 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018900994789199433, |
| "loss": 1.0825, |
| "step": 1974 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018900437484325802, |
| "loss": 1.3795, |
| "step": 1975 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001889988017945217, |
| "loss": 1.0225, |
| "step": 1976 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001889932287457854, |
| "loss": 0.9241, |
| "step": 1977 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001889876556970491, |
| "loss": 0.9494, |
| "step": 1978 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018898208264831276, |
| "loss": 1.1095, |
| "step": 1979 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018897650959957647, |
| "loss": 0.9964, |
| "step": 1980 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018897093655084014, |
| "loss": 1.2874, |
| "step": 1981 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018896536350210383, |
| "loss": 0.9136, |
| "step": 1982 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018895979045336752, |
| "loss": 0.9909, |
| "step": 1983 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001889542174046312, |
| "loss": 0.9906, |
| "step": 1984 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001889486443558949, |
| "loss": 1.2102, |
| "step": 1985 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001889430713071586, |
| "loss": 1.1307, |
| "step": 1986 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018893749825842228, |
| "loss": 1.2534, |
| "step": 1987 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018893192520968598, |
| "loss": 1.0502, |
| "step": 1988 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018892635216094964, |
| "loss": 1.0711, |
| "step": 1989 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018892077911221336, |
| "loss": 1.0575, |
| "step": 1990 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018891520606347702, |
| "loss": 1.047, |
| "step": 1991 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001889096330147407, |
| "loss": 1.2349, |
| "step": 1992 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018890405996600443, |
| "loss": 1.113, |
| "step": 1993 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001888984869172681, |
| "loss": 1.2311, |
| "step": 1994 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018889291386853179, |
| "loss": 1.2178, |
| "step": 1995 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018888734081979548, |
| "loss": 1.2079, |
| "step": 1996 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018888176777105917, |
| "loss": 0.9295, |
| "step": 1997 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018887619472232286, |
| "loss": 1.1131, |
| "step": 1998 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018887062167358652, |
| "loss": 1.1866, |
| "step": 1999 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018886504862485024, |
| "loss": 1.0908, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.17, |
| "eval_loss": 1.108474612236023, |
| "eval_runtime": 334.9214, |
| "eval_samples_per_second": 7.464, |
| "eval_steps_per_second": 0.935, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018885947557611393, |
| "loss": 1.1847, |
| "step": 2001 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001888539025273776, |
| "loss": 1.1325, |
| "step": 2002 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018884832947864132, |
| "loss": 0.8508, |
| "step": 2003 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018884275642990498, |
| "loss": 0.9606, |
| "step": 2004 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018883718338116867, |
| "loss": 0.8268, |
| "step": 2005 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018883161033243236, |
| "loss": 1.3006, |
| "step": 2006 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018882603728369605, |
| "loss": 1.2497, |
| "step": 2007 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018882046423495974, |
| "loss": 1.0249, |
| "step": 2008 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018881489118622343, |
| "loss": 1.0168, |
| "step": 2009 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018880931813748713, |
| "loss": 1.3279, |
| "step": 2010 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018880374508875082, |
| "loss": 1.015, |
| "step": 2011 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001887981720400145, |
| "loss": 0.7831, |
| "step": 2012 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001887925989912782, |
| "loss": 1.0084, |
| "step": 2013 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018878702594254186, |
| "loss": 1.2162, |
| "step": 2014 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018878145289380555, |
| "loss": 0.9519, |
| "step": 2015 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018877587984506925, |
| "loss": 1.0328, |
| "step": 2016 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018877030679633294, |
| "loss": 1.137, |
| "step": 2017 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018876473374759663, |
| "loss": 0.9451, |
| "step": 2018 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018875916069886032, |
| "loss": 1.302, |
| "step": 2019 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.000188753587650124, |
| "loss": 0.9089, |
| "step": 2020 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001887480146013877, |
| "loss": 1.1114, |
| "step": 2021 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001887424415526514, |
| "loss": 0.917, |
| "step": 2022 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018873686850391508, |
| "loss": 1.0666, |
| "step": 2023 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018873129545517875, |
| "loss": 1.052, |
| "step": 2024 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018872572240644247, |
| "loss": 1.1134, |
| "step": 2025 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018872014935770616, |
| "loss": 1.304, |
| "step": 2026 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018871457630896982, |
| "loss": 1.091, |
| "step": 2027 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018870900326023354, |
| "loss": 0.9522, |
| "step": 2028 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001887034302114972, |
| "loss": 1.1781, |
| "step": 2029 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001886978571627609, |
| "loss": 1.1429, |
| "step": 2030 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018869228411402459, |
| "loss": 1.169, |
| "step": 2031 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018868671106528828, |
| "loss": 1.0307, |
| "step": 2032 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018868113801655197, |
| "loss": 1.1688, |
| "step": 2033 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018867556496781563, |
| "loss": 1.1345, |
| "step": 2034 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018866999191907935, |
| "loss": 0.9354, |
| "step": 2035 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018866441887034304, |
| "loss": 1.22, |
| "step": 2036 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001886588458216067, |
| "loss": 1.0848, |
| "step": 2037 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018865327277287042, |
| "loss": 1.297, |
| "step": 2038 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001886476997241341, |
| "loss": 1.174, |
| "step": 2039 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018864212667539778, |
| "loss": 0.9247, |
| "step": 2040 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018863655362666147, |
| "loss": 1.0724, |
| "step": 2041 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018863098057792516, |
| "loss": 1.1957, |
| "step": 2042 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018862540752918885, |
| "loss": 0.8334, |
| "step": 2043 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018861983448045254, |
| "loss": 1.0262, |
| "step": 2044 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018861426143171623, |
| "loss": 1.1988, |
| "step": 2045 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018860868838297993, |
| "loss": 1.0379, |
| "step": 2046 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001886031153342436, |
| "loss": 1.0814, |
| "step": 2047 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001885975422855073, |
| "loss": 1.0915, |
| "step": 2048 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018859196923677097, |
| "loss": 1.0402, |
| "step": 2049 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018858639618803466, |
| "loss": 1.0567, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018858082313929838, |
| "loss": 1.1116, |
| "step": 2051 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018857525009056205, |
| "loss": 1.3408, |
| "step": 2052 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018856967704182574, |
| "loss": 1.0748, |
| "step": 2053 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018856410399308943, |
| "loss": 1.0113, |
| "step": 2054 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018855853094435312, |
| "loss": 1.0389, |
| "step": 2055 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001885529578956168, |
| "loss": 1.1175, |
| "step": 2056 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001885473848468805, |
| "loss": 1.0845, |
| "step": 2057 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001885418117981442, |
| "loss": 0.9602, |
| "step": 2058 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018853623874940786, |
| "loss": 1.2448, |
| "step": 2059 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018853066570067157, |
| "loss": 1.2692, |
| "step": 2060 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018852509265193527, |
| "loss": 1.2293, |
| "step": 2061 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018851951960319893, |
| "loss": 1.1429, |
| "step": 2062 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018851394655446262, |
| "loss": 1.1031, |
| "step": 2063 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001885083735057263, |
| "loss": 0.897, |
| "step": 2064 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018850280045699, |
| "loss": 1.1297, |
| "step": 2065 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001884972274082537, |
| "loss": 0.9605, |
| "step": 2066 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018849165435951739, |
| "loss": 1.2896, |
| "step": 2067 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018848608131078108, |
| "loss": 0.8957, |
| "step": 2068 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018848050826204474, |
| "loss": 1.3459, |
| "step": 2069 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018847493521330846, |
| "loss": 1.1882, |
| "step": 2070 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018846936216457215, |
| "loss": 1.1409, |
| "step": 2071 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018846378911583581, |
| "loss": 1.4032, |
| "step": 2072 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018845821606709953, |
| "loss": 1.2442, |
| "step": 2073 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001884526430183632, |
| "loss": 1.1037, |
| "step": 2074 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001884470699696269, |
| "loss": 0.8906, |
| "step": 2075 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001884414969208906, |
| "loss": 0.9617, |
| "step": 2076 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018843592387215427, |
| "loss": 1.0539, |
| "step": 2077 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018843035082341796, |
| "loss": 1.4232, |
| "step": 2078 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018842477777468162, |
| "loss": 1.0767, |
| "step": 2079 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018841920472594534, |
| "loss": 1.0789, |
| "step": 2080 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018841363167720903, |
| "loss": 1.0517, |
| "step": 2081 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001884080586284727, |
| "loss": 1.421, |
| "step": 2082 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018840248557973642, |
| "loss": 0.9427, |
| "step": 2083 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018839691253100008, |
| "loss": 1.3959, |
| "step": 2084 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018839133948226377, |
| "loss": 0.835, |
| "step": 2085 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001883857664335275, |
| "loss": 1.001, |
| "step": 2086 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018838019338479115, |
| "loss": 1.2754, |
| "step": 2087 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018837462033605485, |
| "loss": 0.9016, |
| "step": 2088 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018836904728731854, |
| "loss": 1.0898, |
| "step": 2089 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018836347423858223, |
| "loss": 1.0918, |
| "step": 2090 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.00018835790118984592, |
| "loss": 1.1316, |
| "step": 2091 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001883523281411096, |
| "loss": 0.9886, |
| "step": 2092 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 0.0001883467550923733, |
| "loss": 0.7953, |
| "step": 2093 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018834118204363696, |
| "loss": 0.999, |
| "step": 2094 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018833560899490066, |
| "loss": 1.2317, |
| "step": 2095 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018833003594616437, |
| "loss": 1.068, |
| "step": 2096 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018832446289742804, |
| "loss": 1.0996, |
| "step": 2097 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018831888984869173, |
| "loss": 1.2349, |
| "step": 2098 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018831331679995542, |
| "loss": 0.9169, |
| "step": 2099 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001883077437512191, |
| "loss": 0.9618, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001883021707024828, |
| "loss": 1.1364, |
| "step": 2101 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001882965976537465, |
| "loss": 1.3388, |
| "step": 2102 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018829102460501019, |
| "loss": 1.2156, |
| "step": 2103 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018828545155627385, |
| "loss": 0.9201, |
| "step": 2104 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018827987850753757, |
| "loss": 1.0079, |
| "step": 2105 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018827430545880126, |
| "loss": 1.0476, |
| "step": 2106 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018826873241006492, |
| "loss": 1.1937, |
| "step": 2107 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018826315936132864, |
| "loss": 1.1291, |
| "step": 2108 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001882575863125923, |
| "loss": 1.1304, |
| "step": 2109 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.000188252013263856, |
| "loss": 1.3582, |
| "step": 2110 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001882464402151197, |
| "loss": 0.9739, |
| "step": 2111 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018824086716638338, |
| "loss": 0.9111, |
| "step": 2112 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018823529411764707, |
| "loss": 0.7705, |
| "step": 2113 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018822972106891076, |
| "loss": 1.2338, |
| "step": 2114 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018822414802017445, |
| "loss": 1.1824, |
| "step": 2115 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018821857497143814, |
| "loss": 1.2548, |
| "step": 2116 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001882130019227018, |
| "loss": 1.2376, |
| "step": 2117 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018820742887396553, |
| "loss": 1.1423, |
| "step": 2118 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001882018558252292, |
| "loss": 1.1186, |
| "step": 2119 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018819628277649288, |
| "loss": 1.3563, |
| "step": 2120 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001881907097277566, |
| "loss": 1.205, |
| "step": 2121 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018818513667902026, |
| "loss": 1.4656, |
| "step": 2122 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018817956363028395, |
| "loss": 1.0459, |
| "step": 2123 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018817399058154764, |
| "loss": 1.158, |
| "step": 2124 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018816841753281134, |
| "loss": 1.0315, |
| "step": 2125 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018816284448407503, |
| "loss": 1.2104, |
| "step": 2126 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001881572714353387, |
| "loss": 1.1108, |
| "step": 2127 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001881516983866024, |
| "loss": 1.0705, |
| "step": 2128 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018814612533786607, |
| "loss": 1.1055, |
| "step": 2129 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018814055228912976, |
| "loss": 0.9544, |
| "step": 2130 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018813497924039348, |
| "loss": 0.8223, |
| "step": 2131 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018812940619165715, |
| "loss": 1.1472, |
| "step": 2132 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018812383314292084, |
| "loss": 1.0918, |
| "step": 2133 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018811826009418453, |
| "loss": 1.0278, |
| "step": 2134 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018811268704544822, |
| "loss": 1.1657, |
| "step": 2135 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001881071139967119, |
| "loss": 0.8155, |
| "step": 2136 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001881015409479756, |
| "loss": 1.2509, |
| "step": 2137 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001880959678992393, |
| "loss": 1.2268, |
| "step": 2138 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018809039485050296, |
| "loss": 1.2848, |
| "step": 2139 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018808482180176668, |
| "loss": 0.8424, |
| "step": 2140 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018807924875303037, |
| "loss": 0.8501, |
| "step": 2141 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018807367570429403, |
| "loss": 1.0142, |
| "step": 2142 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018806810265555772, |
| "loss": 1.0092, |
| "step": 2143 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001880625296068214, |
| "loss": 1.3011, |
| "step": 2144 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001880569565580851, |
| "loss": 0.9019, |
| "step": 2145 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001880513835093488, |
| "loss": 1.2172, |
| "step": 2146 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001880458104606125, |
| "loss": 1.0387, |
| "step": 2147 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018804023741187618, |
| "loss": 1.1656, |
| "step": 2148 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018803466436313987, |
| "loss": 1.0258, |
| "step": 2149 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018802909131440356, |
| "loss": 1.0245, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018802351826566725, |
| "loss": 0.802, |
| "step": 2151 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018801794521693092, |
| "loss": 1.2857, |
| "step": 2152 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018801237216819463, |
| "loss": 1.0113, |
| "step": 2153 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001880067991194583, |
| "loss": 1.1677, |
| "step": 2154 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.000188001226070722, |
| "loss": 1.054, |
| "step": 2155 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001879956530219857, |
| "loss": 1.1631, |
| "step": 2156 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018799007997324937, |
| "loss": 1.3178, |
| "step": 2157 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018798450692451306, |
| "loss": 1.0313, |
| "step": 2158 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018797893387577675, |
| "loss": 0.9866, |
| "step": 2159 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018797336082704044, |
| "loss": 0.7883, |
| "step": 2160 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018796778777830414, |
| "loss": 0.93, |
| "step": 2161 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001879622147295678, |
| "loss": 1.2242, |
| "step": 2162 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018795664168083152, |
| "loss": 1.089, |
| "step": 2163 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018795106863209518, |
| "loss": 1.22, |
| "step": 2164 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018794549558335887, |
| "loss": 1.0789, |
| "step": 2165 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001879399225346226, |
| "loss": 1.2845, |
| "step": 2166 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018793434948588626, |
| "loss": 0.9508, |
| "step": 2167 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018792877643714995, |
| "loss": 1.0546, |
| "step": 2168 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018792320338841364, |
| "loss": 0.9505, |
| "step": 2169 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018791763033967733, |
| "loss": 0.8882, |
| "step": 2170 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018791205729094102, |
| "loss": 1.4098, |
| "step": 2171 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001879064842422047, |
| "loss": 1.0278, |
| "step": 2172 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001879009111934684, |
| "loss": 0.9822, |
| "step": 2173 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001878953381447321, |
| "loss": 1.0687, |
| "step": 2174 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018788976509599578, |
| "loss": 1.1814, |
| "step": 2175 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018788419204725948, |
| "loss": 1.0658, |
| "step": 2176 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018787861899852314, |
| "loss": 0.9306, |
| "step": 2177 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018787304594978683, |
| "loss": 0.9401, |
| "step": 2178 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018786747290105052, |
| "loss": 0.9429, |
| "step": 2179 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001878618998523142, |
| "loss": 0.9031, |
| "step": 2180 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001878563268035779, |
| "loss": 1.3116, |
| "step": 2181 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001878507537548416, |
| "loss": 1.306, |
| "step": 2182 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001878451807061053, |
| "loss": 0.8248, |
| "step": 2183 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018783960765736898, |
| "loss": 1.4121, |
| "step": 2184 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018783403460863267, |
| "loss": 1.0562, |
| "step": 2185 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018782846155989636, |
| "loss": 1.3134, |
| "step": 2186 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018782288851116002, |
| "loss": 1.0375, |
| "step": 2187 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018781731546242374, |
| "loss": 1.101, |
| "step": 2188 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001878117424136874, |
| "loss": 1.0619, |
| "step": 2189 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001878061693649511, |
| "loss": 0.9668, |
| "step": 2190 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018780059631621482, |
| "loss": 1.1236, |
| "step": 2191 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018779502326747848, |
| "loss": 1.1063, |
| "step": 2192 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018778945021874217, |
| "loss": 0.9921, |
| "step": 2193 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018778387717000586, |
| "loss": 1.0504, |
| "step": 2194 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018777830412126955, |
| "loss": 1.13, |
| "step": 2195 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018777273107253324, |
| "loss": 0.8079, |
| "step": 2196 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001877671580237969, |
| "loss": 1.2888, |
| "step": 2197 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018776158497506063, |
| "loss": 1.0939, |
| "step": 2198 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018775601192632432, |
| "loss": 0.9232, |
| "step": 2199 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018775043887758798, |
| "loss": 0.9667, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.18, |
| "eval_loss": 1.0998486280441284, |
| "eval_runtime": 334.2178, |
| "eval_samples_per_second": 7.48, |
| "eval_steps_per_second": 0.937, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001877448658288517, |
| "loss": 0.7618, |
| "step": 2201 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018773929278011536, |
| "loss": 0.8497, |
| "step": 2202 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018773371973137906, |
| "loss": 0.9712, |
| "step": 2203 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018772814668264275, |
| "loss": 1.0125, |
| "step": 2204 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018772257363390644, |
| "loss": 1.2332, |
| "step": 2205 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018771700058517013, |
| "loss": 1.0497, |
| "step": 2206 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018771142753643382, |
| "loss": 1.158, |
| "step": 2207 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001877058544876975, |
| "loss": 0.8751, |
| "step": 2208 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.0001877002814389612, |
| "loss": 0.9566, |
| "step": 2209 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018769470839022487, |
| "loss": 1.1624, |
| "step": 2210 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018768913534148858, |
| "loss": 0.9773, |
| "step": 2211 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018768356229275225, |
| "loss": 1.2221, |
| "step": 2212 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 0.00018767798924401594, |
| "loss": 0.9729, |
| "step": 2213 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018767241619527963, |
| "loss": 1.2627, |
| "step": 2214 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018766684314654332, |
| "loss": 1.113, |
| "step": 2215 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.000187661270097807, |
| "loss": 0.8016, |
| "step": 2216 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001876556970490707, |
| "loss": 1.1018, |
| "step": 2217 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001876501240003344, |
| "loss": 1.0178, |
| "step": 2218 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018764455095159809, |
| "loss": 1.1801, |
| "step": 2219 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018763897790286178, |
| "loss": 1.3784, |
| "step": 2220 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018763340485412547, |
| "loss": 1.0287, |
| "step": 2221 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018762783180538913, |
| "loss": 0.8412, |
| "step": 2222 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018762225875665285, |
| "loss": 1.4055, |
| "step": 2223 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018761668570791654, |
| "loss": 1.1047, |
| "step": 2224 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001876111126591802, |
| "loss": 1.0691, |
| "step": 2225 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001876055396104439, |
| "loss": 1.2944, |
| "step": 2226 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001875999665617076, |
| "loss": 1.0358, |
| "step": 2227 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018759439351297128, |
| "loss": 1.1313, |
| "step": 2228 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018758882046423497, |
| "loss": 1.0325, |
| "step": 2229 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018758324741549866, |
| "loss": 1.0233, |
| "step": 2230 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018757767436676235, |
| "loss": 1.0011, |
| "step": 2231 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018757210131802602, |
| "loss": 0.9339, |
| "step": 2232 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018756652826928973, |
| "loss": 1.016, |
| "step": 2233 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018756095522055343, |
| "loss": 0.7531, |
| "step": 2234 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001875553821718171, |
| "loss": 1.106, |
| "step": 2235 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001875498091230808, |
| "loss": 1.4523, |
| "step": 2236 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018754423607434447, |
| "loss": 1.2235, |
| "step": 2237 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018753866302560816, |
| "loss": 1.0472, |
| "step": 2238 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018753308997687185, |
| "loss": 1.1656, |
| "step": 2239 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018752751692813555, |
| "loss": 1.2238, |
| "step": 2240 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018752194387939924, |
| "loss": 1.3886, |
| "step": 2241 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001875163708306629, |
| "loss": 1.1737, |
| "step": 2242 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018751079778192662, |
| "loss": 1.2153, |
| "step": 2243 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001875052247331903, |
| "loss": 1.2596, |
| "step": 2244 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018749965168445397, |
| "loss": 0.9277, |
| "step": 2245 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001874940786357177, |
| "loss": 0.9458, |
| "step": 2246 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018748850558698136, |
| "loss": 1.0449, |
| "step": 2247 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018748293253824505, |
| "loss": 0.8574, |
| "step": 2248 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018747735948950877, |
| "loss": 1.2902, |
| "step": 2249 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018747178644077243, |
| "loss": 1.2165, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018746621339203612, |
| "loss": 1.1342, |
| "step": 2251 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001874606403432998, |
| "loss": 1.1366, |
| "step": 2252 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001874550672945635, |
| "loss": 1.0504, |
| "step": 2253 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001874494942458272, |
| "loss": 1.184, |
| "step": 2254 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018744392119709089, |
| "loss": 0.9549, |
| "step": 2255 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018743834814835458, |
| "loss": 0.8924, |
| "step": 2256 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018743277509961824, |
| "loss": 1.0921, |
| "step": 2257 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018742720205088193, |
| "loss": 0.6985, |
| "step": 2258 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018742162900214565, |
| "loss": 0.9422, |
| "step": 2259 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018741605595340931, |
| "loss": 0.9877, |
| "step": 2260 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.000187410482904673, |
| "loss": 1.0604, |
| "step": 2261 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001874049098559367, |
| "loss": 0.9651, |
| "step": 2262 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001873993368072004, |
| "loss": 1.057, |
| "step": 2263 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018739376375846408, |
| "loss": 1.2014, |
| "step": 2264 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018738819070972777, |
| "loss": 1.322, |
| "step": 2265 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018738261766099146, |
| "loss": 1.1933, |
| "step": 2266 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018737704461225513, |
| "loss": 1.0775, |
| "step": 2267 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018737147156351884, |
| "loss": 1.0926, |
| "step": 2268 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018736589851478253, |
| "loss": 0.8779, |
| "step": 2269 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001873603254660462, |
| "loss": 0.8844, |
| "step": 2270 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018735475241730992, |
| "loss": 0.9346, |
| "step": 2271 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018734917936857358, |
| "loss": 1.0933, |
| "step": 2272 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018734360631983727, |
| "loss": 1.327, |
| "step": 2273 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018733803327110096, |
| "loss": 1.2275, |
| "step": 2274 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018733246022236465, |
| "loss": 0.9216, |
| "step": 2275 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018732688717362835, |
| "loss": 0.7106, |
| "step": 2276 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.000187321314124892, |
| "loss": 1.0305, |
| "step": 2277 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018731574107615573, |
| "loss": 1.3372, |
| "step": 2278 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018731016802741942, |
| "loss": 1.2397, |
| "step": 2279 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018730459497868308, |
| "loss": 1.194, |
| "step": 2280 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001872990219299468, |
| "loss": 1.3194, |
| "step": 2281 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018729344888121047, |
| "loss": 1.128, |
| "step": 2282 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018728787583247416, |
| "loss": 1.2052, |
| "step": 2283 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018728230278373787, |
| "loss": 1.23, |
| "step": 2284 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018727672973500154, |
| "loss": 1.1783, |
| "step": 2285 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018727115668626523, |
| "loss": 0.7994, |
| "step": 2286 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018726558363752892, |
| "loss": 1.1027, |
| "step": 2287 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001872600105887926, |
| "loss": 1.3199, |
| "step": 2288 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001872544375400563, |
| "loss": 0.9701, |
| "step": 2289 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018724886449131997, |
| "loss": 1.2679, |
| "step": 2290 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018724329144258369, |
| "loss": 1.1202, |
| "step": 2291 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018723771839384735, |
| "loss": 0.9204, |
| "step": 2292 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018723214534511104, |
| "loss": 1.1603, |
| "step": 2293 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018722657229637476, |
| "loss": 1.043, |
| "step": 2294 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018722099924763842, |
| "loss": 1.117, |
| "step": 2295 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018721542619890211, |
| "loss": 1.3688, |
| "step": 2296 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001872098531501658, |
| "loss": 1.0335, |
| "step": 2297 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001872042801014295, |
| "loss": 1.0506, |
| "step": 2298 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001871987070526932, |
| "loss": 1.071, |
| "step": 2299 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018719313400395688, |
| "loss": 1.1846, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018718756095522057, |
| "loss": 1.0655, |
| "step": 2301 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018718198790648423, |
| "loss": 1.4032, |
| "step": 2302 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018717641485774795, |
| "loss": 0.983, |
| "step": 2303 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018717084180901164, |
| "loss": 1.1841, |
| "step": 2304 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001871652687602753, |
| "loss": 1.1848, |
| "step": 2305 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.000187159695711539, |
| "loss": 0.9615, |
| "step": 2306 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001871541226628027, |
| "loss": 1.0203, |
| "step": 2307 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018714854961406638, |
| "loss": 1.3153, |
| "step": 2308 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018714297656533007, |
| "loss": 1.128, |
| "step": 2309 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018713740351659376, |
| "loss": 1.0705, |
| "step": 2310 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018713183046785745, |
| "loss": 1.0897, |
| "step": 2311 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018712625741912112, |
| "loss": 0.8948, |
| "step": 2312 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018712068437038484, |
| "loss": 1.1784, |
| "step": 2313 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018711511132164853, |
| "loss": 1.0466, |
| "step": 2314 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001871095382729122, |
| "loss": 1.1619, |
| "step": 2315 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001871039652241759, |
| "loss": 1.2688, |
| "step": 2316 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018709839217543957, |
| "loss": 1.8173, |
| "step": 2317 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018709281912670326, |
| "loss": 1.0572, |
| "step": 2318 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018708724607796698, |
| "loss": 1.1365, |
| "step": 2319 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018708167302923065, |
| "loss": 1.1554, |
| "step": 2320 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018707609998049434, |
| "loss": 1.0234, |
| "step": 2321 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018707052693175803, |
| "loss": 0.9383, |
| "step": 2322 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018706495388302172, |
| "loss": 1.0396, |
| "step": 2323 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001870593808342854, |
| "loss": 1.1462, |
| "step": 2324 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018705380778554908, |
| "loss": 0.8343, |
| "step": 2325 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.0001870482347368128, |
| "loss": 1.1992, |
| "step": 2326 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018704266168807646, |
| "loss": 0.9566, |
| "step": 2327 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018703708863934015, |
| "loss": 0.8896, |
| "step": 2328 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018703151559060387, |
| "loss": 0.9351, |
| "step": 2329 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018702594254186753, |
| "loss": 1.1172, |
| "step": 2330 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018702036949313122, |
| "loss": 1.1364, |
| "step": 2331 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 0.00018701479644439491, |
| "loss": 0.9794, |
| "step": 2332 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001870092233956586, |
| "loss": 0.938, |
| "step": 2333 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001870036503469223, |
| "loss": 0.9805, |
| "step": 2334 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.000186998077298186, |
| "loss": 1.22, |
| "step": 2335 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018699250424944968, |
| "loss": 1.2233, |
| "step": 2336 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018698693120071334, |
| "loss": 1.2426, |
| "step": 2337 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018698135815197703, |
| "loss": 0.9961, |
| "step": 2338 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018697578510324075, |
| "loss": 0.7817, |
| "step": 2339 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018697021205450442, |
| "loss": 1.073, |
| "step": 2340 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001869646390057681, |
| "loss": 1.078, |
| "step": 2341 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001869590659570318, |
| "loss": 1.0, |
| "step": 2342 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001869534929082955, |
| "loss": 1.2698, |
| "step": 2343 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018694791985955918, |
| "loss": 0.9912, |
| "step": 2344 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018694234681082287, |
| "loss": 0.9149, |
| "step": 2345 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018693677376208656, |
| "loss": 1.0641, |
| "step": 2346 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018693120071335025, |
| "loss": 1.0104, |
| "step": 2347 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018692562766461394, |
| "loss": 1.2689, |
| "step": 2348 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018692005461587764, |
| "loss": 1.2017, |
| "step": 2349 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001869144815671413, |
| "loss": 1.05, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018690890851840502, |
| "loss": 1.2335, |
| "step": 2351 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018690333546966868, |
| "loss": 1.1947, |
| "step": 2352 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018689776242093237, |
| "loss": 0.8512, |
| "step": 2353 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018689218937219606, |
| "loss": 1.0779, |
| "step": 2354 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018688661632345976, |
| "loss": 1.1673, |
| "step": 2355 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018688104327472345, |
| "loss": 0.7876, |
| "step": 2356 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018687547022598714, |
| "loss": 0.9952, |
| "step": 2357 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018686989717725083, |
| "loss": 0.8035, |
| "step": 2358 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018686432412851452, |
| "loss": 1.0653, |
| "step": 2359 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018685875107977818, |
| "loss": 1.0743, |
| "step": 2360 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001868531780310419, |
| "loss": 1.2262, |
| "step": 2361 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018684760498230557, |
| "loss": 1.0778, |
| "step": 2362 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018684203193356926, |
| "loss": 1.0969, |
| "step": 2363 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018683645888483298, |
| "loss": 0.9411, |
| "step": 2364 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018683088583609664, |
| "loss": 1.3003, |
| "step": 2365 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018682531278736033, |
| "loss": 1.193, |
| "step": 2366 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018681973973862402, |
| "loss": 0.9465, |
| "step": 2367 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001868141666898877, |
| "loss": 1.1003, |
| "step": 2368 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001868085936411514, |
| "loss": 0.8098, |
| "step": 2369 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018680302059241507, |
| "loss": 1.0502, |
| "step": 2370 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001867974475436788, |
| "loss": 1.1889, |
| "step": 2371 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018679187449494248, |
| "loss": 1.1923, |
| "step": 2372 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018678630144620614, |
| "loss": 0.8669, |
| "step": 2373 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018678072839746986, |
| "loss": 0.9821, |
| "step": 2374 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018677515534873352, |
| "loss": 1.2617, |
| "step": 2375 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018676958229999722, |
| "loss": 1.168, |
| "step": 2376 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001867640092512609, |
| "loss": 1.1468, |
| "step": 2377 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001867584362025246, |
| "loss": 0.7365, |
| "step": 2378 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001867528631537883, |
| "loss": 1.0452, |
| "step": 2379 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018674729010505198, |
| "loss": 1.0278, |
| "step": 2380 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018674171705631567, |
| "loss": 1.0978, |
| "step": 2381 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018673614400757936, |
| "loss": 1.0494, |
| "step": 2382 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018673057095884305, |
| "loss": 1.0257, |
| "step": 2383 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018672499791010674, |
| "loss": 1.041, |
| "step": 2384 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001867194248613704, |
| "loss": 1.0045, |
| "step": 2385 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001867138518126341, |
| "loss": 1.1708, |
| "step": 2386 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001867082787638978, |
| "loss": 1.1817, |
| "step": 2387 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018670270571516148, |
| "loss": 1.0271, |
| "step": 2388 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018669713266642517, |
| "loss": 1.0665, |
| "step": 2389 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018669155961768886, |
| "loss": 0.8635, |
| "step": 2390 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018668598656895256, |
| "loss": 1.0352, |
| "step": 2391 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018668041352021625, |
| "loss": 1.3933, |
| "step": 2392 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018667484047147994, |
| "loss": 1.2108, |
| "step": 2393 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018666926742274363, |
| "loss": 1.0313, |
| "step": 2394 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001866636943740073, |
| "loss": 1.0139, |
| "step": 2395 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.000186658121325271, |
| "loss": 0.9848, |
| "step": 2396 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001866525482765347, |
| "loss": 1.0428, |
| "step": 2397 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018664697522779837, |
| "loss": 1.1581, |
| "step": 2398 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018664140217906208, |
| "loss": 1.347, |
| "step": 2399 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018663582913032575, |
| "loss": 0.8438, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.2, |
| "eval_loss": 1.0824583768844604, |
| "eval_runtime": 333.2747, |
| "eval_samples_per_second": 7.501, |
| "eval_steps_per_second": 0.939, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018663025608158944, |
| "loss": 0.9966, |
| "step": 2401 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018662468303285313, |
| "loss": 1.0935, |
| "step": 2402 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018661910998411682, |
| "loss": 1.0298, |
| "step": 2403 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001866135369353805, |
| "loss": 0.9423, |
| "step": 2404 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018660796388664418, |
| "loss": 1.0589, |
| "step": 2405 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001866023908379079, |
| "loss": 1.0566, |
| "step": 2406 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001865968177891716, |
| "loss": 1.156, |
| "step": 2407 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018659124474043525, |
| "loss": 0.8761, |
| "step": 2408 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018658567169169897, |
| "loss": 1.1456, |
| "step": 2409 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018658009864296263, |
| "loss": 0.8455, |
| "step": 2410 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018657452559422632, |
| "loss": 1.1348, |
| "step": 2411 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018656895254549002, |
| "loss": 1.1684, |
| "step": 2412 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001865633794967537, |
| "loss": 1.2975, |
| "step": 2413 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001865578064480174, |
| "loss": 1.1063, |
| "step": 2414 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001865522333992811, |
| "loss": 1.0618, |
| "step": 2415 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018654666035054478, |
| "loss": 1.0857, |
| "step": 2416 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018654108730180847, |
| "loss": 0.8344, |
| "step": 2417 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018653551425307213, |
| "loss": 0.9515, |
| "step": 2418 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018652994120433585, |
| "loss": 0.9154, |
| "step": 2419 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018652436815559952, |
| "loss": 0.973, |
| "step": 2420 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001865187951068632, |
| "loss": 1.2343, |
| "step": 2421 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018651322205812693, |
| "loss": 1.0202, |
| "step": 2422 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001865076490093906, |
| "loss": 1.03, |
| "step": 2423 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018650207596065428, |
| "loss": 1.1737, |
| "step": 2424 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018649650291191797, |
| "loss": 0.9595, |
| "step": 2425 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018649092986318166, |
| "loss": 1.0562, |
| "step": 2426 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018648535681444536, |
| "loss": 0.9147, |
| "step": 2427 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018647978376570905, |
| "loss": 0.9393, |
| "step": 2428 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018647421071697274, |
| "loss": 1.1562, |
| "step": 2429 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001864686376682364, |
| "loss": 0.9947, |
| "step": 2430 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018646306461950012, |
| "loss": 1.2146, |
| "step": 2431 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001864574915707638, |
| "loss": 1.0075, |
| "step": 2432 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018645191852202747, |
| "loss": 1.1968, |
| "step": 2433 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018644634547329117, |
| "loss": 1.1285, |
| "step": 2434 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018644077242455486, |
| "loss": 0.9921, |
| "step": 2435 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018643519937581855, |
| "loss": 0.9324, |
| "step": 2436 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018642962632708224, |
| "loss": 1.3127, |
| "step": 2437 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018642405327834593, |
| "loss": 1.0279, |
| "step": 2438 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018641848022960962, |
| "loss": 1.3475, |
| "step": 2439 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018641290718087329, |
| "loss": 1.1904, |
| "step": 2440 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.000186407334132137, |
| "loss": 1.2652, |
| "step": 2441 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001864017610834007, |
| "loss": 0.9396, |
| "step": 2442 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018639618803466436, |
| "loss": 1.0187, |
| "step": 2443 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018639061498592808, |
| "loss": 0.8474, |
| "step": 2444 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018638504193719174, |
| "loss": 0.981, |
| "step": 2445 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018637946888845543, |
| "loss": 1.3782, |
| "step": 2446 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018637389583971912, |
| "loss": 0.9356, |
| "step": 2447 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018636832279098281, |
| "loss": 0.8507, |
| "step": 2448 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001863627497422465, |
| "loss": 1.0694, |
| "step": 2449 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018635717669351017, |
| "loss": 0.9342, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.0001863516036447739, |
| "loss": 1.1939, |
| "step": 2451 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 0.00018634603059603758, |
| "loss": 1.0245, |
| "step": 2452 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018634045754730124, |
| "loss": 0.9447, |
| "step": 2453 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018633488449856496, |
| "loss": 0.9924, |
| "step": 2454 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018632931144982863, |
| "loss": 1.3051, |
| "step": 2455 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018632373840109232, |
| "loss": 1.185, |
| "step": 2456 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018631816535235603, |
| "loss": 0.7267, |
| "step": 2457 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001863125923036197, |
| "loss": 1.1291, |
| "step": 2458 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001863070192548834, |
| "loss": 1.0171, |
| "step": 2459 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018630144620614708, |
| "loss": 1.2079, |
| "step": 2460 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018629587315741077, |
| "loss": 0.988, |
| "step": 2461 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018629030010867446, |
| "loss": 1.0096, |
| "step": 2462 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018628472705993815, |
| "loss": 0.8516, |
| "step": 2463 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018627915401120185, |
| "loss": 0.8989, |
| "step": 2464 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001862735809624655, |
| "loss": 1.0806, |
| "step": 2465 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018626800791372923, |
| "loss": 1.0437, |
| "step": 2466 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018626243486499292, |
| "loss": 0.9838, |
| "step": 2467 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018625686181625658, |
| "loss": 0.9874, |
| "step": 2468 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018625128876752027, |
| "loss": 0.9543, |
| "step": 2469 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018624571571878397, |
| "loss": 1.1484, |
| "step": 2470 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018624014267004766, |
| "loss": 1.141, |
| "step": 2471 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018623456962131135, |
| "loss": 0.9927, |
| "step": 2472 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018622899657257504, |
| "loss": 1.2584, |
| "step": 2473 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018622342352383873, |
| "loss": 1.1777, |
| "step": 2474 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001862178504751024, |
| "loss": 0.9541, |
| "step": 2475 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001862122774263661, |
| "loss": 0.8031, |
| "step": 2476 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001862067043776298, |
| "loss": 1.286, |
| "step": 2477 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018620113132889347, |
| "loss": 1.163, |
| "step": 2478 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018619555828015719, |
| "loss": 1.123, |
| "step": 2479 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018618998523142085, |
| "loss": 0.885, |
| "step": 2480 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018618441218268454, |
| "loss": 0.9494, |
| "step": 2481 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018617883913394826, |
| "loss": 1.106, |
| "step": 2482 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018617326608521192, |
| "loss": 1.1191, |
| "step": 2483 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018616769303647561, |
| "loss": 1.043, |
| "step": 2484 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018616211998773928, |
| "loss": 0.9014, |
| "step": 2485 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.000186156546939003, |
| "loss": 0.7868, |
| "step": 2486 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001861509738902667, |
| "loss": 1.0071, |
| "step": 2487 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018614540084153035, |
| "loss": 1.0229, |
| "step": 2488 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018613982779279407, |
| "loss": 1.0933, |
| "step": 2489 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018613425474405773, |
| "loss": 1.0901, |
| "step": 2490 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018612868169532143, |
| "loss": 1.187, |
| "step": 2491 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018612310864658514, |
| "loss": 1.0152, |
| "step": 2492 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001861175355978488, |
| "loss": 1.2908, |
| "step": 2493 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001861119625491125, |
| "loss": 0.9869, |
| "step": 2494 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001861063895003762, |
| "loss": 0.9594, |
| "step": 2495 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018610081645163988, |
| "loss": 0.9094, |
| "step": 2496 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018609524340290357, |
| "loss": 1.0575, |
| "step": 2497 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018608967035416726, |
| "loss": 1.2168, |
| "step": 2498 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018608409730543095, |
| "loss": 0.9624, |
| "step": 2499 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018607852425669462, |
| "loss": 0.8564, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001860729512079583, |
| "loss": 1.0629, |
| "step": 2501 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018606737815922203, |
| "loss": 0.9288, |
| "step": 2502 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001860618051104857, |
| "loss": 0.9603, |
| "step": 2503 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018605623206174938, |
| "loss": 1.0387, |
| "step": 2504 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018605065901301307, |
| "loss": 1.0262, |
| "step": 2505 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018604508596427677, |
| "loss": 1.2527, |
| "step": 2506 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018603951291554046, |
| "loss": 0.9594, |
| "step": 2507 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018603393986680415, |
| "loss": 0.9938, |
| "step": 2508 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018602836681806784, |
| "loss": 1.0597, |
| "step": 2509 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001860227937693315, |
| "loss": 0.9693, |
| "step": 2510 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018601722072059522, |
| "loss": 1.0334, |
| "step": 2511 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001860116476718589, |
| "loss": 0.9508, |
| "step": 2512 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018600607462312258, |
| "loss": 0.9936, |
| "step": 2513 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001860005015743863, |
| "loss": 1.2444, |
| "step": 2514 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018599492852564996, |
| "loss": 1.3, |
| "step": 2515 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018598935547691365, |
| "loss": 0.837, |
| "step": 2516 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018598378242817734, |
| "loss": 0.9687, |
| "step": 2517 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018597820937944103, |
| "loss": 1.0615, |
| "step": 2518 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018597263633070472, |
| "loss": 1.1632, |
| "step": 2519 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018596706328196841, |
| "loss": 1.1314, |
| "step": 2520 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001859614902332321, |
| "loss": 1.1926, |
| "step": 2521 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001859559171844958, |
| "loss": 0.7978, |
| "step": 2522 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018595034413575946, |
| "loss": 1.4411, |
| "step": 2523 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018594477108702318, |
| "loss": 1.0465, |
| "step": 2524 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018593919803828684, |
| "loss": 1.1212, |
| "step": 2525 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018593362498955053, |
| "loss": 1.0186, |
| "step": 2526 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018592805194081425, |
| "loss": 1.1488, |
| "step": 2527 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018592247889207792, |
| "loss": 1.2756, |
| "step": 2528 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001859169058433416, |
| "loss": 1.3374, |
| "step": 2529 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001859113327946053, |
| "loss": 1.1234, |
| "step": 2530 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.000185905759745869, |
| "loss": 0.9628, |
| "step": 2531 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018590018669713268, |
| "loss": 0.9874, |
| "step": 2532 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018589461364839634, |
| "loss": 1.0404, |
| "step": 2533 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018588904059966006, |
| "loss": 0.998, |
| "step": 2534 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018588346755092373, |
| "loss": 1.0275, |
| "step": 2535 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018587789450218742, |
| "loss": 1.4202, |
| "step": 2536 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018587232145345114, |
| "loss": 0.8911, |
| "step": 2537 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001858667484047148, |
| "loss": 0.794, |
| "step": 2538 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001858611753559785, |
| "loss": 0.8302, |
| "step": 2539 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018585560230724218, |
| "loss": 0.9138, |
| "step": 2540 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018585002925850587, |
| "loss": 1.3651, |
| "step": 2541 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018584445620976956, |
| "loss": 1.0194, |
| "step": 2542 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018583888316103326, |
| "loss": 1.0872, |
| "step": 2543 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018583331011229695, |
| "loss": 1.3586, |
| "step": 2544 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018582773706356064, |
| "loss": 0.8972, |
| "step": 2545 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018582216401482433, |
| "loss": 1.2038, |
| "step": 2546 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018581659096608802, |
| "loss": 1.3701, |
| "step": 2547 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018581101791735168, |
| "loss": 0.9572, |
| "step": 2548 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018580544486861538, |
| "loss": 1.2383, |
| "step": 2549 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018579987181987907, |
| "loss": 0.8599, |
| "step": 2550 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018579429877114276, |
| "loss": 0.9177, |
| "step": 2551 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018578872572240645, |
| "loss": 1.0741, |
| "step": 2552 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018578315267367014, |
| "loss": 0.8423, |
| "step": 2553 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018577757962493383, |
| "loss": 0.8187, |
| "step": 2554 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018577200657619752, |
| "loss": 1.0392, |
| "step": 2555 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018576643352746121, |
| "loss": 1.282, |
| "step": 2556 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001857608604787249, |
| "loss": 1.1133, |
| "step": 2557 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018575528742998857, |
| "loss": 1.1219, |
| "step": 2558 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001857497143812523, |
| "loss": 0.9621, |
| "step": 2559 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018574414133251595, |
| "loss": 0.8138, |
| "step": 2560 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018573856828377964, |
| "loss": 1.0408, |
| "step": 2561 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018573299523504336, |
| "loss": 1.1757, |
| "step": 2562 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018572742218630702, |
| "loss": 1.3519, |
| "step": 2563 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018572184913757072, |
| "loss": 1.1795, |
| "step": 2564 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001857162760888344, |
| "loss": 1.1886, |
| "step": 2565 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001857107030400981, |
| "loss": 0.8103, |
| "step": 2566 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.0001857051299913618, |
| "loss": 1.2916, |
| "step": 2567 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018569955694262545, |
| "loss": 0.6953, |
| "step": 2568 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018569398389388917, |
| "loss": 0.8756, |
| "step": 2569 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018568841084515286, |
| "loss": 0.7784, |
| "step": 2570 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018568283779641653, |
| "loss": 1.0144, |
| "step": 2571 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 0.00018567726474768024, |
| "loss": 0.8321, |
| "step": 2572 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.0001856716916989439, |
| "loss": 1.1439, |
| "step": 2573 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.0001856661186502076, |
| "loss": 1.0528, |
| "step": 2574 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.0001856605456014713, |
| "loss": 0.8741, |
| "step": 2575 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018565497255273498, |
| "loss": 1.1352, |
| "step": 2576 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018564939950399867, |
| "loss": 0.8725, |
| "step": 2577 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018564382645526236, |
| "loss": 1.1423, |
| "step": 2578 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018563825340652606, |
| "loss": 1.3373, |
| "step": 2579 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018563268035778975, |
| "loss": 1.3451, |
| "step": 2580 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.0001856271073090534, |
| "loss": 1.1172, |
| "step": 2581 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018562153426031713, |
| "loss": 1.0888, |
| "step": 2582 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.0001856159612115808, |
| "loss": 1.0076, |
| "step": 2583 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018561038816284448, |
| "loss": 1.1834, |
| "step": 2584 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018560481511410818, |
| "loss": 0.8706, |
| "step": 2585 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018559924206537187, |
| "loss": 1.047, |
| "step": 2586 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018559366901663556, |
| "loss": 1.1157, |
| "step": 2587 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018558809596789925, |
| "loss": 1.3231, |
| "step": 2588 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018558252291916294, |
| "loss": 1.2134, |
| "step": 2589 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018557694987042663, |
| "loss": 1.0736, |
| "step": 2590 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018557137682169032, |
| "loss": 0.9433, |
| "step": 2591 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.000185565803772954, |
| "loss": 1.0592, |
| "step": 2592 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018556023072421768, |
| "loss": 1.1437, |
| "step": 2593 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.0001855546576754814, |
| "loss": 1.0742, |
| "step": 2594 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.0001855490846267451, |
| "loss": 0.8504, |
| "step": 2595 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018554351157800875, |
| "loss": 0.8926, |
| "step": 2596 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018553793852927244, |
| "loss": 1.3279, |
| "step": 2597 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018553236548053613, |
| "loss": 1.2002, |
| "step": 2598 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018552679243179982, |
| "loss": 1.3341, |
| "step": 2599 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 0.00018552121938306352, |
| "loss": 1.122, |
| "step": 2600 |
| }, |
| { |
| "epoch": 0.22, |
| "eval_loss": 1.0684343576431274, |
| "eval_runtime": 333.8565, |
| "eval_samples_per_second": 7.488, |
| "eval_steps_per_second": 0.938, |
| "step": 2600 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 35889, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 200, |
| "total_flos": 5.62877373087744e+16, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
