| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.9995209351346173, |
| "global_step": 815, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0, |
| "learning_rate": 8.000000000000001e-07, |
| "loss": 2.0609, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 1.6000000000000001e-06, |
| "loss": 2.0617, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 2.4000000000000003e-06, |
| "loss": 1.1627, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.0, |
| "learning_rate": 3.2000000000000003e-06, |
| "loss": 1.0683, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.000000000000001e-06, |
| "loss": 0.9079, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 4.800000000000001e-06, |
| "loss": 0.9388, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 5.600000000000001e-06, |
| "loss": 0.7955, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 6.4000000000000006e-06, |
| "loss": 0.7388, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 7.2000000000000005e-06, |
| "loss": 0.7783, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 8.000000000000001e-06, |
| "loss": 0.6871, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 8.8e-06, |
| "loss": 0.7109, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.01, |
| "learning_rate": 9.600000000000001e-06, |
| "loss": 0.682, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.04e-05, |
| "loss": 0.5559, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.1200000000000001e-05, |
| "loss": 0.7089, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.2e-05, |
| "loss": 0.5838, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.2800000000000001e-05, |
| "loss": 0.6805, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.3600000000000002e-05, |
| "loss": 0.682, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.4400000000000001e-05, |
| "loss": 0.5885, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.5200000000000002e-05, |
| "loss": 0.6313, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.02, |
| "learning_rate": 1.6000000000000003e-05, |
| "loss": 0.5472, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.6800000000000002e-05, |
| "loss": 0.5028, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.76e-05, |
| "loss": 0.5996, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.8400000000000003e-05, |
| "loss": 0.5874, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.9200000000000003e-05, |
| "loss": 0.4092, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 2e-05, |
| "loss": 0.5476, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.9999920929407192e-05, |
| "loss": 0.6406, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.9999683718879195e-05, |
| "loss": 0.568, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.03, |
| "learning_rate": 1.9999288372167287e-05, |
| "loss": 0.5859, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.9998734895523525e-05, |
| "loss": 0.5781, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.9998023297700656e-05, |
| "loss": 0.5759, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.9997153589951973e-05, |
| "loss": 0.5874, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.999612578603114e-05, |
| "loss": 0.5971, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.9994939902191964e-05, |
| "loss": 0.5363, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.9993595957188153e-05, |
| "loss": 0.6041, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.999209397227302e-05, |
| "loss": 0.4785, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.04, |
| "learning_rate": 1.9990433971199125e-05, |
| "loss": 0.5196, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.9988615980217926e-05, |
| "loss": 0.5425, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.9986640028079348e-05, |
| "loss": 0.5459, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.9984506146031327e-05, |
| "loss": 0.7733, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.998221436781933e-05, |
| "loss": 0.6413, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.9979764729685812e-05, |
| "loss": 0.4834, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.9977157270369642e-05, |
| "loss": 0.5015, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.9974392031105484e-05, |
| "loss": 0.6706, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.05, |
| "learning_rate": 1.997146905562316e-05, |
| "loss": 0.5767, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.996838839014696e-05, |
| "loss": 0.5857, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.9965150083394883e-05, |
| "loss": 0.5288, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.9961754186577906e-05, |
| "loss": 0.6186, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.995820075339913e-05, |
| "loss": 0.5943, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.995448984005298e-05, |
| "loss": 0.4703, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.9950621505224276e-05, |
| "loss": 0.5553, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.9946595810087324e-05, |
| "loss": 0.5438, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.9942412818304944e-05, |
| "loss": 0.6469, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.06, |
| "learning_rate": 1.9938072596027464e-05, |
| "loss": 0.5699, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.993357521189167e-05, |
| "loss": 0.6206, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.9928920737019735e-05, |
| "loss": 0.5435, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.9924109245018072e-05, |
| "loss": 0.6646, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.991914081197619e-05, |
| "loss": 0.6087, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.9914015516465472e-05, |
| "loss": 0.499, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.990873343953795e-05, |
| "loss": 0.5018, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.9903294664725023e-05, |
| "loss": 0.5944, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.989769927803611e-05, |
| "loss": 0.6198, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.9891947367957323e-05, |
| "loss": 0.5779, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.9886039025450053e-05, |
| "loss": 0.6534, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.9879974343949527e-05, |
| "loss": 0.5579, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.9873753419363336e-05, |
| "loss": 0.6594, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.986737635006992e-05, |
| "loss": 0.5662, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.986084323691701e-05, |
| "loss": 0.5762, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.985415418322003e-05, |
| "loss": 0.632, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.08, |
| "learning_rate": 1.9847309294760473e-05, |
| "loss": 0.5329, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.9840308679784207e-05, |
| "loss": 0.5703, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.9833152448999796e-05, |
| "loss": 0.6123, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.9825840715576722e-05, |
| "loss": 0.5812, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.98183735951436e-05, |
| "loss": 0.6229, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.9810751205786352e-05, |
| "loss": 0.5982, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.9802973668046364e-05, |
| "loss": 0.5695, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.9795041104918527e-05, |
| "loss": 0.569, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.09, |
| "learning_rate": 1.9786953641849336e-05, |
| "loss": 0.4832, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.9778711406734895e-05, |
| "loss": 0.5094, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.9770314529918883e-05, |
| "loss": 0.5591, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.976176314419051e-05, |
| "loss": 0.5515, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.97530573847824e-05, |
| "loss": 0.5607, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.9744197389368467e-05, |
| "loss": 0.6065, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.9735183298061726e-05, |
| "loss": 0.5194, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.972601525341209e-05, |
| "loss": 0.6379, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 1.97166934004041e-05, |
| "loss": 0.5566, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.970721788645465e-05, |
| "loss": 0.6059, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.9697588861410633e-05, |
| "loss": 0.5764, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.96878064775466e-05, |
| "loss": 0.5452, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.9677870889562324e-05, |
| "loss": 0.6418, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.9667782254580373e-05, |
| "loss": 0.5807, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.965754073214362e-05, |
| "loss": 0.6478, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.964714648421271e-05, |
| "loss": 0.5602, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.11, |
| "learning_rate": 1.9636599675163516e-05, |
| "loss": 0.659, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.9625900471784525e-05, |
| "loss": 0.6389, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.9615049043274207e-05, |
| "loss": 0.5816, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.960404556123834e-05, |
| "loss": 0.5778, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.9592890199687295e-05, |
| "loss": 0.5341, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.958158313503328e-05, |
| "loss": 0.6079, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.9570124546087555e-05, |
| "loss": 0.6214, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.955851461405761e-05, |
| "loss": 0.6146, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.12, |
| "learning_rate": 1.954675352254428e-05, |
| "loss": 0.6263, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.9534841457538863e-05, |
| "loss": 0.5899, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.952277860742017e-05, |
| "loss": 0.6246, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.9510565162951538e-05, |
| "loss": 0.6316, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.949820131727783e-05, |
| "loss": 0.5275, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.948568726592236e-05, |
| "loss": 0.5629, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.9473023206783834e-05, |
| "loss": 0.7085, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.9460209340133173e-05, |
| "loss": 0.5051, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.9447245868610385e-05, |
| "loss": 0.5925, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.13, |
| "learning_rate": 1.9434132997221347e-05, |
| "loss": 0.6049, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.9420870933334563e-05, |
| "loss": 0.5393, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.9407459886677885e-05, |
| "loss": 0.6263, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.939390006933519e-05, |
| "loss": 0.5361, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.9380191695743037e-05, |
| "loss": 0.6674, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.936633498268728e-05, |
| "loss": 0.6101, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.9352330149299613e-05, |
| "loss": 0.6384, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.9338177417054135e-05, |
| "loss": 0.5612, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 1.9323877009763834e-05, |
| "loss": 0.5542, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.9309429153577042e-05, |
| "loss": 0.5852, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.9294834076973872e-05, |
| "loss": 0.5427, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.9280092010762597e-05, |
| "loss": 0.4797, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.9265203188075997e-05, |
| "loss": 0.6504, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.925016784436768e-05, |
| "loss": 0.6369, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.9234986217408353e-05, |
| "loss": 0.5935, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.921965854728207e-05, |
| "loss": 0.5971, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.9204185076382416e-05, |
| "loss": 0.5357, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.91885660494087e-05, |
| "loss": 0.6258, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.9172801713362064e-05, |
| "loss": 0.6113, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.915689231754159e-05, |
| "loss": 0.5484, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.9140838113540347e-05, |
| "loss": 0.5906, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.912463935524142e-05, |
| "loss": 0.5952, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.9108296298813898e-05, |
| "loss": 0.5483, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.9091809202708805e-05, |
| "loss": 0.4728, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 1.907517832765504e-05, |
| "loss": 0.5304, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.9058403936655235e-05, |
| "loss": 0.5159, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.9041486294981588e-05, |
| "loss": 0.5608, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.90244256701717e-05, |
| "loss": 0.4885, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.9007222332024308e-05, |
| "loss": 0.5817, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.8989876552595045e-05, |
| "loss": 0.5104, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.8972388606192124e-05, |
| "loss": 0.6529, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.8954758769372e-05, |
| "loss": 0.4812, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.17, |
| "learning_rate": 1.8936987320935003e-05, |
| "loss": 0.5135, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.8919074541920928e-05, |
| "loss": 0.5607, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.8901020715604583e-05, |
| "loss": 0.5699, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.888282612749132e-05, |
| "loss": 0.7095, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.8864491065312512e-05, |
| "loss": 0.5636, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.884601581902101e-05, |
| "loss": 0.7662, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.8827400680786538e-05, |
| "loss": 0.5051, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.880864594499111e-05, |
| "loss": 0.6245, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.18, |
| "learning_rate": 1.878975190822434e-05, |
| "loss": 0.4055, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.8770718869278762e-05, |
| "loss": 0.4225, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.875154712914511e-05, |
| "loss": 0.565, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.873223699100756e-05, |
| "loss": 0.6494, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.8712788760238923e-05, |
| "loss": 0.6793, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.869320274439583e-05, |
| "loss": 0.5608, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.8673479253213848e-05, |
| "loss": 0.5642, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.8653618598602615e-05, |
| "loss": 0.4831, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.8633621094640873e-05, |
| "loss": 0.6482, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 1.8613487057571523e-05, |
| "loss": 0.669, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.8593216805796612e-05, |
| "loss": 0.4977, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.8572810659872307e-05, |
| "loss": 0.5657, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.8552268942503817e-05, |
| "loss": 0.5812, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.8531591978540297e-05, |
| "loss": 0.6266, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.8510780094969708e-05, |
| "loss": 0.5382, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.8489833620913644e-05, |
| "loss": 0.5098, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.8468752887622123e-05, |
| "loss": 0.5627, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.2, |
| "learning_rate": 1.844753822846837e-05, |
| "loss": 0.6356, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.8426189978943512e-05, |
| "loss": 0.5371, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.8404708476651303e-05, |
| "loss": 0.6501, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.8383094061302767e-05, |
| "loss": 0.5644, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.836134707471083e-05, |
| "loss": 0.5281, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.8339467860784913e-05, |
| "loss": 0.5183, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.8317456765525506e-05, |
| "loss": 0.5533, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.8295314137018674e-05, |
| "loss": 0.5765, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 1.8273040325430575e-05, |
| "loss": 0.5498, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.8250635683001897e-05, |
| "loss": 0.5123, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.822810056404232e-05, |
| "loss": 0.5765, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.8205435324924884e-05, |
| "loss": 0.5327, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.8182640324080366e-05, |
| "loss": 0.5504, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.8159715921991612e-05, |
| "loss": 0.5853, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.8136662481187838e-05, |
| "loss": 0.5217, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.8113480366238887e-05, |
| "loss": 0.6271, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.8090169943749477e-05, |
| "loss": 0.5847, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.806673158235339e-05, |
| "loss": 0.5188, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.804316565270765e-05, |
| "loss": 0.6149, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.8019472527486667e-05, |
| "loss": 0.5746, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.799565258137633e-05, |
| "loss": 0.7344, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.797170619106809e-05, |
| "loss": 0.5384, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.7947633735253005e-05, |
| "loss": 0.571, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.7923435594615744e-05, |
| "loss": 0.5596, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.23, |
| "learning_rate": 1.789911215182857e-05, |
| "loss": 0.5955, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.78746637915453e-05, |
| "loss": 0.6544, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.7850090900395186e-05, |
| "loss": 0.6219, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.7825393866976858e-05, |
| "loss": 0.6065, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.7800573081852124e-05, |
| "loss": 0.5689, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.777562893753982e-05, |
| "loss": 0.6463, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.7750561828509606e-05, |
| "loss": 0.4868, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.772537215117571e-05, |
| "loss": 0.6449, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 1.7700060303890685e-05, |
| "loss": 0.6386, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.767462668693908e-05, |
| "loss": 0.6586, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.764907170253113e-05, |
| "loss": 0.6455, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.7623395754796386e-05, |
| "loss": 0.5306, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.759759924977733e-05, |
| "loss": 0.537, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.7571682595422956e-05, |
| "loss": 0.5897, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.7545646201582304e-05, |
| "loss": 0.5917, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.7519490479997992e-05, |
| "loss": 0.4963, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.25, |
| "learning_rate": 1.7493215844299706e-05, |
| "loss": 0.6148, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.7466822709997643e-05, |
| "loss": 0.5681, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.744031149447597e-05, |
| "loss": 0.6188, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.7413682616986185e-05, |
| "loss": 0.5269, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.7386936498640513e-05, |
| "loss": 0.6897, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.736007356240524e-05, |
| "loss": 0.6386, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.733309423309403e-05, |
| "loss": 0.529, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.730599893736119e-05, |
| "loss": 0.4585, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.7278788103694944e-05, |
| "loss": 0.5938, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.26, |
| "learning_rate": 1.7251462162410634e-05, |
| "loss": 0.6414, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.7224021545643944e-05, |
| "loss": 0.6064, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.719646668734404e-05, |
| "loss": 0.6895, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.7168798023266713e-05, |
| "loss": 0.5836, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.71410159909675e-05, |
| "loss": 0.4354, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.711312102979475e-05, |
| "loss": 0.6443, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.7085113580882695e-05, |
| "loss": 0.5828, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.7056994087144443e-05, |
| "loss": 0.4517, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 1.7028762993265008e-05, |
| "loss": 0.5847, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.7000420745694256e-05, |
| "loss": 0.6751, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.6971967792639845e-05, |
| "loss": 0.3529, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.6943404584060156e-05, |
| "loss": 0.6262, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.6914731571657146e-05, |
| "loss": 0.6256, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.688594920886924e-05, |
| "loss": 0.5285, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.6857057950864134e-05, |
| "loss": 0.6378, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.6828058254531606e-05, |
| "loss": 0.5912, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.28, |
| "learning_rate": 1.6798950578476296e-05, |
| "loss": 0.547, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.676973538301044e-05, |
| "loss": 0.6591, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.67404131301466e-05, |
| "loss": 0.5077, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.671098428359037e-05, |
| "loss": 0.5505, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.668144930873301e-05, |
| "loss": 0.6505, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.6651808672644114e-05, |
| "loss": 0.6675, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.662206284406422e-05, |
| "loss": 0.5842, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.6592212293397386e-05, |
| "loss": 0.5377, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 1.6562257492703756e-05, |
| "loss": 0.5663, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.6532198915692105e-05, |
| "loss": 0.5378, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.6502037037712333e-05, |
| "loss": 0.541, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.647177233574795e-05, |
| "loss": 0.5447, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.6441405288408544e-05, |
| "loss": 0.6029, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.64109363759222e-05, |
| "loss": 0.5259, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.6380366080127918e-05, |
| "loss": 0.5866, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.6349694884467976e-05, |
| "loss": 0.4886, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.6318923273980298e-05, |
| "loss": 0.7009, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 1.6288051735290782e-05, |
| "loss": 0.56, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 1.62570807566056e-05, |
| "loss": 0.5893, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 1.6226010827703483e-05, |
| "loss": 0.5215, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 1.619484243992797e-05, |
| "loss": 0.6759, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 1.6163576086179636e-05, |
| "loss": 0.6323, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 1.613221226090831e-05, |
| "loss": 0.607, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 1.6100751460105244e-05, |
| "loss": 0.5978, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.31, |
| "learning_rate": 1.6069194181295275e-05, |
| "loss": 0.5623, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.6037540923528943e-05, |
| "loss": 0.6477, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.6005792187374625e-05, |
| "loss": 0.6373, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.5973948474910595e-05, |
| "loss": 0.4277, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.5942010289717108e-05, |
| "loss": 0.4767, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.5909978136868396e-05, |
| "loss": 0.5801, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.5877852522924733e-05, |
| "loss": 0.5095, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.5845633955924382e-05, |
| "loss": 0.6298, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.5813322945375583e-05, |
| "loss": 0.4727, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 1.5780920002248484e-05, |
| "loss": 0.49, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.5748425638967074e-05, |
| "loss": 0.5025, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.571584036940107e-05, |
| "loss": 0.5592, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.5683164708857778e-05, |
| "loss": 0.5254, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.5650399174073977e-05, |
| "loss": 0.4595, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.561754428320771e-05, |
| "loss": 0.5779, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.5584600555830125e-05, |
| "loss": 0.5553, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.5551568512917228e-05, |
| "loss": 0.6231, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.33, |
| "learning_rate": 1.5518448676841665e-05, |
| "loss": 0.639, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.5485241571364446e-05, |
| "loss": 0.5169, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.5451947721626676e-05, |
| "loss": 0.6414, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.541856765414125e-05, |
| "loss": 0.5767, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.5385101896784497e-05, |
| "loss": 0.5368, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.5351550978787882e-05, |
| "loss": 0.6418, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.5317915430729602e-05, |
| "loss": 0.679, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.5284195784526196e-05, |
| "loss": 0.6343, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 1.5250392573424152e-05, |
| "loss": 0.5204, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 1.521650633199145e-05, |
| "loss": 0.6268, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 1.5182537596109139e-05, |
| "loss": 0.5882, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 1.5148486902962832e-05, |
| "loss": 0.6283, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 1.5114354791034225e-05, |
| "loss": 0.5458, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 1.5080141800092584e-05, |
| "loss": 0.4972, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 1.5045848471186206e-05, |
| "loss": 0.5118, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 1.501147534663386e-05, |
| "loss": 0.5391, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.35, |
| "learning_rate": 1.4977022970016211e-05, |
| "loss": 0.5563, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.494249188616723e-05, |
| "loss": 0.5782, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.490788264116557e-05, |
| "loss": 0.6268, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.4873195782325934e-05, |
| "loss": 0.4464, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.4838431858190423e-05, |
| "loss": 0.556, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.4803591418519849e-05, |
| "loss": 0.5001, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.4768675014285063e-05, |
| "loss": 0.5037, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.4733683197658216e-05, |
| "loss": 0.4744, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 1.4698616522004046e-05, |
| "loss": 0.65, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.4663475541871121e-05, |
| "loss": 0.5298, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.4628260812983066e-05, |
| "loss": 0.5683, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.4592972892229779e-05, |
| "loss": 0.5662, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.4557612337658626e-05, |
| "loss": 0.4492, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.4522179708465603e-05, |
| "loss": 0.7642, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.448667556498651e-05, |
| "loss": 0.4797, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.4451100468688086e-05, |
| "loss": 0.5393, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.4415454982159121e-05, |
| "loss": 0.5399, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.437973966910156e-05, |
| "loss": 0.5199, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.4343955094321605e-05, |
| "loss": 0.4494, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.4308101823720755e-05, |
| "loss": 0.5684, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.4272180424286895e-05, |
| "loss": 0.5599, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.4236191464085286e-05, |
| "loss": 0.4956, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.4200135512249606e-05, |
| "loss": 0.6146, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.4164013138972966e-05, |
| "loss": 0.5776, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.38, |
| "learning_rate": 1.4127824915498852e-05, |
| "loss": 0.3925, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.4091571414112119e-05, |
| "loss": 0.5936, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.405525320812994e-05, |
| "loss": 0.6285, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.4018870871892729e-05, |
| "loss": 0.4788, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.3982424980755064e-05, |
| "loss": 0.5112, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.3945916111076594e-05, |
| "loss": 0.653, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.3909344840212903e-05, |
| "loss": 0.5955, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.3872711746506413e-05, |
| "loss": 0.7006, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.3836017409277205e-05, |
| "loss": 0.588, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 1.3799262408813882e-05, |
| "loss": 0.6106, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.3762447326364377e-05, |
| "loss": 0.5551, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.3725572744126772e-05, |
| "loss": 0.5657, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.3688639245240078e-05, |
| "loss": 0.525, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.365164741377503e-05, |
| "loss": 0.5116, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.3614597834724834e-05, |
| "loss": 0.5437, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.3577491093995928e-05, |
| "loss": 0.6001, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.3540327778398703e-05, |
| "loss": 0.5839, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 1.3503108475638244e-05, |
| "loss": 0.5819, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.3465833774305016e-05, |
| "loss": 0.6025, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.3428504263865565e-05, |
| "loss": 0.6385, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.3391120534653198e-05, |
| "loss": 0.4742, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.3353683177858634e-05, |
| "loss": 0.633, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.331619278552068e-05, |
| "loss": 0.5259, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.327864995051684e-05, |
| "loss": 0.6526, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.324105526655396e-05, |
| "loss": 0.6007, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 1.3203409328158827e-05, |
| "loss": 0.5424, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.3165712730668778e-05, |
| "loss": 0.5087, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.3127966070222273e-05, |
| "loss": 0.7112, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.3090169943749475e-05, |
| "loss": 0.62, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.3052324948962811e-05, |
| "loss": 0.6589, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.3014431684347515e-05, |
| "loss": 0.6035, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.2976490749152166e-05, |
| "loss": 0.5496, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.2938502743379212e-05, |
| "loss": 0.5194, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.42, |
| "learning_rate": 1.2900468267775477e-05, |
| "loss": 0.4988, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.2862387923822672e-05, |
| "loss": 0.4649, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.2824262313727869e-05, |
| "loss": 0.465, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.278609204041399e-05, |
| "loss": 0.6568, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.2747877707510252e-05, |
| "loss": 0.6085, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.2709619919342662e-05, |
| "loss": 0.5197, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.2671319280924408e-05, |
| "loss": 0.675, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.2632976397946325e-05, |
| "loss": 0.4796, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.43, |
| "learning_rate": 1.2594591876767317e-05, |
| "loss": 0.7135, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.2556166324404747e-05, |
| "loss": 0.5429, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.2517700348524864e-05, |
| "loss": 0.5058, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.2479194557433163e-05, |
| "loss": 0.7637, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.244064956006479e-05, |
| "loss": 0.5292, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.2402065965974912e-05, |
| "loss": 0.5596, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.2363444385329052e-05, |
| "loss": 0.5425, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.2324785428893464e-05, |
| "loss": 0.5623, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.2286089708025477e-05, |
| "loss": 0.5131, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.2247357834663804e-05, |
| "loss": 0.5603, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.2208590421318882e-05, |
| "loss": 0.5538, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.2169788081063181e-05, |
| "loss": 0.5815, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.2130951427521514e-05, |
| "loss": 0.6195, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.2092081074861324e-05, |
| "loss": 0.6948, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.2053177637782969e-05, |
| "loss": 0.427, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.2014241731510026e-05, |
| "loss": 0.6659, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.1975273971779528e-05, |
| "loss": 0.5934, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.45, |
| "learning_rate": 1.1936274974832247e-05, |
| "loss": 0.5807, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.1897245357402948e-05, |
| "loss": 0.5779, |
| "step": 372 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.1858185736710621e-05, |
| "loss": 0.5506, |
| "step": 373 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.181909673044875e-05, |
| "loss": 0.5654, |
| "step": 374 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.1779978956775507e-05, |
| "loss": 0.6878, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.1740833034303997e-05, |
| "loss": 0.4305, |
| "step": 376 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.1701659582092493e-05, |
| "loss": 0.5167, |
| "step": 377 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.1662459219634603e-05, |
| "loss": 0.5442, |
| "step": 378 |
| }, |
| { |
| "epoch": 0.46, |
| "learning_rate": 1.1623232566849512e-05, |
| "loss": 0.4127, |
| "step": 379 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.158398024407215e-05, |
| "loss": 0.6508, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.1544702872043411e-05, |
| "loss": 0.7112, |
| "step": 381 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.1505401071900313e-05, |
| "loss": 0.5279, |
| "step": 382 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.1466075465166182e-05, |
| "loss": 0.5537, |
| "step": 383 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.1426726673740817e-05, |
| "loss": 0.6383, |
| "step": 384 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.1387355319890685e-05, |
| "loss": 0.646, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.1347962026239032e-05, |
| "loss": 0.5558, |
| "step": 386 |
| }, |
| { |
| "epoch": 0.47, |
| "learning_rate": 1.1308547415756084e-05, |
| "loss": 0.5827, |
| "step": 387 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.1269112111749155e-05, |
| "loss": 0.4887, |
| "step": 388 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.1229656737852823e-05, |
| "loss": 0.4916, |
| "step": 389 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.119018191801905e-05, |
| "loss": 0.4968, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.1150688276507313e-05, |
| "loss": 0.6529, |
| "step": 391 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.1111176437874747e-05, |
| "loss": 0.7072, |
| "step": 392 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.1071647026966245e-05, |
| "loss": 0.6109, |
| "step": 393 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.1032100668904606e-05, |
| "loss": 0.4233, |
| "step": 394 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 1.0992537989080618e-05, |
| "loss": 0.5682, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.0952959613143193e-05, |
| "loss": 0.5379, |
| "step": 396 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.0913366166989462e-05, |
| "loss": 0.6789, |
| "step": 397 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.0873758276754877e-05, |
| "loss": 0.44, |
| "step": 398 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.083413656880331e-05, |
| "loss": 0.512, |
| "step": 399 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.0794501669717146e-05, |
| "loss": 0.5473, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.075485420628738e-05, |
| "loss": 0.4838, |
| "step": 401 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.0715194805503697e-05, |
| "loss": 0.5978, |
| "step": 402 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 1.0675524094544564e-05, |
| "loss": 0.5881, |
| "step": 403 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.0635842700767309e-05, |
| "loss": 0.5636, |
| "step": 404 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.05961512516982e-05, |
| "loss": 0.5508, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.0556450375022516e-05, |
| "loss": 0.6022, |
| "step": 406 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.0516740698574624e-05, |
| "loss": 0.4875, |
| "step": 407 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.0477022850328063e-05, |
| "loss": 0.5067, |
| "step": 408 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.0437297458385587e-05, |
| "loss": 0.4638, |
| "step": 409 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.039756515096926e-05, |
| "loss": 0.5478, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.5, |
| "learning_rate": 1.0357826556410495e-05, |
| "loss": 0.425, |
| "step": 411 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.0318082303140145e-05, |
| "loss": 0.5791, |
| "step": 412 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.0278333019678542e-05, |
| "loss": 0.5872, |
| "step": 413 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.0238579334625565e-05, |
| "loss": 0.6167, |
| "step": 414 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.0198821876650702e-05, |
| "loss": 0.5502, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.0159061274483113e-05, |
| "loss": 0.5996, |
| "step": 416 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.0119298156901669e-05, |
| "loss": 0.6092, |
| "step": 417 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.0079533152725025e-05, |
| "loss": 0.5073, |
| "step": 418 |
| }, |
| { |
| "epoch": 0.51, |
| "learning_rate": 1.0039766890801676e-05, |
| "loss": 0.5053, |
| "step": 419 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 1e-05, |
| "loss": 0.6447, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.960233109198327e-06, |
| "loss": 0.4794, |
| "step": 421 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.920466847274978e-06, |
| "loss": 0.4794, |
| "step": 422 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.880701843098335e-06, |
| "loss": 0.6036, |
| "step": 423 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.840938725516889e-06, |
| "loss": 0.6568, |
| "step": 424 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.801178123349298e-06, |
| "loss": 0.4632, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.761420665374439e-06, |
| "loss": 0.5784, |
| "step": 426 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.72166698032146e-06, |
| "loss": 0.6247, |
| "step": 427 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 9.681917696859855e-06, |
| "loss": 0.4805, |
| "step": 428 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.642173443589508e-06, |
| "loss": 0.589, |
| "step": 429 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.602434849030747e-06, |
| "loss": 0.5272, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.562702541614416e-06, |
| "loss": 0.5351, |
| "step": 431 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.522977149671942e-06, |
| "loss": 0.6636, |
| "step": 432 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.48325930142538e-06, |
| "loss": 0.6028, |
| "step": 433 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.443549624977488e-06, |
| "loss": 0.5579, |
| "step": 434 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.403848748301802e-06, |
| "loss": 0.5568, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 9.364157299232693e-06, |
| "loss": 0.5308, |
| "step": 436 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.324475905455439e-06, |
| "loss": 0.5737, |
| "step": 437 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.284805194496305e-06, |
| "loss": 0.4532, |
| "step": 438 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.245145793712625e-06, |
| "loss": 0.4631, |
| "step": 439 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.205498330282857e-06, |
| "loss": 0.5277, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.165863431196693e-06, |
| "loss": 0.5677, |
| "step": 441 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.126241723245123e-06, |
| "loss": 0.6744, |
| "step": 442 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.086633833010538e-06, |
| "loss": 0.5684, |
| "step": 443 |
| }, |
| { |
| "epoch": 0.54, |
| "learning_rate": 9.047040386856807e-06, |
| "loss": 0.542, |
| "step": 444 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 9.007462010919387e-06, |
| "loss": 0.5561, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.967899331095399e-06, |
| "loss": 0.526, |
| "step": 446 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.92835297303376e-06, |
| "loss": 0.5821, |
| "step": 447 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.88882356212526e-06, |
| "loss": 0.584, |
| "step": 448 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.849311723492689e-06, |
| "loss": 0.6277, |
| "step": 449 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.809818081980954e-06, |
| "loss": 0.6401, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.77034326214718e-06, |
| "loss": 0.6108, |
| "step": 451 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 8.73088788825085e-06, |
| "loss": 0.5503, |
| "step": 452 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.69145258424392e-06, |
| "loss": 0.5291, |
| "step": 453 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.65203797376097e-06, |
| "loss": 0.5102, |
| "step": 454 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.61264468010932e-06, |
| "loss": 0.5068, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.573273326259184e-06, |
| "loss": 0.628, |
| "step": 456 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.533924534833822e-06, |
| "loss": 0.6384, |
| "step": 457 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.494598928099688e-06, |
| "loss": 0.5289, |
| "step": 458 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.455297127956589e-06, |
| "loss": 0.6462, |
| "step": 459 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 8.416019755927851e-06, |
| "loss": 0.5612, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.376767433150491e-06, |
| "loss": 0.5404, |
| "step": 461 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.337540780365402e-06, |
| "loss": 0.5564, |
| "step": 462 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.298340417907512e-06, |
| "loss": 0.567, |
| "step": 463 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.259166965696004e-06, |
| "loss": 0.4957, |
| "step": 464 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.2200210432245e-06, |
| "loss": 0.5601, |
| "step": 465 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.180903269551254e-06, |
| "loss": 0.5895, |
| "step": 466 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.14181426328938e-06, |
| "loss": 0.5821, |
| "step": 467 |
| }, |
| { |
| "epoch": 0.57, |
| "learning_rate": 8.102754642597055e-06, |
| "loss": 0.5428, |
| "step": 468 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 8.063725025167755e-06, |
| "loss": 0.5913, |
| "step": 469 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 8.024726028220474e-06, |
| "loss": 0.4581, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 7.985758268489975e-06, |
| "loss": 0.5537, |
| "step": 471 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 7.946822362217033e-06, |
| "loss": 0.5797, |
| "step": 472 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 7.90791892513868e-06, |
| "loss": 0.5318, |
| "step": 473 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 7.869048572478488e-06, |
| "loss": 0.5732, |
| "step": 474 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 7.83021191893682e-06, |
| "loss": 0.5898, |
| "step": 475 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 7.79140957868112e-06, |
| "loss": 0.5905, |
| "step": 476 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 7.752642165336195e-06, |
| "loss": 0.5652, |
| "step": 477 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.713910291974527e-06, |
| "loss": 0.555, |
| "step": 478 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.675214571106538e-06, |
| "loss": 0.5772, |
| "step": 479 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.636555614670953e-06, |
| "loss": 0.5617, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.597934034025093e-06, |
| "loss": 0.4953, |
| "step": 481 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.559350439935213e-06, |
| "loss": 0.5632, |
| "step": 482 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.5208054425668405e-06, |
| "loss": 0.6101, |
| "step": 483 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.48229965147514e-06, |
| "loss": 0.5806, |
| "step": 484 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 7.443833675595254e-06, |
| "loss": 0.619, |
| "step": 485 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.405408123232687e-06, |
| "loss": 0.5195, |
| "step": 486 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.367023602053679e-06, |
| "loss": 0.6557, |
| "step": 487 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.328680719075596e-06, |
| "loss": 0.452, |
| "step": 488 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.29038008065734e-06, |
| "loss": 0.5612, |
| "step": 489 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.252122292489747e-06, |
| "loss": 0.5479, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.213907959586015e-06, |
| "loss": 0.6204, |
| "step": 491 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.175737686272131e-06, |
| "loss": 0.5646, |
| "step": 492 |
| }, |
| { |
| "epoch": 0.6, |
| "learning_rate": 7.137612076177329e-06, |
| "loss": 0.4528, |
| "step": 493 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 7.099531732224524e-06, |
| "loss": 0.5371, |
| "step": 494 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 7.061497256620793e-06, |
| "loss": 0.5719, |
| "step": 495 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 7.0235092508478385e-06, |
| "loss": 0.5563, |
| "step": 496 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 6.985568315652489e-06, |
| "loss": 0.4069, |
| "step": 497 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 6.947675051037193e-06, |
| "loss": 0.4976, |
| "step": 498 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 6.909830056250527e-06, |
| "loss": 0.6324, |
| "step": 499 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 6.872033929777731e-06, |
| "loss": 0.5583, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.61, |
| "learning_rate": 6.834287269331226e-06, |
| "loss": 0.5776, |
| "step": 501 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.7965906718411766e-06, |
| "loss": 0.512, |
| "step": 502 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.758944733446043e-06, |
| "loss": 0.6016, |
| "step": 503 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.721350049483163e-06, |
| "loss": 0.5053, |
| "step": 504 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.683807214479323e-06, |
| "loss": 0.6061, |
| "step": 505 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.646316822141368e-06, |
| "loss": 0.4553, |
| "step": 506 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.608879465346804e-06, |
| "loss": 0.482, |
| "step": 507 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.571495736134435e-06, |
| "loss": 0.6818, |
| "step": 508 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 6.5341662256949844e-06, |
| "loss": 0.6059, |
| "step": 509 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.496891524361757e-06, |
| "loss": 0.5527, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.459672221601301e-06, |
| "loss": 0.5475, |
| "step": 511 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.422508906004079e-06, |
| "loss": 0.4486, |
| "step": 512 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.38540216527517e-06, |
| "loss": 0.611, |
| "step": 513 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.348352586224972e-06, |
| "loss": 0.4929, |
| "step": 514 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.311360754759923e-06, |
| "loss": 0.4924, |
| "step": 515 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.274427255873231e-06, |
| "loss": 0.5938, |
| "step": 516 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 6.237552673635625e-06, |
| "loss": 0.6196, |
| "step": 517 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 6.20073759118612e-06, |
| "loss": 0.536, |
| "step": 518 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 6.1639825907227964e-06, |
| "loss": 0.5342, |
| "step": 519 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 6.127288253493591e-06, |
| "loss": 0.5688, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 6.0906551597871e-06, |
| "loss": 0.5241, |
| "step": 521 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 6.0540838889234076e-06, |
| "loss": 0.4907, |
| "step": 522 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 6.017575019244935e-06, |
| "loss": 0.5352, |
| "step": 523 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 5.9811291281072725e-06, |
| "loss": 0.5815, |
| "step": 524 |
| }, |
| { |
| "epoch": 0.64, |
| "learning_rate": 5.944746791870062e-06, |
| "loss": 0.5706, |
| "step": 525 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.9084285858878835e-06, |
| "loss": 0.4357, |
| "step": 526 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.872175084501154e-06, |
| "loss": 0.5235, |
| "step": 527 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.835986861027038e-06, |
| "loss": 0.4825, |
| "step": 528 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.799864487750395e-06, |
| "loss": 0.4695, |
| "step": 529 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.7638085359147235e-06, |
| "loss": 0.5062, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.72781957571311e-06, |
| "loss": 0.5901, |
| "step": 531 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.6918981762792446e-06, |
| "loss": 0.6307, |
| "step": 532 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.656044905678402e-06, |
| "loss": 0.6259, |
| "step": 533 |
| }, |
| { |
| "epoch": 0.65, |
| "learning_rate": 5.620260330898441e-06, |
| "loss": 0.5619, |
| "step": 534 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.584545017840886e-06, |
| "loss": 0.6941, |
| "step": 535 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.548899531311915e-06, |
| "loss": 0.5202, |
| "step": 536 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.51332443501349e-06, |
| "loss": 0.6196, |
| "step": 537 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.477820291534401e-06, |
| "loss": 0.5323, |
| "step": 538 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.442387662341377e-06, |
| "loss": 0.6519, |
| "step": 539 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.40702710777022e-06, |
| "loss": 0.5783, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.371739187016935e-06, |
| "loss": 0.4833, |
| "step": 541 |
| }, |
| { |
| "epoch": 0.66, |
| "learning_rate": 5.33652445812888e-06, |
| "loss": 0.5678, |
| "step": 542 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.3013834779959564e-06, |
| "loss": 0.7164, |
| "step": 543 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.26631680234179e-06, |
| "loss": 0.453, |
| "step": 544 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.231324985714942e-06, |
| "loss": 0.544, |
| "step": 545 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.196408581480152e-06, |
| "loss": 0.5213, |
| "step": 546 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.161568141809583e-06, |
| "loss": 0.6047, |
| "step": 547 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.126804217674068e-06, |
| "loss": 0.5118, |
| "step": 548 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.092117358834434e-06, |
| "loss": 0.5184, |
| "step": 549 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 5.057508113832772e-06, |
| "loss": 0.4618, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 5.0229770299837886e-06, |
| "loss": 0.5749, |
| "step": 551 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.988524653366144e-06, |
| "loss": 0.5626, |
| "step": 552 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.954151528813796e-06, |
| "loss": 0.4327, |
| "step": 553 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.91985819990742e-06, |
| "loss": 0.6465, |
| "step": 554 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.885645208965779e-06, |
| "loss": 0.5033, |
| "step": 555 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.851513097037171e-06, |
| "loss": 0.5334, |
| "step": 556 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.8174624038908645e-06, |
| "loss": 0.6025, |
| "step": 557 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.783493668008551e-06, |
| "loss": 0.5617, |
| "step": 558 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.749607426575849e-06, |
| "loss": 0.5354, |
| "step": 559 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.7158042154738094e-06, |
| "loss": 0.5769, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.682084569270402e-06, |
| "loss": 0.5404, |
| "step": 561 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.648449021212118e-06, |
| "loss": 0.5062, |
| "step": 562 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.614898103215507e-06, |
| "loss": 0.4504, |
| "step": 563 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.5814323458587565e-06, |
| "loss": 0.5203, |
| "step": 564 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.548052278373327e-06, |
| "loss": 0.5625, |
| "step": 565 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 4.5147584286355575e-06, |
| "loss": 0.5452, |
| "step": 566 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.4815513231583385e-06, |
| "loss": 0.519, |
| "step": 567 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.448431487082776e-06, |
| "loss": 0.5522, |
| "step": 568 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.415399444169878e-06, |
| "loss": 0.6247, |
| "step": 569 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.382455716792291e-06, |
| "loss": 0.4886, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.349600825926028e-06, |
| "loss": 0.4883, |
| "step": 571 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.316835291142223e-06, |
| "loss": 0.5447, |
| "step": 572 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.284159630598935e-06, |
| "loss": 0.5352, |
| "step": 573 |
| }, |
| { |
| "epoch": 0.7, |
| "learning_rate": 4.251574361032926e-06, |
| "loss": 0.6183, |
| "step": 574 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.219079997751515e-06, |
| "loss": 0.6006, |
| "step": 575 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.186677054624421e-06, |
| "loss": 0.4818, |
| "step": 576 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.154366044075623e-06, |
| "loss": 0.5417, |
| "step": 577 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.12214747707527e-06, |
| "loss": 0.5962, |
| "step": 578 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.090021863131608e-06, |
| "loss": 0.5575, |
| "step": 579 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.057989710282897e-06, |
| "loss": 0.4924, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 4.026051525089403e-06, |
| "loss": 0.636, |
| "step": 581 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.994207812625378e-06, |
| "loss": 0.5569, |
| "step": 582 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 3.962459076471059e-06, |
| "loss": 0.6194, |
| "step": 583 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.930805818704731e-06, |
| "loss": 0.4695, |
| "step": 584 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.899248539894756e-06, |
| "loss": 0.561, |
| "step": 585 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.867787739091691e-06, |
| "loss": 0.5154, |
| "step": 586 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.8364239138203676e-06, |
| "loss": 0.5684, |
| "step": 587 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.8051575600720336e-06, |
| "loss": 0.5292, |
| "step": 588 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.7739891722965207e-06, |
| "loss": 0.4915, |
| "step": 589 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.7429192433944016e-06, |
| "loss": 0.5364, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.72, |
| "learning_rate": 3.7119482647092187e-06, |
| "loss": 0.5628, |
| "step": 591 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.6810767260197044e-06, |
| "loss": 0.5665, |
| "step": 592 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.650305115532028e-06, |
| "loss": 0.5763, |
| "step": 593 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.619633919872083e-06, |
| "loss": 0.5347, |
| "step": 594 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.589063624077802e-06, |
| "loss": 0.6078, |
| "step": 595 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.5585947115914586e-06, |
| "loss": 0.6073, |
| "step": 596 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.528227664252052e-06, |
| "loss": 0.4946, |
| "step": 597 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.497962962287671e-06, |
| "loss": 0.5556, |
| "step": 598 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 3.467801084307896e-06, |
| "loss": 0.4763, |
| "step": 599 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.4377425072962467e-06, |
| "loss": 0.5127, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.4077877066026177e-06, |
| "loss": 0.637, |
| "step": 601 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.377937155935781e-06, |
| "loss": 0.5593, |
| "step": 602 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.3481913273558885e-06, |
| "loss": 0.5527, |
| "step": 603 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.318550691266993e-06, |
| "loss": 0.5271, |
| "step": 604 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.2890157164096315e-06, |
| "loss": 0.5101, |
| "step": 605 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.2595868698534007e-06, |
| "loss": 0.5387, |
| "step": 606 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 3.2302646169895634e-06, |
| "loss": 0.562, |
| "step": 607 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.2010494215237086e-06, |
| "loss": 0.4971, |
| "step": 608 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.1719417454683954e-06, |
| "loss": 0.5409, |
| "step": 609 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.1429420491358696e-06, |
| "loss": 0.5866, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.114050791130766e-06, |
| "loss": 0.5779, |
| "step": 611 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.085268428342858e-06, |
| "loss": 0.4699, |
| "step": 612 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.056595415939849e-06, |
| "loss": 0.6173, |
| "step": 613 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 3.0280322073601585e-06, |
| "loss": 0.5483, |
| "step": 614 |
| }, |
| { |
| "epoch": 0.75, |
| "learning_rate": 2.999579254305748e-06, |
| "loss": 0.5824, |
| "step": 615 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 2.9712370067349915e-06, |
| "loss": 0.5668, |
| "step": 616 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 2.9430059128555586e-06, |
| "loss": 0.6005, |
| "step": 617 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 2.9148864191173075e-06, |
| "loss": 0.5247, |
| "step": 618 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 2.8868789702052513e-06, |
| "loss": 0.5157, |
| "step": 619 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 2.8589840090325028e-06, |
| "loss": 0.4628, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 2.8312019767332888e-06, |
| "loss": 0.4138, |
| "step": 621 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 2.8035333126559627e-06, |
| "loss": 0.5048, |
| "step": 622 |
| }, |
| { |
| "epoch": 0.76, |
| "learning_rate": 2.775978454356055e-06, |
| "loss": 0.5119, |
| "step": 623 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 2.7485378375893634e-06, |
| "loss": 0.4522, |
| "step": 624 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 2.721211896305059e-06, |
| "loss": 0.5843, |
| "step": 625 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 2.6940010626388134e-06, |
| "loss": 0.6001, |
| "step": 626 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 2.6669057669059727e-06, |
| "loss": 0.6444, |
| "step": 627 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 2.6399264375947632e-06, |
| "loss": 0.6018, |
| "step": 628 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 2.613063501359492e-06, |
| "loss": 0.6042, |
| "step": 629 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 2.5863173830138212e-06, |
| "loss": 0.4609, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.77, |
| "learning_rate": 2.559688505524033e-06, |
| "loss": 0.5247, |
| "step": 631 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.5331772900023566e-06, |
| "loss": 0.5996, |
| "step": 632 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.5067841557002994e-06, |
| "loss": 0.5239, |
| "step": 633 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.48050952000201e-06, |
| "loss": 0.5562, |
| "step": 634 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.454353798417698e-06, |
| "loss": 0.5061, |
| "step": 635 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.4283174045770464e-06, |
| "loss": 0.6099, |
| "step": 636 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.402400750222669e-06, |
| "loss": 0.6004, |
| "step": 637 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.3766042452036164e-06, |
| "loss": 0.611, |
| "step": 638 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.350928297468872e-06, |
| "loss": 0.5936, |
| "step": 639 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 2.325373313060919e-06, |
| "loss": 0.4792, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.2999396961093157e-06, |
| "loss": 0.586, |
| "step": 641 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.274627848824288e-06, |
| "loss": 0.6163, |
| "step": 642 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.249438171490398e-06, |
| "loss": 0.4845, |
| "step": 643 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.224371062460184e-06, |
| "loss": 0.5021, |
| "step": 644 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.19942691814788e-06, |
| "loss": 0.5239, |
| "step": 645 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.1746061330231427e-06, |
| "loss": 0.484, |
| "step": 646 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.1499090996048146e-06, |
| "loss": 0.4892, |
| "step": 647 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 2.1253362084547037e-06, |
| "loss": 0.5326, |
| "step": 648 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.100887848171431e-06, |
| "loss": 0.5735, |
| "step": 649 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.0765644053842583e-06, |
| "loss": 0.6161, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.0523662647469954e-06, |
| "loss": 0.6498, |
| "step": 651 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.0282938089319125e-06, |
| "loss": 0.5519, |
| "step": 652 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 2.0043474186236723e-06, |
| "loss": 0.5435, |
| "step": 653 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 1.980527472513337e-06, |
| "loss": 0.5513, |
| "step": 654 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 1.9568343472923524e-06, |
| "loss": 0.6049, |
| "step": 655 |
| }, |
| { |
| "epoch": 0.8, |
| "learning_rate": 1.9332684176466134e-06, |
| "loss": 0.6951, |
| "step": 656 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.9098300562505266e-06, |
| "loss": 0.4251, |
| "step": 657 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.8865196337611136e-06, |
| "loss": 0.5339, |
| "step": 658 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.863337518812165e-06, |
| "loss": 0.5143, |
| "step": 659 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.840284078008393e-06, |
| "loss": 0.6168, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.8173596759196388e-06, |
| "loss": 0.4514, |
| "step": 661 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.79456467507512e-06, |
| "loss": 0.4451, |
| "step": 662 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.7718994359576835e-06, |
| "loss": 0.5861, |
| "step": 663 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.7493643169981045e-06, |
| "loss": 0.5268, |
| "step": 664 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.7269596745694295e-06, |
| "loss": 0.509, |
| "step": 665 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.7046858629813257e-06, |
| "loss": 0.4789, |
| "step": 666 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.682543234474494e-06, |
| "loss": 0.5495, |
| "step": 667 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.6605321392150875e-06, |
| "loss": 0.5065, |
| "step": 668 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.6386529252891725e-06, |
| "loss": 0.6325, |
| "step": 669 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.6169059386972342e-06, |
| "loss": 0.5761, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.5952915233486987e-06, |
| "loss": 0.4692, |
| "step": 671 |
| }, |
| { |
| "epoch": 0.82, |
| "learning_rate": 1.57381002105649e-06, |
| "loss": 0.61, |
| "step": 672 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 1.5524617715316348e-06, |
| "loss": 0.5462, |
| "step": 673 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 1.5312471123778795e-06, |
| "loss": 0.6237, |
| "step": 674 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 1.5101663790863597e-06, |
| "loss": 0.6188, |
| "step": 675 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 1.4892199050302968e-06, |
| "loss": 0.5557, |
| "step": 676 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 1.4684080214597064e-06, |
| "loss": 0.5326, |
| "step": 677 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 1.4477310574961856e-06, |
| "loss": 0.5854, |
| "step": 678 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 1.4271893401276972e-06, |
| "loss": 0.4828, |
| "step": 679 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 1.4067831942033904e-06, |
| "loss": 0.5964, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.3865129424284785e-06, |
| "loss": 0.5067, |
| "step": 681 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.3663789053591292e-06, |
| "loss": 0.5634, |
| "step": 682 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.3463814013973875e-06, |
| "loss": 0.5593, |
| "step": 683 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.3265207467861552e-06, |
| "loss": 0.5708, |
| "step": 684 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.3067972556041753e-06, |
| "loss": 0.44, |
| "step": 685 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.287211239761078e-06, |
| "loss": 0.5631, |
| "step": 686 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.267763008992442e-06, |
| "loss": 0.4403, |
| "step": 687 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.2484528708548904e-06, |
| "loss": 0.48, |
| "step": 688 |
| }, |
| { |
| "epoch": 0.84, |
| "learning_rate": 1.2292811307212415e-06, |
| "loss": 0.577, |
| "step": 689 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.2102480917756632e-06, |
| "loss": 0.5179, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.1913540550088899e-06, |
| "loss": 0.5526, |
| "step": 691 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.1725993192134633e-06, |
| "loss": 0.4579, |
| "step": 692 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.1539841809789965e-06, |
| "loss": 0.5592, |
| "step": 693 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.1355089346874902e-06, |
| "loss": 0.6017, |
| "step": 694 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.1171738725086833e-06, |
| "loss": 0.5109, |
| "step": 695 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.0989792843954196e-06, |
| "loss": 0.5385, |
| "step": 696 |
| }, |
| { |
| "epoch": 0.85, |
| "learning_rate": 1.080925458079074e-06, |
| "loss": 0.5643, |
| "step": 697 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.0630126790649998e-06, |
| "loss": 0.5271, |
| "step": 698 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.0452412306280023e-06, |
| "loss": 0.3891, |
| "step": 699 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.0276113938078768e-06, |
| "loss": 0.534, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 1.010123447404956e-06, |
| "loss": 0.5535, |
| "step": 701 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 9.927776679756918e-07, |
| "loss": 0.6365, |
| "step": 702 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 9.755743298283027e-07, |
| "loss": 0.5563, |
| "step": 703 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 9.585137050184124e-07, |
| "loss": 0.5016, |
| "step": 704 |
| }, |
| { |
| "epoch": 0.86, |
| "learning_rate": 9.415960633447674e-07, |
| "loss": 0.5933, |
| "step": 705 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 9.248216723449599e-07, |
| "loss": 0.5801, |
| "step": 706 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 9.081907972911941e-07, |
| "loss": 0.5007, |
| "step": 707 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 8.917037011861041e-07, |
| "loss": 0.5454, |
| "step": 708 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 8.753606447585816e-07, |
| "loss": 0.4534, |
| "step": 709 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 8.591618864596541e-07, |
| "loss": 0.4756, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 8.431076824584106e-07, |
| "loss": 0.6238, |
| "step": 711 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 8.271982866379369e-07, |
| "loss": 0.4893, |
| "step": 712 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 8.114339505913016e-07, |
| "loss": 0.4915, |
| "step": 713 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.95814923617586e-07, |
| "loss": 0.4149, |
| "step": 714 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.803414527179343e-07, |
| "loss": 0.5395, |
| "step": 715 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.650137825916482e-07, |
| "loss": 0.5657, |
| "step": 716 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.49832155632324e-07, |
| "loss": 0.525, |
| "step": 717 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.347968119240068e-07, |
| "loss": 0.3744, |
| "step": 718 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.199079892374072e-07, |
| "loss": 0.5329, |
| "step": 719 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 7.051659230261299e-07, |
| "loss": 0.5871, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.88, |
| "learning_rate": 6.905708464229588e-07, |
| "loss": 0.5257, |
| "step": 721 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.761229902361677e-07, |
| "loss": 0.5626, |
| "step": 722 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.618225829458647e-07, |
| "loss": 0.5072, |
| "step": 723 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.476698507003864e-07, |
| "loss": 0.5223, |
| "step": 724 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.336650173127224e-07, |
| "loss": 0.4932, |
| "step": 725 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.198083042569625e-07, |
| "loss": 0.6346, |
| "step": 726 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 6.060999306648129e-07, |
| "loss": 0.6828, |
| "step": 727 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 5.925401133221198e-07, |
| "loss": 0.5541, |
| "step": 728 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 5.791290666654392e-07, |
| "loss": 0.4976, |
| "step": 729 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 5.658670027786561e-07, |
| "loss": 0.6199, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 5.527541313896179e-07, |
| "loss": 0.4479, |
| "step": 731 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 5.397906598668301e-07, |
| "loss": 0.5514, |
| "step": 732 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 5.269767932161696e-07, |
| "loss": 0.4535, |
| "step": 733 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 5.143127340776389e-07, |
| "loss": 0.6185, |
| "step": 734 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 5.017986827221733e-07, |
| "loss": 0.4964, |
| "step": 735 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 4.894348370484648e-07, |
| "loss": 0.4748, |
| "step": 736 |
| }, |
| { |
| "epoch": 0.9, |
| "learning_rate": 4.772213925798331e-07, |
| "loss": 0.5474, |
| "step": 737 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 4.651585424611382e-07, |
| "loss": 0.5774, |
| "step": 738 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 4.5324647745572103e-07, |
| "loss": 0.4721, |
| "step": 739 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 4.4148538594239176e-07, |
| "loss": 0.4261, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 4.2987545391244613e-07, |
| "loss": 0.5503, |
| "step": 741 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 4.1841686496672306e-07, |
| "loss": 0.5644, |
| "step": 742 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 4.071098003127072e-07, |
| "loss": 0.4625, |
| "step": 743 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.959544387616632e-07, |
| "loss": 0.5904, |
| "step": 744 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.8495095672579584e-07, |
| "loss": 0.56, |
| "step": 745 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 3.74099528215478e-07, |
| "loss": 0.5291, |
| "step": 746 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.634003248364881e-07, |
| "loss": 0.5246, |
| "step": 747 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.5285351578729254e-07, |
| "loss": 0.5326, |
| "step": 748 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.424592678563843e-07, |
| "loss": 0.3966, |
| "step": 749 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.322177454196285e-07, |
| "loss": 0.491, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.221291104376789e-07, |
| "loss": 0.5086, |
| "step": 751 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.121935224534034e-07, |
| "loss": 0.4697, |
| "step": 752 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.024111385893691e-07, |
| "loss": 0.6453, |
| "step": 753 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 2.927821135453535e-07, |
| "loss": 0.5405, |
| "step": 754 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 2.8330659959589944e-07, |
| "loss": 0.523, |
| "step": 755 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 2.739847465879097e-07, |
| "loss": 0.553, |
| "step": 756 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 2.6481670193827413e-07, |
| "loss": 0.5055, |
| "step": 757 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 2.558026106315348e-07, |
| "loss": 0.4959, |
| "step": 758 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 2.469426152176013e-07, |
| "loss": 0.3971, |
| "step": 759 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 2.3823685580949273e-07, |
| "loss": 0.4743, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 2.2968547008111796e-07, |
| "loss": 0.6122, |
| "step": 761 |
| }, |
| { |
| "epoch": 0.93, |
| "learning_rate": 2.2128859326510765e-07, |
| "loss": 0.6303, |
| "step": 762 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 2.1304635815066587e-07, |
| "loss": 0.5057, |
| "step": 763 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 2.0495889508147625e-07, |
| "loss": 0.5201, |
| "step": 764 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 1.9702633195363918e-07, |
| "loss": 0.5643, |
| "step": 765 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 1.8924879421364672e-07, |
| "loss": 0.6162, |
| "step": 766 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 1.8162640485640425e-07, |
| "loss": 0.55, |
| "step": 767 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 1.74159284423282e-07, |
| "loss": 0.5446, |
| "step": 768 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 1.6684755100020435e-07, |
| "loss": 0.5262, |
| "step": 769 |
| }, |
| { |
| "epoch": 0.94, |
| "learning_rate": 1.5969132021579347e-07, |
| "loss": 0.4811, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 1.52690705239531e-07, |
| "loss": 0.4781, |
| "step": 771 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 1.4584581677997035e-07, |
| "loss": 0.4447, |
| "step": 772 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 1.3915676308299154e-07, |
| "loss": 0.5534, |
| "step": 773 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 1.3262364993008146e-07, |
| "loss": 0.4286, |
| "step": 774 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 1.262465806366664e-07, |
| "loss": 0.5617, |
| "step": 775 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 1.2002565605047646e-07, |
| "loss": 0.5761, |
| "step": 776 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 1.1396097454994925e-07, |
| "loss": 0.4448, |
| "step": 777 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 1.080526320426789e-07, |
| "loss": 0.5825, |
| "step": 778 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 1.0230072196389384e-07, |
| "loss": 0.5849, |
| "step": 779 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 9.670533527498139e-08, |
| "loss": 0.4981, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 9.126656046204996e-08, |
| "loss": 0.4984, |
| "step": 781 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 8.598448353453137e-08, |
| "loss": 0.6348, |
| "step": 782 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 8.085918802381298e-08, |
| "loss": 0.6819, |
| "step": 783 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 7.589075498192878e-08, |
| "loss": 0.5116, |
| "step": 784 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 7.10792629802659e-08, |
| "loss": 0.6499, |
| "step": 785 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 6.642478810833019e-08, |
| "loss": 0.5087, |
| "step": 786 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 6.192740397253926e-08, |
| "loss": 0.5191, |
| "step": 787 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 5.7587181695057945e-08, |
| "loss": 0.5053, |
| "step": 788 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 5.340418991267804e-08, |
| "loss": 0.5209, |
| "step": 789 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 4.937849477572587e-08, |
| "loss": 0.5879, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 4.551015994702201e-08, |
| "loss": 0.5049, |
| "step": 791 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 4.179924660087098e-08, |
| "loss": 0.7043, |
| "step": 792 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.8245813422098653e-08, |
| "loss": 0.5545, |
| "step": 793 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.4849916605117496e-08, |
| "loss": 0.4486, |
| "step": 794 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.161160985304168e-08, |
| "loss": 0.5127, |
| "step": 795 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.8530944376838898e-08, |
| "loss": 0.509, |
| "step": 796 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.5607968894518775e-08, |
| "loss": 0.5205, |
| "step": 797 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.2842729630361272e-08, |
| "loss": 0.5652, |
| "step": 798 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 2.023527031418726e-08, |
| "loss": 0.531, |
| "step": 799 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 1.77856321806702e-08, |
| "loss": 0.4995, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 1.5493853968675533e-08, |
| "loss": 0.5647, |
| "step": 801 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 1.3359971920655634e-08, |
| "loss": 0.5119, |
| "step": 802 |
| }, |
| { |
| "epoch": 0.98, |
| "learning_rate": 1.1384019782075818e-08, |
| "loss": 0.5839, |
| "step": 803 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 9.56602880087587e-09, |
| "loss": 0.5425, |
| "step": 804 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 7.906027726981568e-09, |
| "loss": 0.6194, |
| "step": 805 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 6.404042811846145e-09, |
| "loss": 0.5197, |
| "step": 806 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 5.0600978080384e-09, |
| "loss": 0.499, |
| "step": 807 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 3.874213968863005e-09, |
| "loss": 0.5129, |
| "step": 808 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 2.846410048027437e-09, |
| "loss": 0.4216, |
| "step": 809 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 1.9767022993444353e-09, |
| "loss": 0.5141, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.99, |
| "learning_rate": 1.265104476476653e-09, |
| "loss": 0.5714, |
| "step": 811 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 7.116278327157223e-10, |
| "loss": 0.5361, |
| "step": 812 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 3.1628112080683837e-10, |
| "loss": 0.5643, |
| "step": 813 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 7.90705928099822e-11, |
| "loss": 0.5627, |
| "step": 814 |
| }, |
| { |
| "epoch": 1.0, |
| "learning_rate": 0.0, |
| "loss": 0.429, |
| "step": 815 |
| }, |
| { |
| "epoch": 1.0, |
| "step": 815, |
| "total_flos": 1.5252221738523034e+18, |
| "train_loss": 0.566269972463327, |
| "train_runtime": 22903.1396, |
| "train_samples_per_second": 4.557, |
| "train_steps_per_second": 0.036 |
| } |
| ], |
| "max_steps": 815, |
| "num_train_epochs": 1, |
| "total_flos": 1.5252221738523034e+18, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|