{ "best_metric": 0.5480253402747288, "best_model_checkpoint": "esm2_t12_35M-lora-binding-sites_2023-09-13_00-24-11/checkpoint-270198", "epoch": 3.0, "eval_steps": 500, "global_step": 270198, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 0.0005701566148067436, "loss": 0.4436, "step": 200 }, { "epoch": 0.0, "learning_rate": 0.0005701560386498644, "loss": 0.1877, "step": 400 }, { "epoch": 0.01, "learning_rate": 0.0005701550771046327, "loss": 0.1187, "step": 600 }, { "epoch": 0.01, "learning_rate": 0.0005701537301723485, "loss": 0.1312, "step": 800 }, { "epoch": 0.01, "learning_rate": 0.0005701519978548325, "loss": 0.1103, "step": 1000 }, { "epoch": 0.01, "learning_rate": 0.0005701498801544267, "loss": 0.1073, "step": 1200 }, { "epoch": 0.02, "learning_rate": 0.0005701473770739939, "loss": 0.0926, "step": 1400 }, { "epoch": 0.02, "learning_rate": 0.000570144488616918, "loss": 0.0842, "step": 1600 }, { "epoch": 0.02, "learning_rate": 0.0005701412147871038, "loss": 0.0846, "step": 1800 }, { "epoch": 0.02, "learning_rate": 0.0005701375555889771, "loss": 0.073, "step": 2000 }, { "epoch": 0.02, "learning_rate": 0.0005701335322088747, "loss": 0.0787, "step": 2200 }, { "epoch": 0.03, "learning_rate": 0.0005701291042162594, "loss": 0.084, "step": 2400 }, { "epoch": 0.03, "learning_rate": 0.0005701242908717034, "loss": 0.0826, "step": 2600 }, { "epoch": 0.03, "learning_rate": 0.0005701190921817139, "loss": 0.0625, "step": 2800 }, { "epoch": 0.03, "learning_rate": 0.0005701135081533188, "loss": 0.0601, "step": 3000 }, { "epoch": 0.04, "learning_rate": 0.0005701075387940671, "loss": 0.0628, "step": 3200 }, { "epoch": 0.04, "learning_rate": 0.0005701012168439152, "loss": 0.0897, "step": 3400 }, { "epoch": 0.04, "learning_rate": 0.0005700944787742291, "loss": 0.0656, "step": 3600 }, { "epoch": 0.04, "learning_rate": 0.0005700873553994116, "loss": 0.0627, "step": 3800 }, { "epoch": 0.04, "learning_rate": 0.0005700798467290926, "loss": 0.0645, "step": 4000 }, { "epoch": 0.05, "learning_rate": 0.0005700719527734229, "loss": 0.0674, "step": 4200 }, { "epoch": 0.05, "learning_rate": 0.0005700636735430738, "loss": 0.0641, "step": 4400 }, { "epoch": 0.05, "learning_rate": 0.0005700550090492382, "loss": 0.058, "step": 4600 }, { "epoch": 0.05, "learning_rate": 0.0005700459593036292, "loss": 0.0581, "step": 4800 }, { "epoch": 0.06, "learning_rate": 0.0005700365243184809, "loss": 0.0476, "step": 5000 }, { "epoch": 0.06, "learning_rate": 0.0005700267041065481, "loss": 0.0527, "step": 5200 }, { "epoch": 0.06, "learning_rate": 0.0005700164986811065, "loss": 0.0612, "step": 5400 }, { "epoch": 0.06, "learning_rate": 0.0005700059080559526, "loss": 0.0528, "step": 5600 }, { "epoch": 0.06, "learning_rate": 0.0005699949322454036, "loss": 0.065, "step": 5800 }, { "epoch": 0.07, "learning_rate": 0.0005699835712642972, "loss": 0.0473, "step": 6000 }, { "epoch": 0.07, "learning_rate": 0.0005699718251279921, "loss": 0.0607, "step": 6200 }, { "epoch": 0.07, "learning_rate": 0.0005699596938523677, "loss": 0.0503, "step": 6400 }, { "epoch": 0.07, "learning_rate": 0.0005699471774538237, "loss": 0.0633, "step": 6600 }, { "epoch": 0.08, "learning_rate": 0.0005699342759492807, "loss": 0.052, "step": 6800 }, { "epoch": 0.08, "learning_rate": 0.0005699209893561799, "loss": 0.048, "step": 7000 }, { "epoch": 0.08, "learning_rate": 0.0005699073176924832, "loss": 0.0506, "step": 7200 }, { "epoch": 0.08, "learning_rate": 0.0005698933322180377, "loss": 0.0727, "step": 7400 }, { "epoch": 0.08, "learning_rate": 0.000569878892394234, "loss": 0.0464, "step": 7600 }, { "epoch": 0.09, "learning_rate": 0.000569864067556744, "loss": 0.0436, "step": 7800 }, { "epoch": 0.09, "learning_rate": 0.0005698489347324025, "loss": 0.0564, "step": 8000 }, { "epoch": 0.09, "learning_rate": 0.0005698333418529975, "loss": 0.0503, "step": 8200 }, { "epoch": 0.09, "learning_rate": 0.0005698173640214845, "loss": 0.054, "step": 8400 }, { "epoch": 0.1, "learning_rate": 0.0005698010012594639, "loss": 0.0548, "step": 8600 }, { "epoch": 0.1, "learning_rate": 0.0005697842535890557, "loss": 0.0625, "step": 8800 }, { "epoch": 0.1, "learning_rate": 0.0005697671210329006, "loss": 0.0547, "step": 9000 }, { "epoch": 0.1, "learning_rate": 0.0005697496036141599, "loss": 0.0458, "step": 9200 }, { "epoch": 0.1, "learning_rate": 0.0005697317013565146, "loss": 0.0381, "step": 9400 }, { "epoch": 0.11, "learning_rate": 0.0005697134142841663, "loss": 0.0493, "step": 9600 }, { "epoch": 0.11, "learning_rate": 0.000569694742421837, "loss": 0.0441, "step": 9800 }, { "epoch": 0.11, "learning_rate": 0.0005696756857947685, "loss": 0.0496, "step": 10000 }, { "epoch": 0.11, "learning_rate": 0.0005696562444287228, "loss": 0.0453, "step": 10200 }, { "epoch": 0.12, "learning_rate": 0.0005696364183499823, "loss": 0.0599, "step": 10400 }, { "epoch": 0.12, "learning_rate": 0.0005696162075853492, "loss": 0.0482, "step": 10600 }, { "epoch": 0.12, "learning_rate": 0.0005695957160960544, "loss": 0.056, "step": 10800 }, { "epoch": 0.12, "learning_rate": 0.0005695747379652068, "loss": 0.0673, "step": 11000 }, { "epoch": 0.12, "learning_rate": 0.0005695533752318507, "loss": 0.0419, "step": 11200 }, { "epoch": 0.13, "learning_rate": 0.0005695316279248656, "loss": 0.0457, "step": 11400 }, { "epoch": 0.13, "learning_rate": 0.0005695094960736512, "loss": 0.0658, "step": 11600 }, { "epoch": 0.13, "learning_rate": 0.0005694869797081268, "loss": 0.0485, "step": 11800 }, { "epoch": 0.13, "learning_rate": 0.0005694640788587318, "loss": 0.0581, "step": 12000 }, { "epoch": 0.14, "learning_rate": 0.0005694407935564252, "loss": 0.0523, "step": 12200 }, { "epoch": 0.14, "learning_rate": 0.0005694171238326854, "loss": 0.0488, "step": 12400 }, { "epoch": 0.14, "learning_rate": 0.0005693930697195113, "loss": 0.0434, "step": 12600 }, { "epoch": 0.14, "learning_rate": 0.0005693686312494207, "loss": 0.0441, "step": 12800 }, { "epoch": 0.14, "learning_rate": 0.0005693439335253714, "loss": 0.0474, "step": 13000 }, { "epoch": 0.15, "learning_rate": 0.0005693187283624477, "loss": 0.0466, "step": 13200 }, { "epoch": 0.15, "learning_rate": 0.0005692931389431076, "loss": 0.0504, "step": 13400 }, { "epoch": 0.15, "learning_rate": 0.0005692671653019445, "loss": 0.0574, "step": 13600 }, { "epoch": 0.15, "learning_rate": 0.0005692408074740716, "loss": 0.0455, "step": 13800 }, { "epoch": 0.16, "learning_rate": 0.0005692140654951213, "loss": 0.047, "step": 14000 }, { "epoch": 0.16, "learning_rate": 0.0005691869394012452, "loss": 0.0399, "step": 14200 }, { "epoch": 0.16, "learning_rate": 0.0005691594292291145, "loss": 0.046, "step": 14400 }, { "epoch": 0.16, "learning_rate": 0.0005691315350159191, "loss": 0.0509, "step": 14600 }, { "epoch": 0.16, "learning_rate": 0.0005691032567993687, "loss": 0.0493, "step": 14800 }, { "epoch": 0.17, "learning_rate": 0.0005690745946176916, "loss": 0.0433, "step": 15000 }, { "epoch": 0.17, "learning_rate": 0.0005690455485096356, "loss": 0.0525, "step": 15200 }, { "epoch": 0.17, "learning_rate": 0.0005690161185144672, "loss": 0.0487, "step": 15400 }, { "epoch": 0.17, "learning_rate": 0.000568986304671972, "loss": 0.0493, "step": 15600 }, { "epoch": 0.18, "learning_rate": 0.0005689562589653545, "loss": 0.0596, "step": 15800 }, { "epoch": 0.18, "learning_rate": 0.0005689256794683665, "loss": 0.0381, "step": 16000 }, { "epoch": 0.18, "learning_rate": 0.0005688947162463136, "loss": 0.0463, "step": 16200 }, { "epoch": 0.18, "learning_rate": 0.0005688633693410542, "loss": 0.0411, "step": 16400 }, { "epoch": 0.18, "learning_rate": 0.0005688316387949652, "loss": 0.0532, "step": 16600 }, { "epoch": 0.19, "learning_rate": 0.0005687995246509422, "loss": 0.0407, "step": 16800 }, { "epoch": 0.19, "learning_rate": 0.0005687670269523996, "loss": 0.0426, "step": 17000 }, { "epoch": 0.19, "learning_rate": 0.0005687341457432698, "loss": 0.0415, "step": 17200 }, { "epoch": 0.19, "learning_rate": 0.0005687008810680043, "loss": 0.0449, "step": 17400 }, { "epoch": 0.2, "learning_rate": 0.0005686672329715724, "loss": 0.0411, "step": 17600 }, { "epoch": 0.2, "learning_rate": 0.0005686332014994621, "loss": 0.0465, "step": 17800 }, { "epoch": 0.2, "learning_rate": 0.0005685987866976797, "loss": 0.0386, "step": 18000 }, { "epoch": 0.2, "learning_rate": 0.0005685639886127493, "loss": 0.0379, "step": 18200 }, { "epoch": 0.2, "learning_rate": 0.0005685288072917138, "loss": 0.05, "step": 18400 }, { "epoch": 0.21, "learning_rate": 0.0005684934215577838, "loss": 0.0409, "step": 18600 }, { "epoch": 0.21, "learning_rate": 0.0005684574758233195, "loss": 0.0417, "step": 18800 }, { "epoch": 0.21, "learning_rate": 0.0005684211469967415, "loss": 0.0418, "step": 19000 }, { "epoch": 0.21, "learning_rate": 0.0005683844351271619, "loss": 0.0478, "step": 19200 }, { "epoch": 0.22, "learning_rate": 0.0005683473402642101, "loss": 0.04, "step": 19400 }, { "epoch": 0.22, "learning_rate": 0.0005683098624580339, "loss": 0.0448, "step": 19600 }, { "epoch": 0.22, "learning_rate": 0.0005682720017592983, "loss": 0.0374, "step": 19800 }, { "epoch": 0.22, "learning_rate": 0.0005682337582191861, "loss": 0.0382, "step": 20000 }, { "epoch": 0.22, "learning_rate": 0.0005681951318893979, "loss": 0.0431, "step": 20200 }, { "epoch": 0.23, "learning_rate": 0.0005681561228221512, "loss": 0.0437, "step": 20400 }, { "epoch": 0.23, "learning_rate": 0.0005681167310701815, "loss": 0.0421, "step": 20600 }, { "epoch": 0.23, "learning_rate": 0.0005680769566867412, "loss": 0.0441, "step": 20800 }, { "epoch": 0.23, "learning_rate": 0.0005680370014619785, "loss": 0.0625, "step": 21000 }, { "epoch": 0.24, "learning_rate": 0.0005679964638899052, "loss": 0.0418, "step": 21200 }, { "epoch": 0.24, "learning_rate": 0.0005679555438489471, "loss": 0.0454, "step": 21400 }, { "epoch": 0.24, "learning_rate": 0.0005679142413944229, "loss": 0.0416, "step": 21600 }, { "epoch": 0.24, "learning_rate": 0.0005678725565821682, "loss": 0.0462, "step": 21800 }, { "epoch": 0.24, "learning_rate": 0.0005678304894685357, "loss": 0.0456, "step": 22000 }, { "epoch": 0.25, "learning_rate": 0.0005677880401103944, "loss": 0.0454, "step": 22200 }, { "epoch": 0.25, "learning_rate": 0.0005677452085651307, "loss": 0.0422, "step": 22400 }, { "epoch": 0.25, "learning_rate": 0.0005677019948906472, "loss": 0.039, "step": 22600 }, { "epoch": 0.25, "learning_rate": 0.0005676583991453631, "loss": 0.0356, "step": 22800 }, { "epoch": 0.26, "learning_rate": 0.0005676144213882145, "loss": 0.0386, "step": 23000 }, { "epoch": 0.26, "learning_rate": 0.000567570284427209, "loss": 0.0431, "step": 23200 }, { "epoch": 0.26, "learning_rate": 0.0005675255447345165, "loss": 0.0377, "step": 23400 }, { "epoch": 0.26, "learning_rate": 0.0005674804232095615, "loss": 0.0419, "step": 23600 }, { "epoch": 0.26, "learning_rate": 0.0005674349199133426, "loss": 0.04, "step": 23800 }, { "epoch": 0.27, "learning_rate": 0.0005673890349073741, "loss": 0.05, "step": 24000 }, { "epoch": 0.27, "learning_rate": 0.0005673427682536871, "loss": 0.0306, "step": 24200 }, { "epoch": 0.27, "learning_rate": 0.0005672961200148277, "loss": 0.0366, "step": 24400 }, { "epoch": 0.27, "learning_rate": 0.0005672490902538587, "loss": 0.0463, "step": 24600 }, { "epoch": 0.28, "learning_rate": 0.0005672016790343581, "loss": 0.0421, "step": 24800 }, { "epoch": 0.28, "learning_rate": 0.0005671538864204198, "loss": 0.0478, "step": 25000 }, { "epoch": 0.28, "learning_rate": 0.0005671059542948227, "loss": 0.0521, "step": 25200 }, { "epoch": 0.28, "learning_rate": 0.0005670574009925137, "loss": 0.0449, "step": 25400 }, { "epoch": 0.28, "learning_rate": 0.0005670084664908124, "loss": 0.0485, "step": 25600 }, { "epoch": 0.29, "learning_rate": 0.0005669591508558719, "loss": 0.0308, "step": 25800 }, { "epoch": 0.29, "learning_rate": 0.0005669094541543607, "loss": 0.0354, "step": 26000 }, { "epoch": 0.29, "learning_rate": 0.0005668596277895911, "loss": 0.0415, "step": 26200 }, { "epoch": 0.29, "learning_rate": 0.0005668091710614934, "loss": 0.0555, "step": 26400 }, { "epoch": 0.3, "learning_rate": 0.0005667583334695783, "loss": 0.0399, "step": 26600 }, { "epoch": 0.3, "learning_rate": 0.000566707115082572, "loss": 0.052, "step": 26800 }, { "epoch": 0.3, "learning_rate": 0.0005666555159697149, "loss": 0.0398, "step": 27000 }, { "epoch": 0.3, "learning_rate": 0.0005666035362007624, "loss": 0.053, "step": 27200 }, { "epoch": 0.3, "learning_rate": 0.0005665511758459846, "loss": 0.0454, "step": 27400 }, { "epoch": 0.31, "learning_rate": 0.0005664984349761658, "loss": 0.0322, "step": 27600 }, { "epoch": 0.31, "learning_rate": 0.0005664453136626051, "loss": 0.0508, "step": 27800 }, { "epoch": 0.31, "learning_rate": 0.0005663918119771154, "loss": 0.0366, "step": 28000 }, { "epoch": 0.31, "learning_rate": 0.0005663379299920243, "loss": 0.0429, "step": 28200 }, { "epoch": 0.32, "learning_rate": 0.0005662836677801735, "loss": 0.0375, "step": 28400 }, { "epoch": 0.32, "learning_rate": 0.0005662290254149182, "loss": 0.0408, "step": 28600 }, { "epoch": 0.32, "learning_rate": 0.0005661740029701281, "loss": 0.0417, "step": 28800 }, { "epoch": 0.32, "learning_rate": 0.0005661186005201862, "loss": 0.0442, "step": 29000 }, { "epoch": 0.32, "learning_rate": 0.0005660628181399899, "loss": 0.0393, "step": 29200 }, { "epoch": 0.33, "learning_rate": 0.0005660066559049495, "loss": 0.0413, "step": 29400 }, { "epoch": 0.33, "learning_rate": 0.0005659501138909893, "loss": 0.036, "step": 29600 }, { "epoch": 0.33, "learning_rate": 0.0005658931921745466, "loss": 0.0391, "step": 29800 }, { "epoch": 0.33, "learning_rate": 0.0005658358908325724, "loss": 0.035, "step": 30000 }, { "epoch": 0.34, "learning_rate": 0.000565778209942531, "loss": 0.0397, "step": 30200 }, { "epoch": 0.34, "learning_rate": 0.0005657201495823991, "loss": 0.0363, "step": 30400 }, { "epoch": 0.34, "learning_rate": 0.0005656620029730321, "loss": 0.0536, "step": 30600 }, { "epoch": 0.34, "learning_rate": 0.0005656031858050686, "loss": 0.0525, "step": 30800 }, { "epoch": 0.34, "learning_rate": 0.0005655439894036252, "loss": 0.0343, "step": 31000 }, { "epoch": 0.35, "learning_rate": 0.0005654844138487275, "loss": 0.0463, "step": 31200 }, { "epoch": 0.35, "learning_rate": 0.0005654244592209141, "loss": 0.0399, "step": 31400 }, { "epoch": 0.35, "learning_rate": 0.0005653641256012361, "loss": 0.0584, "step": 31600 }, { "epoch": 0.35, "learning_rate": 0.0005653034130712565, "loss": 0.0398, "step": 31800 }, { "epoch": 0.36, "learning_rate": 0.0005652423217130511, "loss": 0.0473, "step": 32000 }, { "epoch": 0.36, "learning_rate": 0.0005651808516092075, "loss": 0.043, "step": 32200 }, { "epoch": 0.36, "learning_rate": 0.0005651190028428255, "loss": 0.0478, "step": 32400 }, { "epoch": 0.36, "learning_rate": 0.000565057087575819, "loss": 0.0531, "step": 32600 }, { "epoch": 0.36, "learning_rate": 0.0005649944836279704, "loss": 0.0346, "step": 32800 }, { "epoch": 0.37, "learning_rate": 0.0005649315012695288, "loss": 0.0452, "step": 33000 }, { "epoch": 0.37, "learning_rate": 0.0005648681405856382, "loss": 0.0385, "step": 33200 }, { "epoch": 0.37, "learning_rate": 0.0005648044016619544, "loss": 0.0418, "step": 33400 }, { "epoch": 0.37, "learning_rate": 0.0005647402845846439, "loss": 0.0386, "step": 33600 }, { "epoch": 0.38, "learning_rate": 0.0005646757894403851, "loss": 0.0363, "step": 33800 }, { "epoch": 0.38, "learning_rate": 0.0005646109163163668, "loss": 0.0398, "step": 34000 }, { "epoch": 0.38, "learning_rate": 0.0005645456653002893, "loss": 0.0469, "step": 34200 }, { "epoch": 0.38, "learning_rate": 0.0005644800364803637, "loss": 0.0393, "step": 34400 }, { "epoch": 0.38, "learning_rate": 0.0005644143609174058, "loss": 0.0453, "step": 34600 }, { "epoch": 0.39, "learning_rate": 0.0005643479786443663, "loss": 0.0347, "step": 34800 }, { "epoch": 0.39, "learning_rate": 0.0005642812188347254, "loss": 0.037, "step": 35000 }, { "epoch": 0.39, "learning_rate": 0.0005642140815787342, "loss": 0.036, "step": 35200 }, { "epoch": 0.39, "learning_rate": 0.0005641465669671531, "loss": 0.057, "step": 35400 }, { "epoch": 0.4, "learning_rate": 0.0005640786750912536, "loss": 0.0368, "step": 35600 }, { "epoch": 0.4, "learning_rate": 0.0005640104060428169, "loss": 0.0402, "step": 35800 }, { "epoch": 0.4, "learning_rate": 0.0005639417599141336, "loss": 0.0456, "step": 36000 }, { "epoch": 0.4, "learning_rate": 0.0005638727367980045, "loss": 0.0489, "step": 36200 }, { "epoch": 0.4, "learning_rate": 0.0005638033367877403, "loss": 0.041, "step": 36400 }, { "epoch": 0.41, "learning_rate": 0.0005637335599771606, "loss": 0.0665, "step": 36600 }, { "epoch": 0.41, "learning_rate": 0.0005636634064605948, "loss": 0.0418, "step": 36800 }, { "epoch": 0.41, "learning_rate": 0.0005635928763328815, "loss": 0.0341, "step": 37000 }, { "epoch": 0.41, "learning_rate": 0.0005635223251590105, "loss": 0.0583, "step": 37200 }, { "epoch": 0.42, "learning_rate": 0.0005634510439774146, "loss": 0.0392, "step": 37400 }, { "epoch": 0.42, "learning_rate": 0.0005633793864717577, "loss": 0.0599, "step": 37600 }, { "epoch": 0.42, "learning_rate": 0.0005633073527389121, "loss": 0.0436, "step": 37800 }, { "epoch": 0.42, "learning_rate": 0.0005632349428762577, "loss": 0.037, "step": 38000 }, { "epoch": 0.42, "learning_rate": 0.0005631621569816834, "loss": 0.0318, "step": 38200 }, { "epoch": 0.43, "learning_rate": 0.0005630889951535865, "loss": 0.0625, "step": 38400 }, { "epoch": 0.43, "learning_rate": 0.0005630154574908725, "loss": 0.0433, "step": 38600 }, { "epoch": 0.43, "learning_rate": 0.0005629415440929546, "loss": 0.0314, "step": 38800 }, { "epoch": 0.43, "learning_rate": 0.0005628672550597543, "loss": 0.0363, "step": 39000 }, { "epoch": 0.44, "learning_rate": 0.0005627925904917011, "loss": 0.05, "step": 39200 }, { "epoch": 0.44, "learning_rate": 0.0005627175504897316, "loss": 0.0477, "step": 39400 }, { "epoch": 0.44, "learning_rate": 0.0005626425131654336, "loss": 0.0529, "step": 39600 }, { "epoch": 0.44, "learning_rate": 0.0005625667244763705, "loss": 0.0282, "step": 39800 }, { "epoch": 0.44, "learning_rate": 0.0005624905606587331, "loss": 0.0447, "step": 40000 }, { "epoch": 0.45, "learning_rate": 0.0005624140218154851, "loss": 0.0393, "step": 40200 }, { "epoch": 0.45, "learning_rate": 0.0005623371080500971, "loss": 0.0411, "step": 40400 }, { "epoch": 0.45, "learning_rate": 0.0005622598194665467, "loss": 0.0455, "step": 40600 }, { "epoch": 0.45, "learning_rate": 0.000562182156169318, "loss": 0.0378, "step": 40800 }, { "epoch": 0.46, "learning_rate": 0.0005621041182634019, "loss": 0.0384, "step": 41000 }, { "epoch": 0.46, "learning_rate": 0.0005620257058542957, "loss": 0.0497, "step": 41200 }, { "epoch": 0.46, "learning_rate": 0.0005619469190480028, "loss": 0.0352, "step": 41400 }, { "epoch": 0.46, "learning_rate": 0.0005618677579510328, "loss": 0.0429, "step": 41600 }, { "epoch": 0.46, "learning_rate": 0.0005617886212774089, "loss": 0.0566, "step": 41800 }, { "epoch": 0.47, "learning_rate": 0.0005617087137907502, "loss": 0.0357, "step": 42000 }, { "epoch": 0.47, "learning_rate": 0.0005616284323354373, "loss": 0.0534, "step": 42200 }, { "epoch": 0.47, "learning_rate": 0.0005615477770200006, "loss": 0.0653, "step": 42400 }, { "epoch": 0.47, "learning_rate": 0.0005614667479534755, "loss": 0.057, "step": 42600 }, { "epoch": 0.48, "learning_rate": 0.0005613853452454032, "loss": 0.0451, "step": 42800 }, { "epoch": 0.48, "learning_rate": 0.0005613035690058294, "loss": 0.0394, "step": 43000 }, { "epoch": 0.48, "learning_rate": 0.0005612214193453053, "loss": 0.044, "step": 43200 }, { "epoch": 0.48, "learning_rate": 0.0005611388963748866, "loss": 0.0376, "step": 43400 }, { "epoch": 0.48, "learning_rate": 0.0005610560002061337, "loss": 0.0443, "step": 43600 }, { "epoch": 0.49, "learning_rate": 0.0005609727309511118, "loss": 0.0435, "step": 43800 }, { "epoch": 0.49, "learning_rate": 0.0005608895078611188, "loss": 0.0332, "step": 44000 }, { "epoch": 0.49, "learning_rate": 0.0005608054946357917, "loss": 0.0486, "step": 44200 }, { "epoch": 0.49, "learning_rate": 0.0005607211086628469, "loss": 0.0341, "step": 44400 }, { "epoch": 0.5, "learning_rate": 0.0005606363500563632, "loss": 0.0381, "step": 44600 }, { "epoch": 0.5, "learning_rate": 0.0005605512189309235, "loss": 0.041, "step": 44800 }, { "epoch": 0.5, "learning_rate": 0.0005604657154016139, "loss": 0.0353, "step": 45000 }, { "epoch": 0.5, "learning_rate": 0.0005603798395840247, "loss": 0.0489, "step": 45200 }, { "epoch": 0.5, "learning_rate": 0.0005602935915942485, "loss": 0.0508, "step": 45400 }, { "epoch": 0.51, "learning_rate": 0.0005602069715488817, "loss": 0.0339, "step": 45600 }, { "epoch": 0.51, "learning_rate": 0.0005601199795650235, "loss": 0.0382, "step": 45800 }, { "epoch": 0.51, "learning_rate": 0.0005600326157602758, "loss": 0.0422, "step": 46000 }, { "epoch": 0.51, "learning_rate": 0.0005599448802527431, "loss": 0.0317, "step": 46200 }, { "epoch": 0.52, "learning_rate": 0.000559856773161033, "loss": 0.0398, "step": 46400 }, { "epoch": 0.52, "learning_rate": 0.0005597682946042545, "loss": 0.0389, "step": 46600 }, { "epoch": 0.52, "learning_rate": 0.0005596794447020196, "loss": 0.0422, "step": 46800 }, { "epoch": 0.52, "learning_rate": 0.0005595902235744417, "loss": 0.0353, "step": 47000 }, { "epoch": 0.52, "learning_rate": 0.0005595010802262207, "loss": 0.0427, "step": 47200 }, { "epoch": 0.53, "learning_rate": 0.0005594111188649213, "loss": 0.0357, "step": 47400 }, { "epoch": 0.53, "learning_rate": 0.000559320786641021, "loss": 0.0519, "step": 47600 }, { "epoch": 0.53, "learning_rate": 0.0005592300836766373, "loss": 0.0521, "step": 47800 }, { "epoch": 0.53, "learning_rate": 0.0005591390100943891, "loss": 0.0474, "step": 48000 }, { "epoch": 0.54, "learning_rate": 0.0005590475660173964, "loss": 0.043, "step": 48200 }, { "epoch": 0.54, "learning_rate": 0.0005589557515692793, "loss": 0.0485, "step": 48400 }, { "epoch": 0.54, "learning_rate": 0.0005588635668741596, "loss": 0.0355, "step": 48600 }, { "epoch": 0.54, "learning_rate": 0.0005587710120566592, "loss": 0.0333, "step": 48800 }, { "epoch": 0.54, "learning_rate": 0.0005586780872419001, "loss": 0.0435, "step": 49000 }, { "epoch": 0.55, "learning_rate": 0.0005585847925555047, "loss": 0.0405, "step": 49200 }, { "epoch": 0.55, "learning_rate": 0.0005584911281235955, "loss": 0.0409, "step": 49400 }, { "epoch": 0.55, "learning_rate": 0.0005583975651622658, "loss": 0.0394, "step": 49600 }, { "epoch": 0.55, "learning_rate": 0.0005583031634668374, "loss": 0.0541, "step": 49800 }, { "epoch": 0.56, "learning_rate": 0.0005582083924066216, "loss": 0.0369, "step": 50000 }, { "epoch": 0.56, "learning_rate": 0.0005581132521097367, "loss": 0.0335, "step": 50200 }, { "epoch": 0.56, "learning_rate": 0.0005580177427048001, "loss": 0.0558, "step": 50400 }, { "epoch": 0.56, "learning_rate": 0.0005579218643209284, "loss": 0.0388, "step": 50600 }, { "epoch": 0.56, "learning_rate": 0.0005578256170877366, "loss": 0.0408, "step": 50800 }, { "epoch": 0.57, "learning_rate": 0.0005577290011353388, "loss": 0.0396, "step": 51000 }, { "epoch": 0.57, "learning_rate": 0.0005576320165943473, "loss": 0.0371, "step": 51200 }, { "epoch": 0.57, "learning_rate": 0.0005575346635958727, "loss": 0.0431, "step": 51400 }, { "epoch": 0.57, "learning_rate": 0.000557437431794138, "loss": 0.0415, "step": 51600 }, { "epoch": 0.58, "learning_rate": 0.000557339344116661, "loss": 0.0284, "step": 51800 }, { "epoch": 0.58, "learning_rate": 0.0005572408883773566, "loss": 0.0447, "step": 52000 }, { "epoch": 0.58, "learning_rate": 0.0005571420647093241, "loss": 0.0466, "step": 52200 }, { "epoch": 0.58, "learning_rate": 0.0005570428732461607, "loss": 0.035, "step": 52400 }, { "epoch": 0.58, "learning_rate": 0.0005569433141219605, "loss": 0.0378, "step": 52600 }, { "epoch": 0.59, "learning_rate": 0.0005568433874713144, "loss": 0.0318, "step": 52800 }, { "epoch": 0.59, "learning_rate": 0.0005567430934293107, "loss": 0.0422, "step": 53000 }, { "epoch": 0.59, "learning_rate": 0.0005566424321315337, "loss": 0.0415, "step": 53200 }, { "epoch": 0.59, "learning_rate": 0.0005565414037140649, "loss": 0.0501, "step": 53400 }, { "epoch": 0.6, "learning_rate": 0.0005564400083134816, "loss": 0.0399, "step": 53600 }, { "epoch": 0.6, "learning_rate": 0.0005563382460668574, "loss": 0.0345, "step": 53800 }, { "epoch": 0.6, "learning_rate": 0.0005562361171117617, "loss": 0.0315, "step": 54000 }, { "epoch": 0.6, "learning_rate": 0.0005561336215862598, "loss": 0.0425, "step": 54200 }, { "epoch": 0.6, "learning_rate": 0.0005560307596289125, "loss": 0.0402, "step": 54400 }, { "epoch": 0.61, "learning_rate": 0.0005559280484309493, "loss": 0.0552, "step": 54600 }, { "epoch": 0.61, "learning_rate": 0.000555824455857993, "loss": 0.0539, "step": 54800 }, { "epoch": 0.61, "learning_rate": 0.0005557204972711436, "loss": 0.046, "step": 55000 }, { "epoch": 0.61, "learning_rate": 0.0005556161728109399, "loss": 0.0483, "step": 55200 }, { "epoch": 0.62, "learning_rate": 0.0005555114826184149, "loss": 0.0355, "step": 55400 }, { "epoch": 0.62, "learning_rate": 0.0005554064268350967, "loss": 0.0371, "step": 55600 }, { "epoch": 0.62, "learning_rate": 0.0005553010056030069, "loss": 0.0488, "step": 55800 }, { "epoch": 0.62, "learning_rate": 0.0005551952190646615, "loss": 0.0288, "step": 56000 }, { "epoch": 0.62, "learning_rate": 0.0005550890673630703, "loss": 0.034, "step": 56200 }, { "epoch": 0.63, "learning_rate": 0.0005549830841330922, "loss": 0.0469, "step": 56400 }, { "epoch": 0.63, "learning_rate": 0.0005548762043600333, "loss": 0.0463, "step": 56600 }, { "epoch": 0.63, "learning_rate": 0.0005547689598549955, "loss": 0.0567, "step": 56800 }, { "epoch": 0.63, "learning_rate": 0.0005546613507629596, "loss": 0.0432, "step": 57000 }, { "epoch": 0.64, "learning_rate": 0.0005545533772293995, "loss": 0.0486, "step": 57200 }, { "epoch": 0.64, "learning_rate": 0.0005544450394002813, "loss": 0.0362, "step": 57400 }, { "epoch": 0.64, "learning_rate": 0.0005543363374220639, "loss": 0.0489, "step": 57600 }, { "epoch": 0.64, "learning_rate": 0.0005542272714416986, "loss": 0.0479, "step": 57800 }, { "epoch": 0.64, "learning_rate": 0.0005541178416066283, "loss": 0.035, "step": 58000 }, { "epoch": 0.65, "learning_rate": 0.0005540080480647882, "loss": 0.0363, "step": 58200 }, { "epoch": 0.65, "learning_rate": 0.0005538978909646053, "loss": 0.0395, "step": 58400 }, { "epoch": 0.65, "learning_rate": 0.0005537873704549974, "loss": 0.0286, "step": 58600 }, { "epoch": 0.65, "learning_rate": 0.0005536764866853745, "loss": 0.0426, "step": 58800 }, { "epoch": 0.66, "learning_rate": 0.0005535652398056369, "loss": 0.0327, "step": 59000 }, { "epoch": 0.66, "learning_rate": 0.0005534536299661762, "loss": 0.0343, "step": 59200 }, { "epoch": 0.66, "learning_rate": 0.0005533416573178747, "loss": 0.0426, "step": 59400 }, { "epoch": 0.66, "learning_rate": 0.0005532293220121047, "loss": 0.0326, "step": 59600 }, { "epoch": 0.66, "learning_rate": 0.0005531166242007295, "loss": 0.0401, "step": 59800 }, { "epoch": 0.67, "learning_rate": 0.0005530035640361017, "loss": 0.0547, "step": 60000 }, { "epoch": 0.67, "learning_rate": 0.0005528901416710644, "loss": 0.0406, "step": 60200 }, { "epoch": 0.67, "learning_rate": 0.00055277635725895, "loss": 0.0392, "step": 60400 }, { "epoch": 0.67, "learning_rate": 0.0005526622109535803, "loss": 0.0331, "step": 60600 }, { "epoch": 0.68, "learning_rate": 0.000552548276349058, "loss": 0.0497, "step": 60800 }, { "epoch": 0.68, "learning_rate": 0.0005524334085281352, "loss": 0.0493, "step": 61000 }, { "epoch": 0.68, "learning_rate": 0.0005523181792775799, "loss": 0.0509, "step": 61200 }, { "epoch": 0.68, "learning_rate": 0.0005522025887531671, "loss": 0.0278, "step": 61400 }, { "epoch": 0.68, "learning_rate": 0.0005520866371111605, "loss": 0.0386, "step": 61600 }, { "epoch": 0.69, "learning_rate": 0.0005519703245083118, "loss": 0.0357, "step": 61800 }, { "epoch": 0.69, "learning_rate": 0.0005518536511018605, "loss": 0.0352, "step": 62000 }, { "epoch": 0.69, "learning_rate": 0.0005517366170495342, "loss": 0.0404, "step": 62200 }, { "epoch": 0.69, "learning_rate": 0.000551619222509548, "loss": 0.0434, "step": 62400 }, { "epoch": 0.7, "learning_rate": 0.0005515014676406042, "loss": 0.0357, "step": 62600 }, { "epoch": 0.7, "learning_rate": 0.0005513833526018923, "loss": 0.0374, "step": 62800 }, { "epoch": 0.7, "learning_rate": 0.0005512648775530885, "loss": 0.0444, "step": 63000 }, { "epoch": 0.7, "learning_rate": 0.0005511460426543559, "loss": 0.038, "step": 63200 }, { "epoch": 0.7, "learning_rate": 0.0005510268480663441, "loss": 0.0444, "step": 63400 }, { "epoch": 0.71, "learning_rate": 0.0005509072939501887, "loss": 0.0357, "step": 63600 }, { "epoch": 0.71, "learning_rate": 0.0005507879809285815, "loss": 0.0418, "step": 63800 }, { "epoch": 0.71, "learning_rate": 0.0005506677100371086, "loss": 0.0314, "step": 64000 }, { "epoch": 0.71, "learning_rate": 0.0005505470801030006, "loss": 0.0377, "step": 64200 }, { "epoch": 0.72, "learning_rate": 0.0005504260912893339, "loss": 0.0358, "step": 64400 }, { "epoch": 0.72, "learning_rate": 0.0005503047437596697, "loss": 0.0405, "step": 64600 }, { "epoch": 0.72, "learning_rate": 0.0005501830376780543, "loss": 0.0348, "step": 64800 }, { "epoch": 0.72, "learning_rate": 0.0005500609732090188, "loss": 0.0349, "step": 65000 }, { "epoch": 0.72, "learning_rate": 0.0005499385505175786, "loss": 0.0395, "step": 65200 }, { "epoch": 0.73, "learning_rate": 0.0005498157697692334, "loss": 0.0415, "step": 65400 }, { "epoch": 0.73, "learning_rate": 0.000549692631129967, "loss": 0.0302, "step": 65600 }, { "epoch": 0.73, "learning_rate": 0.0005495691347662473, "loss": 0.0423, "step": 65800 }, { "epoch": 0.73, "learning_rate": 0.0005494452808450252, "loss": 0.0412, "step": 66000 }, { "epoch": 0.74, "learning_rate": 0.000549321691479022, "loss": 0.0338, "step": 66200 }, { "epoch": 0.74, "learning_rate": 0.0005491977484533271, "loss": 0.0386, "step": 66400 }, { "epoch": 0.74, "learning_rate": 0.0005490728264358403, "loss": 0.0319, "step": 66600 }, { "epoch": 0.74, "learning_rate": 0.0005489475475317964, "loss": 0.0425, "step": 66800 }, { "epoch": 0.74, "learning_rate": 0.0005488219119105563, "loss": 0.0395, "step": 67000 }, { "epoch": 0.75, "learning_rate": 0.0005486959197419632, "loss": 0.0537, "step": 67200 }, { "epoch": 0.75, "learning_rate": 0.0005485695711963426, "loss": 0.0456, "step": 67400 }, { "epoch": 0.75, "learning_rate": 0.000548442866444501, "loss": 0.0408, "step": 67600 }, { "epoch": 0.75, "learning_rate": 0.0005483158056577273, "loss": 0.0371, "step": 67800 }, { "epoch": 0.76, "learning_rate": 0.0005481883890077914, "loss": 0.0415, "step": 68000 }, { "epoch": 0.76, "learning_rate": 0.0005480606166669441, "loss": 0.0458, "step": 68200 }, { "epoch": 0.76, "learning_rate": 0.0005479324888079173, "loss": 0.0385, "step": 68400 }, { "epoch": 0.76, "learning_rate": 0.0005478046489035769, "loss": 0.0437, "step": 68600 }, { "epoch": 0.76, "learning_rate": 0.0005476758123037319, "loss": 0.0331, "step": 68800 }, { "epoch": 0.77, "learning_rate": 0.0005475466207059137, "loss": 0.042, "step": 69000 }, { "epoch": 0.77, "learning_rate": 0.0005474170742847724, "loss": 0.0377, "step": 69200 }, { "epoch": 0.77, "learning_rate": 0.0005472871732154382, "loss": 0.0399, "step": 69400 }, { "epoch": 0.77, "learning_rate": 0.0005471569176735208, "loss": 0.0568, "step": 69600 }, { "epoch": 0.77, "learning_rate": 0.0005470263078351086, "loss": 0.031, "step": 69800 }, { "epoch": 0.78, "learning_rate": 0.0005468953438767697, "loss": 0.0308, "step": 70000 }, { "epoch": 0.78, "learning_rate": 0.0005467640259755503, "loss": 0.0417, "step": 70200 }, { "epoch": 0.78, "learning_rate": 0.0005466323543089754, "loss": 0.0467, "step": 70400 }, { "epoch": 0.78, "learning_rate": 0.0005465003290550484, "loss": 0.0316, "step": 70600 }, { "epoch": 0.79, "learning_rate": 0.0005463679503922502, "loss": 0.0411, "step": 70800 }, { "epoch": 0.79, "learning_rate": 0.0005462352184995399, "loss": 0.0365, "step": 71000 }, { "epoch": 0.79, "learning_rate": 0.0005461021335563539, "loss": 0.0358, "step": 71200 }, { "epoch": 0.79, "learning_rate": 0.0005459686957426061, "loss": 0.0497, "step": 71400 }, { "epoch": 0.79, "learning_rate": 0.0005458349052386872, "loss": 0.0415, "step": 71600 }, { "epoch": 0.8, "learning_rate": 0.0005457007622254647, "loss": 0.0367, "step": 71800 }, { "epoch": 0.8, "learning_rate": 0.0005455662668842827, "loss": 0.0343, "step": 72000 }, { "epoch": 0.8, "learning_rate": 0.0005454314193969619, "loss": 0.0434, "step": 72200 }, { "epoch": 0.8, "learning_rate": 0.0005452962199457985, "loss": 0.0465, "step": 72400 }, { "epoch": 0.81, "learning_rate": 0.0005451606687135648, "loss": 0.036, "step": 72600 }, { "epoch": 0.81, "learning_rate": 0.0005450247658835089, "loss": 0.0375, "step": 72800 }, { "epoch": 0.81, "learning_rate": 0.0005448885116393537, "loss": 0.0358, "step": 73000 }, { "epoch": 0.81, "learning_rate": 0.0005447519061652976, "loss": 0.0473, "step": 73200 }, { "epoch": 0.81, "learning_rate": 0.0005446149496460137, "loss": 0.0347, "step": 73400 }, { "epoch": 0.82, "learning_rate": 0.000544478329676004, "loss": 0.0562, "step": 73600 }, { "epoch": 0.82, "learning_rate": 0.0005443406733750912, "loss": 0.0354, "step": 73800 }, { "epoch": 0.82, "learning_rate": 0.0005442026665848845, "loss": 0.052, "step": 74000 }, { "epoch": 0.82, "learning_rate": 0.0005440643094919513, "loss": 0.0347, "step": 74200 }, { "epoch": 0.83, "learning_rate": 0.0005439256022833326, "loss": 0.0345, "step": 74400 }, { "epoch": 0.83, "learning_rate": 0.0005437865451465428, "loss": 0.0502, "step": 74600 }, { "epoch": 0.83, "learning_rate": 0.0005436471382695693, "loss": 0.0384, "step": 74800 }, { "epoch": 0.83, "learning_rate": 0.0005435073818408721, "loss": 0.0327, "step": 75000 }, { "epoch": 0.83, "learning_rate": 0.0005433672760493841, "loss": 0.0468, "step": 75200 }, { "epoch": 0.84, "learning_rate": 0.0005432268210845104, "loss": 0.0531, "step": 75400 }, { "epoch": 0.84, "learning_rate": 0.000543086017136128, "loss": 0.0342, "step": 75600 }, { "epoch": 0.84, "learning_rate": 0.0005429448643945857, "loss": 0.041, "step": 75800 }, { "epoch": 0.84, "learning_rate": 0.0005428040714242554, "loss": 0.0525, "step": 76000 }, { "epoch": 0.85, "learning_rate": 0.0005426622234109044, "loss": 0.0412, "step": 76200 }, { "epoch": 0.85, "learning_rate": 0.0005425200271773083, "loss": 0.0394, "step": 76400 }, { "epoch": 0.85, "learning_rate": 0.000542377482915698, "loss": 0.031, "step": 76600 }, { "epoch": 0.85, "learning_rate": 0.0005422345908187753, "loss": 0.0324, "step": 76800 }, { "epoch": 0.85, "learning_rate": 0.0005420913510797115, "loss": 0.0365, "step": 77000 }, { "epoch": 0.86, "learning_rate": 0.0005419477638921487, "loss": 0.0373, "step": 77200 }, { "epoch": 0.86, "learning_rate": 0.0005418038294501984, "loss": 0.0402, "step": 77400 }, { "epoch": 0.86, "learning_rate": 0.0005416595479484413, "loss": 0.0352, "step": 77600 }, { "epoch": 0.86, "learning_rate": 0.0005415149195819273, "loss": 0.0357, "step": 77800 }, { "epoch": 0.87, "learning_rate": 0.0005413699445461759, "loss": 0.0294, "step": 78000 }, { "epoch": 0.87, "learning_rate": 0.0005412246230371741, "loss": 0.0369, "step": 78200 }, { "epoch": 0.87, "learning_rate": 0.0005410789552513784, "loss": 0.0463, "step": 78400 }, { "epoch": 0.87, "learning_rate": 0.0005409329413857127, "loss": 0.0447, "step": 78600 }, { "epoch": 0.87, "learning_rate": 0.000540786581637569, "loss": 0.0476, "step": 78800 }, { "epoch": 0.88, "learning_rate": 0.0005406398762048069, "loss": 0.032, "step": 79000 }, { "epoch": 0.88, "learning_rate": 0.0005404928252857535, "loss": 0.0512, "step": 79200 }, { "epoch": 0.88, "learning_rate": 0.0005403454290792025, "loss": 0.0407, "step": 79400 }, { "epoch": 0.88, "learning_rate": 0.0005401984273489655, "loss": 0.0435, "step": 79600 }, { "epoch": 0.89, "learning_rate": 0.0005400503428896132, "loss": 0.037, "step": 79800 }, { "epoch": 0.89, "learning_rate": 0.0005399019137409428, "loss": 0.0396, "step": 80000 }, { "epoch": 0.89, "learning_rate": 0.0005397531401036115, "loss": 0.0398, "step": 80200 }, { "epoch": 0.89, "learning_rate": 0.0005396040221787419, "loss": 0.0346, "step": 80400 }, { "epoch": 0.89, "learning_rate": 0.0005394545601679225, "loss": 0.033, "step": 80600 }, { "epoch": 0.9, "learning_rate": 0.0005393047542732067, "loss": 0.0449, "step": 80800 }, { "epoch": 0.9, "learning_rate": 0.0005391546046971128, "loss": 0.0292, "step": 81000 }, { "epoch": 0.9, "learning_rate": 0.0005390041116426239, "loss": 0.0419, "step": 81200 }, { "epoch": 0.9, "learning_rate": 0.0005388532753131871, "loss": 0.0417, "step": 81400 }, { "epoch": 0.91, "learning_rate": 0.0005387020959127137, "loss": 0.0393, "step": 81600 }, { "epoch": 0.91, "learning_rate": 0.0005385505736455791, "loss": 0.0408, "step": 81800 }, { "epoch": 0.91, "learning_rate": 0.0005383994688932983, "loss": 0.0354, "step": 82000 }, { "epoch": 0.91, "learning_rate": 0.0005382472632195911, "loss": 0.0335, "step": 82200 }, { "epoch": 0.91, "learning_rate": 0.0005380947152940981, "loss": 0.0378, "step": 82400 }, { "epoch": 0.92, "learning_rate": 0.0005379418253230443, "loss": 0.0355, "step": 82600 }, { "epoch": 0.92, "learning_rate": 0.0005377885935131172, "loss": 0.0325, "step": 82800 }, { "epoch": 0.92, "learning_rate": 0.0005376350200714668, "loss": 0.0506, "step": 83000 }, { "epoch": 0.92, "learning_rate": 0.0005374811052057044, "loss": 0.0378, "step": 83200 }, { "epoch": 0.93, "learning_rate": 0.0005373268491239034, "loss": 0.0337, "step": 83400 }, { "epoch": 0.93, "learning_rate": 0.000537172252034598, "loss": 0.0499, "step": 83600 }, { "epoch": 0.93, "learning_rate": 0.0005370173141467838, "loss": 0.0344, "step": 83800 }, { "epoch": 0.93, "learning_rate": 0.0005368620356699166, "loss": 0.0406, "step": 84000 }, { "epoch": 0.93, "learning_rate": 0.0005367071957545386, "loss": 0.0563, "step": 84200 }, { "epoch": 0.94, "learning_rate": 0.0005365512384300956, "loss": 0.0382, "step": 84400 }, { "epoch": 0.94, "learning_rate": 0.0005363949411466742, "loss": 0.0396, "step": 84600 }, { "epoch": 0.94, "learning_rate": 0.0005362383041155683, "loss": 0.0408, "step": 84800 }, { "epoch": 0.94, "learning_rate": 0.0005360813275485313, "loss": 0.0319, "step": 85000 }, { "epoch": 0.95, "learning_rate": 0.0005359240116577751, "loss": 0.0491, "step": 85200 }, { "epoch": 0.95, "learning_rate": 0.0005357663566559707, "loss": 0.0347, "step": 85400 }, { "epoch": 0.95, "learning_rate": 0.0005356083627562474, "loss": 0.0417, "step": 85600 }, { "epoch": 0.95, "learning_rate": 0.0005354500301721926, "loss": 0.0595, "step": 85800 }, { "epoch": 0.95, "learning_rate": 0.0005352913591178517, "loss": 0.033, "step": 86000 }, { "epoch": 0.96, "learning_rate": 0.0005351331456953341, "loss": 0.0459, "step": 86200 }, { "epoch": 0.96, "learning_rate": 0.0005349738000340555, "loss": 0.0398, "step": 86400 }, { "epoch": 0.96, "learning_rate": 0.0005348141165462928, "loss": 0.0376, "step": 86600 }, { "epoch": 0.96, "learning_rate": 0.0005346540954479174, "loss": 0.0344, "step": 86800 }, { "epoch": 0.97, "learning_rate": 0.0005344937369552576, "loss": 0.0353, "step": 87000 }, { "epoch": 0.97, "learning_rate": 0.0005343330412850971, "loss": 0.0447, "step": 87200 }, { "epoch": 0.97, "learning_rate": 0.0005341720086546761, "loss": 0.0326, "step": 87400 }, { "epoch": 0.97, "learning_rate": 0.0005340106392816899, "loss": 0.0356, "step": 87600 }, { "epoch": 0.97, "learning_rate": 0.0005338489333842892, "loss": 0.0323, "step": 87800 }, { "epoch": 0.98, "learning_rate": 0.0005336868911810796, "loss": 0.0381, "step": 88000 }, { "epoch": 0.98, "learning_rate": 0.0005335245128911213, "loss": 0.0434, "step": 88200 }, { "epoch": 0.98, "learning_rate": 0.000533361798733929, "loss": 0.0331, "step": 88400 }, { "epoch": 0.98, "learning_rate": 0.0005331987489294713, "loss": 0.0513, "step": 88600 }, { "epoch": 0.99, "learning_rate": 0.0005330353636981706, "loss": 0.0333, "step": 88800 }, { "epoch": 0.99, "learning_rate": 0.0005328716432609027, "loss": 0.0448, "step": 89000 }, { "epoch": 0.99, "learning_rate": 0.0005327084089490138, "loss": 0.0479, "step": 89200 }, { "epoch": 0.99, "learning_rate": 0.0005325440204375134, "loss": 0.0394, "step": 89400 }, { "epoch": 0.99, "learning_rate": 0.0005323792973842787, "loss": 0.042, "step": 89600 }, { "epoch": 1.0, "learning_rate": 0.0005322142400119943, "loss": 0.0421, "step": 89800 }, { "epoch": 1.0, "learning_rate": 0.0005320488485437965, "loss": 0.0374, "step": 90000 }, { "epoch": 1.0, "eval_accuracy": 0.9394310337232581, "eval_auc": 0.8882998322705576, "eval_f1": 0.5088262320154422, "eval_loss": 0.30141326785087585, "eval_mcc": 0.5283107198885248, "eval_precision": 0.3662816185295454, "eval_recall": 0.8330034442739891, "eval_runtime": 4672.3874, "eval_samples_per_second": 24.286, "eval_steps_per_second": 4.857, "step": 90066 }, { "epoch": 1.0, "learning_rate": 0.0005318831232032731, "loss": 0.0455, "step": 90200 }, { "epoch": 1.0, "learning_rate": 0.0005317170642144641, "loss": 0.0368, "step": 90400 }, { "epoch": 1.01, "learning_rate": 0.0005315506718018594, "loss": 0.0327, "step": 90600 }, { "epoch": 1.01, "learning_rate": 0.0005313839461904002, "loss": 0.0466, "step": 90800 }, { "epoch": 1.01, "learning_rate": 0.0005312168876054782, "loss": 0.0552, "step": 91000 }, { "epoch": 1.01, "learning_rate": 0.0005310503340569336, "loss": 0.0451, "step": 91200 }, { "epoch": 1.01, "learning_rate": 0.0005308826118651043, "loss": 0.0366, "step": 91400 }, { "epoch": 1.02, "learning_rate": 0.0005307145573775516, "loss": 0.0465, "step": 91600 }, { "epoch": 1.02, "learning_rate": 0.000530546170821464, "loss": 0.0544, "step": 91800 }, { "epoch": 1.02, "learning_rate": 0.0005303774524244784, "loss": 0.0426, "step": 92000 }, { "epoch": 1.02, "learning_rate": 0.0005302084024146802, "loss": 0.0382, "step": 92200 }, { "epoch": 1.03, "learning_rate": 0.0005300390210206035, "loss": 0.0355, "step": 92400 }, { "epoch": 1.03, "learning_rate": 0.0005298693084712301, "loss": 0.034, "step": 92600 }, { "epoch": 1.03, "learning_rate": 0.0005296992649959896, "loss": 0.0306, "step": 92800 }, { "epoch": 1.03, "learning_rate": 0.000529528890824759, "loss": 0.0372, "step": 93000 }, { "epoch": 1.03, "learning_rate": 0.0005293590405326986, "loss": 0.0482, "step": 93200 }, { "epoch": 1.04, "learning_rate": 0.000529188007311506, "loss": 0.0436, "step": 93400 }, { "epoch": 1.04, "learning_rate": 0.0005290166440854781, "loss": 0.04, "step": 93600 }, { "epoch": 1.04, "learning_rate": 0.0005288449510862758, "loss": 0.0351, "step": 93800 }, { "epoch": 1.04, "learning_rate": 0.000528672928546006, "loss": 0.045, "step": 94000 }, { "epoch": 1.05, "learning_rate": 0.0005285005766972211, "loss": 0.0339, "step": 94200 }, { "epoch": 1.05, "learning_rate": 0.0005283278957729188, "loss": 0.056, "step": 94400 }, { "epoch": 1.05, "learning_rate": 0.0005281548860065413, "loss": 0.0264, "step": 94600 }, { "epoch": 1.05, "learning_rate": 0.0005279815476319757, "loss": 0.0509, "step": 94800 }, { "epoch": 1.05, "learning_rate": 0.0005278078808835534, "loss": 0.0352, "step": 95000 }, { "epoch": 1.06, "learning_rate": 0.0005276338859960492, "loss": 0.0338, "step": 95200 }, { "epoch": 1.06, "learning_rate": 0.000527459563204682, "loss": 0.0374, "step": 95400 }, { "epoch": 1.06, "learning_rate": 0.0005272849127451139, "loss": 0.0332, "step": 95600 }, { "epoch": 1.06, "learning_rate": 0.0005271099348534497, "loss": 0.0344, "step": 95800 }, { "epoch": 1.07, "learning_rate": 0.0005269346297662373, "loss": 0.0352, "step": 96000 }, { "epoch": 1.07, "learning_rate": 0.0005267589977204665, "loss": 0.0331, "step": 96200 }, { "epoch": 1.07, "learning_rate": 0.0005265830389535692, "loss": 0.0327, "step": 96400 }, { "epoch": 1.07, "learning_rate": 0.0005264067537034191, "loss": 0.0404, "step": 96600 }, { "epoch": 1.07, "learning_rate": 0.0005262310260769464, "loss": 0.0446, "step": 96800 }, { "epoch": 1.08, "learning_rate": 0.0005260540902051132, "loss": 0.0328, "step": 97000 }, { "epoch": 1.08, "learning_rate": 0.0005258768285650981, "loss": 0.0359, "step": 97200 }, { "epoch": 1.08, "learning_rate": 0.000525699241396536, "loss": 0.0362, "step": 97400 }, { "epoch": 1.08, "learning_rate": 0.000525521328939502, "loss": 0.0318, "step": 97600 }, { "epoch": 1.09, "learning_rate": 0.0005253430914345107, "loss": 0.0404, "step": 97800 }, { "epoch": 1.09, "learning_rate": 0.0005251645291225166, "loss": 0.0553, "step": 98000 }, { "epoch": 1.09, "learning_rate": 0.0005249856422449128, "loss": 0.0351, "step": 98200 }, { "epoch": 1.09, "learning_rate": 0.0005248064310435313, "loss": 0.0406, "step": 98400 }, { "epoch": 1.09, "learning_rate": 0.000524626895760643, "loss": 0.043, "step": 98600 }, { "epoch": 1.1, "learning_rate": 0.0005244470366389562, "loss": 0.0368, "step": 98800 }, { "epoch": 1.1, "learning_rate": 0.0005242677556397457, "loss": 0.0265, "step": 99000 }, { "epoch": 1.1, "learning_rate": 0.0005240872511864926, "loss": 0.0333, "step": 99200 }, { "epoch": 1.1, "learning_rate": 0.0005239064236239713, "loss": 0.0327, "step": 99400 }, { "epoch": 1.11, "learning_rate": 0.0005237252731966375, "loss": 0.0369, "step": 99600 }, { "epoch": 1.11, "learning_rate": 0.0005235438001493833, "loss": 0.0418, "step": 99800 }, { "epoch": 1.11, "learning_rate": 0.0005233620047275368, "loss": 0.0365, "step": 100000 }, { "epoch": 1.11, "learning_rate": 0.0005231798871768622, "loss": 0.0378, "step": 100200 }, { "epoch": 1.11, "learning_rate": 0.0005229974477435589, "loss": 0.0313, "step": 100400 }, { "epoch": 1.12, "learning_rate": 0.0005228146866742617, "loss": 0.0358, "step": 100600 }, { "epoch": 1.12, "learning_rate": 0.0005226316042160402, "loss": 0.0349, "step": 100800 }, { "epoch": 1.12, "learning_rate": 0.0005224491184328257, "loss": 0.0602, "step": 101000 }, { "epoch": 1.12, "learning_rate": 0.0005222653955435513, "loss": 0.0458, "step": 101200 }, { "epoch": 1.13, "learning_rate": 0.0005220813520079234, "loss": 0.0352, "step": 101400 }, { "epoch": 1.13, "learning_rate": 0.0005218969880747456, "loss": 0.0398, "step": 101600 }, { "epoch": 1.13, "learning_rate": 0.0005217123039932539, "loss": 0.0407, "step": 101800 }, { "epoch": 1.13, "learning_rate": 0.0005215273000131176, "loss": 0.0462, "step": 102000 }, { "epoch": 1.13, "learning_rate": 0.0005213419763844384, "loss": 0.0385, "step": 102200 }, { "epoch": 1.14, "learning_rate": 0.0005211563333577499, "loss": 0.0444, "step": 102400 }, { "epoch": 1.14, "learning_rate": 0.0005209703711840176, "loss": 0.0368, "step": 102600 }, { "epoch": 1.14, "learning_rate": 0.0005207840901146387, "loss": 0.0314, "step": 102800 }, { "epoch": 1.14, "learning_rate": 0.000520597490401441, "loss": 0.0557, "step": 103000 }, { "epoch": 1.15, "learning_rate": 0.0005204105722966836, "loss": 0.0315, "step": 103200 }, { "epoch": 1.15, "learning_rate": 0.0005202233360530556, "loss": 0.0384, "step": 103400 }, { "epoch": 1.15, "learning_rate": 0.0005200357819236765, "loss": 0.0431, "step": 103600 }, { "epoch": 1.15, "learning_rate": 0.0005198479101620953, "loss": 0.0299, "step": 103800 }, { "epoch": 1.15, "learning_rate": 0.0005196597210222905, "loss": 0.0436, "step": 104000 }, { "epoch": 1.16, "learning_rate": 0.0005194712147586696, "loss": 0.0432, "step": 104200 }, { "epoch": 1.16, "learning_rate": 0.000519282391626069, "loss": 0.0332, "step": 104400 }, { "epoch": 1.16, "learning_rate": 0.0005190932518797532, "loss": 0.037, "step": 104600 }, { "epoch": 1.16, "learning_rate": 0.0005189037957754149, "loss": 0.0394, "step": 104800 }, { "epoch": 1.17, "learning_rate": 0.0005187140235691744, "loss": 0.0358, "step": 105000 }, { "epoch": 1.17, "learning_rate": 0.0005185239355175794, "loss": 0.0349, "step": 105200 }, { "epoch": 1.17, "learning_rate": 0.0005183344846804048, "loss": 0.0472, "step": 105400 }, { "epoch": 1.17, "learning_rate": 0.0005181447216564026, "loss": 0.0398, "step": 105600 }, { "epoch": 1.17, "learning_rate": 0.0005179536907617519, "loss": 0.0447, "step": 105800 }, { "epoch": 1.18, "learning_rate": 0.0005177623450496188, "loss": 0.0507, "step": 106000 }, { "epoch": 1.18, "learning_rate": 0.0005175706847786785, "loss": 0.0401, "step": 106200 }, { "epoch": 1.18, "learning_rate": 0.0005173787102080307, "loss": 0.0342, "step": 106400 }, { "epoch": 1.18, "learning_rate": 0.0005171864215972007, "loss": 0.0344, "step": 106600 }, { "epoch": 1.19, "learning_rate": 0.0005169938192061378, "loss": 0.034, "step": 106800 }, { "epoch": 1.19, "learning_rate": 0.0005168009032952157, "loss": 0.0349, "step": 107000 }, { "epoch": 1.19, "learning_rate": 0.0005166076741252321, "loss": 0.0298, "step": 107200 }, { "epoch": 1.19, "learning_rate": 0.0005164141319574078, "loss": 0.0421, "step": 107400 }, { "epoch": 1.19, "learning_rate": 0.000516220277053387, "loss": 0.0383, "step": 107600 }, { "epoch": 1.2, "learning_rate": 0.0005160270812889729, "loss": 0.0418, "step": 107800 }, { "epoch": 1.2, "learning_rate": 0.0005158326032595872, "loss": 0.0538, "step": 108000 }, { "epoch": 1.2, "learning_rate": 0.0005156378132801571, "loss": 0.0346, "step": 108200 }, { "epoch": 1.2, "learning_rate": 0.0005154427116140137, "loss": 0.0342, "step": 108400 }, { "epoch": 1.21, "learning_rate": 0.0005152472985249094, "loss": 0.0472, "step": 108600 }, { "epoch": 1.21, "learning_rate": 0.0005150515742770175, "loss": 0.0333, "step": 108800 }, { "epoch": 1.21, "learning_rate": 0.0005148555391349324, "loss": 0.045, "step": 109000 }, { "epoch": 1.21, "learning_rate": 0.000514659193363668, "loss": 0.0396, "step": 109200 }, { "epoch": 1.21, "learning_rate": 0.0005144625372286588, "loss": 0.0336, "step": 109400 }, { "epoch": 1.22, "learning_rate": 0.0005142655709957586, "loss": 0.0383, "step": 109600 }, { "epoch": 1.22, "learning_rate": 0.0005140682949312405, "loss": 0.0341, "step": 109800 }, { "epoch": 1.22, "learning_rate": 0.0005138707093017963, "loss": 0.0358, "step": 110000 }, { "epoch": 1.22, "learning_rate": 0.0005136738046181087, "loss": 0.0476, "step": 110200 }, { "epoch": 1.23, "learning_rate": 0.000513475602205047, "loss": 0.0324, "step": 110400 }, { "epoch": 1.23, "learning_rate": 0.0005132770910283037, "loss": 0.0349, "step": 110600 }, { "epoch": 1.23, "learning_rate": 0.0005130782713562405, "loss": 0.0472, "step": 110800 }, { "epoch": 1.23, "learning_rate": 0.0005128791434576359, "loss": 0.0334, "step": 111000 }, { "epoch": 1.23, "learning_rate": 0.0005126797076016853, "loss": 0.0515, "step": 111200 }, { "epoch": 1.24, "learning_rate": 0.0005124799640580006, "loss": 0.0407, "step": 111400 }, { "epoch": 1.24, "learning_rate": 0.000512279913096609, "loss": 0.0324, "step": 111600 }, { "epoch": 1.24, "learning_rate": 0.000512079554987954, "loss": 0.0435, "step": 111800 }, { "epoch": 1.24, "learning_rate": 0.0005118788900028939, "loss": 0.0368, "step": 112000 }, { "epoch": 1.25, "learning_rate": 0.000511677918412702, "loss": 0.0402, "step": 112200 }, { "epoch": 1.25, "learning_rate": 0.0005114766404890663, "loss": 0.0397, "step": 112400 }, { "epoch": 1.25, "learning_rate": 0.0005112750565040884, "loss": 0.0415, "step": 112600 }, { "epoch": 1.25, "learning_rate": 0.0005110731667302841, "loss": 0.039, "step": 112800 }, { "epoch": 1.25, "learning_rate": 0.0005108709714405825, "loss": 0.0431, "step": 113000 }, { "epoch": 1.26, "learning_rate": 0.0005106684709083255, "loss": 0.0433, "step": 113200 }, { "epoch": 1.26, "learning_rate": 0.0005104656654072679, "loss": 0.0341, "step": 113400 }, { "epoch": 1.26, "learning_rate": 0.0005102625552115767, "loss": 0.0372, "step": 113600 }, { "epoch": 1.26, "learning_rate": 0.0005100591405958309, "loss": 0.0591, "step": 113800 }, { "epoch": 1.27, "learning_rate": 0.0005098554218350208, "loss": 0.0379, "step": 114000 }, { "epoch": 1.27, "learning_rate": 0.0005096513992045481, "loss": 0.0447, "step": 114200 }, { "epoch": 1.27, "learning_rate": 0.0005094470729802251, "loss": 0.0288, "step": 114400 }, { "epoch": 1.27, "learning_rate": 0.0005092434673400295, "loss": 0.0462, "step": 114600 }, { "epoch": 1.27, "learning_rate": 0.0005090385362716005, "loss": 0.0417, "step": 114800 }, { "epoch": 1.28, "learning_rate": 0.0005088333024378334, "loss": 0.0467, "step": 115000 }, { "epoch": 1.28, "learning_rate": 0.0005086277661161777, "loss": 0.0366, "step": 115200 }, { "epoch": 1.28, "learning_rate": 0.0005084219275844923, "loss": 0.0416, "step": 115400 }, { "epoch": 1.28, "learning_rate": 0.0005082157871210442, "loss": 0.0407, "step": 115600 }, { "epoch": 1.29, "learning_rate": 0.0005080093450045091, "loss": 0.0359, "step": 115800 }, { "epoch": 1.29, "learning_rate": 0.0005078026015139702, "loss": 0.0327, "step": 116000 }, { "epoch": 1.29, "learning_rate": 0.0005075955569289179, "loss": 0.0341, "step": 116200 }, { "epoch": 1.29, "learning_rate": 0.0005073882115292502, "loss": 0.0414, "step": 116400 }, { "epoch": 1.29, "learning_rate": 0.0005071805655952713, "loss": 0.0314, "step": 116600 }, { "epoch": 1.3, "learning_rate": 0.0005069726194076919, "loss": 0.0343, "step": 116800 }, { "epoch": 1.3, "learning_rate": 0.0005067654152241447, "loss": 0.0333, "step": 117000 }, { "epoch": 1.3, "learning_rate": 0.0005065579143376581, "loss": 0.0376, "step": 117200 }, { "epoch": 1.3, "learning_rate": 0.0005063490720702915, "loss": 0.0607, "step": 117400 }, { "epoch": 1.31, "learning_rate": 0.0005061399306733968, "loss": 0.0458, "step": 117600 }, { "epoch": 1.31, "learning_rate": 0.0005059304904297063, "loss": 0.0365, "step": 117800 }, { "epoch": 1.31, "learning_rate": 0.0005057207516223563, "loss": 0.0338, "step": 118000 }, { "epoch": 1.31, "learning_rate": 0.0005055107145348869, "loss": 0.0481, "step": 118200 }, { "epoch": 1.31, "learning_rate": 0.000505300379451241, "loss": 0.044, "step": 118400 }, { "epoch": 1.32, "learning_rate": 0.0005050897466557649, "loss": 0.047, "step": 118600 }, { "epoch": 1.32, "learning_rate": 0.0005048788164332068, "loss": 0.0333, "step": 118800 }, { "epoch": 1.32, "learning_rate": 0.0005046675890687175, "loss": 0.0329, "step": 119000 }, { "epoch": 1.32, "learning_rate": 0.0005044560648478492, "loss": 0.0451, "step": 119200 }, { "epoch": 1.33, "learning_rate": 0.0005042442440565555, "loss": 0.0471, "step": 119400 }, { "epoch": 1.33, "learning_rate": 0.0005040321269811908, "loss": 0.04, "step": 119600 }, { "epoch": 1.33, "learning_rate": 0.0005038197139085103, "loss": 0.0398, "step": 119800 }, { "epoch": 1.33, "learning_rate": 0.000503607005125669, "loss": 0.0383, "step": 120000 }, { "epoch": 1.33, "learning_rate": 0.0005033940009202221, "loss": 0.0425, "step": 120200 }, { "epoch": 1.34, "learning_rate": 0.0005031807015801236, "loss": 0.0351, "step": 120400 }, { "epoch": 1.34, "learning_rate": 0.0005029671073937273, "loss": 0.0453, "step": 120600 }, { "epoch": 1.34, "learning_rate": 0.0005027542888257382, "loss": 0.039, "step": 120800 }, { "epoch": 1.34, "learning_rate": 0.0005025401072840216, "loss": 0.0397, "step": 121000 }, { "epoch": 1.35, "learning_rate": 0.0005023256317620081, "loss": 0.0395, "step": 121200 }, { "epoch": 1.35, "learning_rate": 0.0005021108625496411, "loss": 0.0362, "step": 121400 }, { "epoch": 1.35, "learning_rate": 0.0005018957999372611, "loss": 0.0383, "step": 121600 }, { "epoch": 1.35, "learning_rate": 0.000501680444215605, "loss": 0.046, "step": 121800 }, { "epoch": 1.35, "learning_rate": 0.0005014647956758061, "loss": 0.0308, "step": 122000 }, { "epoch": 1.36, "learning_rate": 0.0005012488546093935, "loss": 0.0411, "step": 122200 }, { "epoch": 1.36, "learning_rate": 0.0005010326213082918, "loss": 0.0299, "step": 122400 }, { "epoch": 1.36, "learning_rate": 0.0005008160960648208, "loss": 0.033, "step": 122600 }, { "epoch": 1.36, "learning_rate": 0.0005005992791716946, "loss": 0.0394, "step": 122800 }, { "epoch": 1.37, "learning_rate": 0.0005003821709220219, "loss": 0.0388, "step": 123000 }, { "epoch": 1.37, "learning_rate": 0.0005001647716093052, "loss": 0.0417, "step": 123200 }, { "epoch": 1.37, "learning_rate": 0.0004999481707006514, "loss": 0.0326, "step": 123400 }, { "epoch": 1.37, "learning_rate": 0.0004997301915955692, "loss": 0.0421, "step": 123600 }, { "epoch": 1.37, "learning_rate": 0.0004995119223088352, "loss": 0.0387, "step": 123800 }, { "epoch": 1.38, "learning_rate": 0.0004992933631355216, "loss": 0.0518, "step": 124000 }, { "epoch": 1.38, "learning_rate": 0.0004990745143710922, "loss": 0.0399, "step": 124200 }, { "epoch": 1.38, "learning_rate": 0.0004988553763114026, "loss": 0.0343, "step": 124400 }, { "epoch": 1.38, "learning_rate": 0.0004986359492526991, "loss": 0.0328, "step": 124600 }, { "epoch": 1.39, "learning_rate": 0.0004984162334916192, "loss": 0.0398, "step": 124800 }, { "epoch": 1.39, "learning_rate": 0.0004981962293251899, "loss": 0.0418, "step": 125000 }, { "epoch": 1.39, "learning_rate": 0.0004979759370508291, "loss": 0.0404, "step": 125200 }, { "epoch": 1.39, "learning_rate": 0.0004977553569663433, "loss": 0.029, "step": 125400 }, { "epoch": 1.39, "learning_rate": 0.0004975344893699287, "loss": 0.0383, "step": 125600 }, { "epoch": 1.4, "learning_rate": 0.0004973133345601698, "loss": 0.0396, "step": 125800 }, { "epoch": 1.4, "learning_rate": 0.0004970918928360394, "loss": 0.0441, "step": 126000 }, { "epoch": 1.4, "learning_rate": 0.0004968701644968985, "loss": 0.0282, "step": 126200 }, { "epoch": 1.4, "learning_rate": 0.0004966481498424952, "loss": 0.0351, "step": 126400 }, { "epoch": 1.41, "learning_rate": 0.0004964269613872782, "loss": 0.0425, "step": 126600 }, { "epoch": 1.41, "learning_rate": 0.0004962043764309675, "loss": 0.0369, "step": 126800 }, { "epoch": 1.41, "learning_rate": 0.0004959815060594544, "loss": 0.0443, "step": 127000 }, { "epoch": 1.41, "learning_rate": 0.000495758350574031, "loss": 0.0373, "step": 127200 }, { "epoch": 1.41, "learning_rate": 0.000495534910276375, "loss": 0.039, "step": 127400 }, { "epoch": 1.42, "learning_rate": 0.0004953111854685486, "loss": 0.0292, "step": 127600 }, { "epoch": 1.42, "learning_rate": 0.0004950871764529994, "loss": 0.0486, "step": 127800 }, { "epoch": 1.42, "learning_rate": 0.0004948628835325587, "loss": 0.0321, "step": 128000 }, { "epoch": 1.42, "learning_rate": 0.0004946383070104417, "loss": 0.0514, "step": 128200 }, { "epoch": 1.43, "learning_rate": 0.000494413447190247, "loss": 0.0341, "step": 128400 }, { "epoch": 1.43, "learning_rate": 0.0004941883043759562, "loss": 0.0448, "step": 128600 }, { "epoch": 1.43, "learning_rate": 0.0004939628788719335, "loss": 0.0435, "step": 128800 }, { "epoch": 1.43, "learning_rate": 0.0004937371709829253, "loss": 0.0346, "step": 129000 }, { "epoch": 1.43, "learning_rate": 0.0004935123116650722, "loss": 0.0399, "step": 129200 }, { "epoch": 1.44, "learning_rate": 0.0004932860413299694, "loss": 0.0329, "step": 129400 }, { "epoch": 1.44, "learning_rate": 0.0004930594895248787, "loss": 0.0453, "step": 129600 }, { "epoch": 1.44, "learning_rate": 0.0004928326565560686, "loss": 0.0359, "step": 129800 }, { "epoch": 1.44, "learning_rate": 0.0004926055427301884, "loss": 0.0287, "step": 130000 }, { "epoch": 1.45, "learning_rate": 0.0004923781483542667, "loss": 0.0274, "step": 130200 }, { "epoch": 1.45, "learning_rate": 0.0004921504737357116, "loss": 0.0441, "step": 130400 }, { "epoch": 1.45, "learning_rate": 0.0004919225191823099, "loss": 0.0473, "step": 130600 }, { "epoch": 1.45, "learning_rate": 0.0004916942850022269, "loss": 0.0404, "step": 130800 }, { "epoch": 1.45, "learning_rate": 0.0004914657715040059, "loss": 0.0341, "step": 131000 }, { "epoch": 1.46, "learning_rate": 0.0004912369789965678, "loss": 0.0394, "step": 131200 }, { "epoch": 1.46, "learning_rate": 0.0004910090538380014, "loss": 0.0481, "step": 131400 }, { "epoch": 1.46, "learning_rate": 0.0004907797056315803, "loss": 0.0433, "step": 131600 }, { "epoch": 1.46, "learning_rate": 0.0004905500793434152, "loss": 0.0384, "step": 131800 }, { "epoch": 1.47, "learning_rate": 0.000490320175283931, "loss": 0.0304, "step": 132000 }, { "epoch": 1.47, "learning_rate": 0.0004900899937639288, "loss": 0.0372, "step": 132200 }, { "epoch": 1.47, "learning_rate": 0.0004898595350945842, "loss": 0.0318, "step": 132400 }, { "epoch": 1.47, "learning_rate": 0.0004896287995874479, "loss": 0.0374, "step": 132600 }, { "epoch": 1.47, "learning_rate": 0.0004893977875544445, "loss": 0.034, "step": 132800 }, { "epoch": 1.48, "learning_rate": 0.0004891664993078729, "loss": 0.0327, "step": 133000 }, { "epoch": 1.48, "learning_rate": 0.0004889349351604048, "loss": 0.0298, "step": 133200 }, { "epoch": 1.48, "learning_rate": 0.0004887030954250854, "loss": 0.0386, "step": 133400 }, { "epoch": 1.48, "learning_rate": 0.000488470980415332, "loss": 0.036, "step": 133600 }, { "epoch": 1.49, "learning_rate": 0.0004882385904449345, "loss": 0.0312, "step": 133800 }, { "epoch": 1.49, "learning_rate": 0.00048800708983380163, "loss": 0.0619, "step": 134000 }, { "epoch": 1.49, "learning_rate": 0.000487774152255848, "loss": 0.0279, "step": 134200 }, { "epoch": 1.49, "learning_rate": 0.00048754094065927244, "loss": 0.0359, "step": 134400 }, { "epoch": 1.49, "learning_rate": 0.00048730745535934694, "loss": 0.0404, "step": 134600 }, { "epoch": 1.5, "learning_rate": 0.0004870736966717138, "loss": 0.0365, "step": 134800 }, { "epoch": 1.5, "learning_rate": 0.0004868396649123845, "loss": 0.0382, "step": 135000 }, { "epoch": 1.5, "learning_rate": 0.00048660536039774033, "loss": 0.049, "step": 135200 }, { "epoch": 1.5, "learning_rate": 0.00048637078344453057, "loss": 0.0389, "step": 135400 }, { "epoch": 1.51, "learning_rate": 0.0004861359343698732, "loss": 0.0471, "step": 135600 }, { "epoch": 1.51, "learning_rate": 0.00048590081349125406, "loss": 0.04, "step": 135800 }, { "epoch": 1.51, "learning_rate": 0.0004856654211265263, "loss": 0.0389, "step": 136000 }, { "epoch": 1.51, "learning_rate": 0.00048542975759391024, "loss": 0.0334, "step": 136200 }, { "epoch": 1.51, "learning_rate": 0.0004851950035571125, "loss": 0.044, "step": 136400 }, { "epoch": 1.52, "learning_rate": 0.0004849587999967043, "loss": 0.0349, "step": 136600 }, { "epoch": 1.52, "learning_rate": 0.0004847223262236687, "loss": 0.0447, "step": 136800 }, { "epoch": 1.52, "learning_rate": 0.000484485582557688, "loss": 0.0376, "step": 137000 }, { "epoch": 1.52, "learning_rate": 0.0004842485693188092, "loss": 0.0356, "step": 137200 }, { "epoch": 1.53, "learning_rate": 0.00048401128682744386, "loss": 0.045, "step": 137400 }, { "epoch": 1.53, "learning_rate": 0.0004837737354043673, "loss": 0.048, "step": 137600 }, { "epoch": 1.53, "learning_rate": 0.00048353591537071857, "loss": 0.0413, "step": 137800 }, { "epoch": 1.53, "learning_rate": 0.00048329782704799985, "loss": 0.035, "step": 138000 }, { "epoch": 1.53, "learning_rate": 0.00048305947075807595, "loss": 0.039, "step": 138200 }, { "epoch": 1.54, "learning_rate": 0.00048282084682317394, "loss": 0.0356, "step": 138400 }, { "epoch": 1.54, "learning_rate": 0.00048258195556588265, "loss": 0.0464, "step": 138600 }, { "epoch": 1.54, "learning_rate": 0.0004823427973091527, "loss": 0.0324, "step": 138800 }, { "epoch": 1.54, "learning_rate": 0.000482103372376295, "loss": 0.0324, "step": 139000 }, { "epoch": 1.55, "learning_rate": 0.0004818648802094239, "loss": 0.0503, "step": 139200 }, { "epoch": 1.55, "learning_rate": 0.0004816249242250218, "loss": 0.0286, "step": 139400 }, { "epoch": 1.55, "learning_rate": 0.0004813847025349644, "loss": 0.0412, "step": 139600 }, { "epoch": 1.55, "learning_rate": 0.0004811442154640004, "loss": 0.0297, "step": 139800 }, { "epoch": 1.55, "learning_rate": 0.0004809034633372375, "loss": 0.0551, "step": 140000 }, { "epoch": 1.56, "learning_rate": 0.0004806624464801416, "loss": 0.0401, "step": 140200 }, { "epoch": 1.56, "learning_rate": 0.0004804211652185367, "loss": 0.0436, "step": 140400 }, { "epoch": 1.56, "learning_rate": 0.0004801796198786039, "loss": 0.0511, "step": 140600 }, { "epoch": 1.56, "learning_rate": 0.00047993781078688175, "loss": 0.0406, "step": 140800 }, { "epoch": 1.57, "learning_rate": 0.0004796957382702646, "loss": 0.0353, "step": 141000 }, { "epoch": 1.57, "learning_rate": 0.000479453402656004, "loss": 0.0455, "step": 141200 }, { "epoch": 1.57, "learning_rate": 0.000479210804271706, "loss": 0.036, "step": 141400 }, { "epoch": 1.57, "learning_rate": 0.0004789691584017465, "loss": 0.0384, "step": 141600 }, { "epoch": 1.57, "learning_rate": 0.00047872603677136624, "loss": 0.0341, "step": 141800 }, { "epoch": 1.58, "learning_rate": 0.0004784826533542542, "loss": 0.0314, "step": 142000 }, { "epoch": 1.58, "learning_rate": 0.00047823900847943334, "loss": 0.0412, "step": 142200 }, { "epoch": 1.58, "learning_rate": 0.0004779951024762804, "loss": 0.0362, "step": 142400 }, { "epoch": 1.58, "learning_rate": 0.00047775093567452477, "loss": 0.0471, "step": 142600 }, { "epoch": 1.59, "learning_rate": 0.00047750650840424887, "loss": 0.0353, "step": 142800 }, { "epoch": 1.59, "learning_rate": 0.00047726182099588676, "loss": 0.0334, "step": 143000 }, { "epoch": 1.59, "learning_rate": 0.00047701687378022453, "loss": 0.0454, "step": 143200 }, { "epoch": 1.59, "learning_rate": 0.00047677166708839924, "loss": 0.0343, "step": 143400 }, { "epoch": 1.59, "learning_rate": 0.00047652620125189905, "loss": 0.0413, "step": 143600 }, { "epoch": 1.6, "learning_rate": 0.000476280476602562, "loss": 0.0356, "step": 143800 }, { "epoch": 1.6, "learning_rate": 0.00047603572403064705, "loss": 0.0552, "step": 144000 }, { "epoch": 1.6, "learning_rate": 0.0004757894840424632, "loss": 0.0279, "step": 144200 }, { "epoch": 1.6, "learning_rate": 0.0004755429862373897, "loss": 0.043, "step": 144400 }, { "epoch": 1.61, "learning_rate": 0.0004752962309486598, "loss": 0.0356, "step": 144600 }, { "epoch": 1.61, "learning_rate": 0.0004750492185098552, "loss": 0.0378, "step": 144800 }, { "epoch": 1.61, "learning_rate": 0.0004748019492549047, "loss": 0.0426, "step": 145000 }, { "epoch": 1.61, "learning_rate": 0.00047455442351808483, "loss": 0.0414, "step": 145200 }, { "epoch": 1.61, "learning_rate": 0.00047430664163401835, "loss": 0.0336, "step": 145400 }, { "epoch": 1.62, "learning_rate": 0.00047405860393767474, "loss": 0.0404, "step": 145600 }, { "epoch": 1.62, "learning_rate": 0.00047381031076436906, "loss": 0.0404, "step": 145800 }, { "epoch": 1.62, "learning_rate": 0.0004735630058254431, "loss": 0.059, "step": 146000 }, { "epoch": 1.62, "learning_rate": 0.00047331420397872965, "loss": 0.0397, "step": 146200 }, { "epoch": 1.63, "learning_rate": 0.0004730651476613874, "loss": 0.0463, "step": 146400 }, { "epoch": 1.63, "learning_rate": 0.0004728158372101083, "loss": 0.0549, "step": 146600 }, { "epoch": 1.63, "learning_rate": 0.000472566272961928, "loss": 0.0313, "step": 146800 }, { "epoch": 1.63, "learning_rate": 0.00047231645525422556, "loss": 0.048, "step": 147000 }, { "epoch": 1.63, "learning_rate": 0.0004720663844247223, "loss": 0.044, "step": 147200 }, { "epoch": 1.64, "learning_rate": 0.0004718160608114819, "loss": 0.0396, "step": 147400 }, { "epoch": 1.64, "learning_rate": 0.00047156548475290976, "loss": 0.0358, "step": 147600 }, { "epoch": 1.64, "learning_rate": 0.00047131465658775245, "loss": 0.0375, "step": 147800 }, { "epoch": 1.64, "learning_rate": 0.0004710635766550974, "loss": 0.029, "step": 148000 }, { "epoch": 1.65, "learning_rate": 0.0004708122452943726, "loss": 0.0423, "step": 148200 }, { "epoch": 1.65, "learning_rate": 0.00047056066284534564, "loss": 0.0483, "step": 148400 }, { "epoch": 1.65, "learning_rate": 0.00047030882964812376, "loss": 0.0382, "step": 148600 }, { "epoch": 1.65, "learning_rate": 0.00047005674604315304, "loss": 0.042, "step": 148800 }, { "epoch": 1.65, "learning_rate": 0.00046980441237121827, "loss": 0.0379, "step": 149000 }, { "epoch": 1.66, "learning_rate": 0.000469551828973442, "loss": 0.0376, "step": 149200 }, { "epoch": 1.66, "learning_rate": 0.0004692989961912846, "loss": 0.0435, "step": 149400 }, { "epoch": 1.66, "learning_rate": 0.0004690459143665435, "loss": 0.0486, "step": 149600 }, { "epoch": 1.66, "learning_rate": 0.00046879258384135286, "loss": 0.0385, "step": 149800 }, { "epoch": 1.67, "learning_rate": 0.0004685390049581831, "loss": 0.0371, "step": 150000 }, { "epoch": 1.67, "learning_rate": 0.0004682864478107021, "loss": 0.0485, "step": 150200 }, { "epoch": 1.67, "learning_rate": 0.0004680336454600183, "loss": 0.0583, "step": 150400 }, { "epoch": 1.67, "learning_rate": 0.0004677793260326723, "loss": 0.0379, "step": 150600 }, { "epoch": 1.67, "learning_rate": 0.00046752475961714146, "loss": 0.0332, "step": 150800 }, { "epoch": 1.68, "learning_rate": 0.00046726994655756706, "loss": 0.041, "step": 151000 }, { "epoch": 1.68, "learning_rate": 0.0004670148871984234, "loss": 0.0397, "step": 151200 }, { "epoch": 1.68, "learning_rate": 0.0004667595818845182, "loss": 0.0315, "step": 151400 }, { "epoch": 1.68, "learning_rate": 0.0004665040309609913, "loss": 0.0431, "step": 151600 }, { "epoch": 1.69, "learning_rate": 0.00046624823477331483, "loss": 0.0392, "step": 151800 }, { "epoch": 1.69, "learning_rate": 0.00046599219366729225, "loss": 0.0432, "step": 152000 }, { "epoch": 1.69, "learning_rate": 0.00046573590798905845, "loss": 0.0406, "step": 152200 }, { "epoch": 1.69, "learning_rate": 0.0004654793780850786, "loss": 0.0382, "step": 152400 }, { "epoch": 1.69, "learning_rate": 0.0004652226043021482, "loss": 0.041, "step": 152600 }, { "epoch": 1.7, "learning_rate": 0.000464966872679177, "loss": 0.0469, "step": 152800 }, { "epoch": 1.7, "learning_rate": 0.0004647096133951073, "loss": 0.0318, "step": 153000 }, { "epoch": 1.7, "learning_rate": 0.00046445211127271014, "loss": 0.0353, "step": 153200 }, { "epoch": 1.7, "learning_rate": 0.0004641943666600953, "loss": 0.0459, "step": 153400 }, { "epoch": 1.71, "learning_rate": 0.00046393637990570024, "loss": 0.0397, "step": 153600 }, { "epoch": 1.71, "learning_rate": 0.0004636781513582901, "loss": 0.0484, "step": 153800 }, { "epoch": 1.71, "learning_rate": 0.0004634196813669566, "loss": 0.0404, "step": 154000 }, { "epoch": 1.71, "learning_rate": 0.00046316097028111804, "loss": 0.0386, "step": 154200 }, { "epoch": 1.71, "learning_rate": 0.0004629020184505186, "loss": 0.0376, "step": 154400 }, { "epoch": 1.72, "learning_rate": 0.00046264282622522786, "loss": 0.0358, "step": 154600 }, { "epoch": 1.72, "learning_rate": 0.00046238339395564043, "loss": 0.035, "step": 154800 }, { "epoch": 1.72, "learning_rate": 0.00046212372199247546, "loss": 0.0402, "step": 155000 }, { "epoch": 1.72, "learning_rate": 0.000461863810686776, "loss": 0.0308, "step": 155200 }, { "epoch": 1.73, "learning_rate": 0.00046160366038990876, "loss": 0.0393, "step": 155400 }, { "epoch": 1.73, "learning_rate": 0.0004613432714535636, "loss": 0.0313, "step": 155600 }, { "epoch": 1.73, "learning_rate": 0.00046108264422975296, "loss": 0.0397, "step": 155800 }, { "epoch": 1.73, "learning_rate": 0.00046082177907081127, "loss": 0.0405, "step": 156000 }, { "epoch": 1.73, "learning_rate": 0.0004605606763293948, "loss": 0.0354, "step": 156200 }, { "epoch": 1.74, "learning_rate": 0.0004602993363584809, "loss": 0.0326, "step": 156400 }, { "epoch": 1.74, "learning_rate": 0.0004600390679842479, "loss": 0.0397, "step": 156600 }, { "epoch": 1.74, "learning_rate": 0.0004597772557962863, "loss": 0.0365, "step": 156800 }, { "epoch": 1.74, "learning_rate": 0.0004595152074379112, "loss": 0.0279, "step": 157000 }, { "epoch": 1.75, "learning_rate": 0.0004592529232633784, "loss": 0.0319, "step": 157200 }, { "epoch": 1.75, "learning_rate": 0.00045899040362726236, "loss": 0.0389, "step": 157400 }, { "epoch": 1.75, "learning_rate": 0.00045872764888445585, "loss": 0.0362, "step": 157600 }, { "epoch": 1.75, "learning_rate": 0.00045846465939016964, "loss": 0.0396, "step": 157800 }, { "epoch": 1.75, "learning_rate": 0.0004582014354999317, "loss": 0.0402, "step": 158000 }, { "epoch": 1.76, "learning_rate": 0.00045793797756958697, "loss": 0.0423, "step": 158200 }, { "epoch": 1.76, "learning_rate": 0.0004576742859552967, "loss": 0.0363, "step": 158400 }, { "epoch": 1.76, "learning_rate": 0.00045741036101353813, "loss": 0.0415, "step": 158600 }, { "epoch": 1.76, "learning_rate": 0.00045714620310110377, "loss": 0.035, "step": 158800 }, { "epoch": 1.77, "learning_rate": 0.0004568818125751013, "loss": 0.038, "step": 159000 }, { "epoch": 1.77, "learning_rate": 0.00045661718979295277, "loss": 0.0356, "step": 159200 }, { "epoch": 1.77, "learning_rate": 0.00045635233511239404, "loss": 0.0316, "step": 159400 }, { "epoch": 1.77, "learning_rate": 0.00045608724889147485, "loss": 0.028, "step": 159600 }, { "epoch": 1.77, "learning_rate": 0.0004558232586500442, "loss": 0.0487, "step": 159800 }, { "epoch": 1.78, "learning_rate": 0.0004555577115770279, "loss": 0.0402, "step": 160000 }, { "epoch": 1.78, "learning_rate": 0.00045529193403787995, "loss": 0.0438, "step": 160200 }, { "epoch": 1.78, "learning_rate": 0.0004550259263918975, "loss": 0.0326, "step": 160400 }, { "epoch": 1.78, "learning_rate": 0.0004547596889986888, "loss": 0.0466, "step": 160600 }, { "epoch": 1.79, "learning_rate": 0.00045449322221817234, "loss": 0.0371, "step": 160800 }, { "epoch": 1.79, "learning_rate": 0.00045422652641057736, "loss": 0.0333, "step": 161000 }, { "epoch": 1.79, "learning_rate": 0.00045395960193644213, "loss": 0.0361, "step": 161200 }, { "epoch": 1.79, "learning_rate": 0.00045369244915661436, "loss": 0.0344, "step": 161400 }, { "epoch": 1.79, "learning_rate": 0.00045342506843225013, "loss": 0.0303, "step": 161600 }, { "epoch": 1.8, "learning_rate": 0.000453157460124814, "loss": 0.0409, "step": 161800 }, { "epoch": 1.8, "learning_rate": 0.0004528909643383358, "loss": 0.0521, "step": 162000 }, { "epoch": 1.8, "learning_rate": 0.00045262290308377407, "loss": 0.0362, "step": 162200 }, { "epoch": 1.8, "learning_rate": 0.0004523546153305649, "loss": 0.0438, "step": 162400 }, { "epoch": 1.81, "learning_rate": 0.00045208610144139883, "loss": 0.039, "step": 162600 }, { "epoch": 1.81, "learning_rate": 0.00045181736177927237, "loss": 0.0324, "step": 162800 }, { "epoch": 1.81, "learning_rate": 0.00045154839670748676, "loss": 0.0299, "step": 163000 }, { "epoch": 1.81, "learning_rate": 0.0004512792065896484, "loss": 0.0429, "step": 163200 }, { "epoch": 1.81, "learning_rate": 0.0004510097917896677, "loss": 0.0392, "step": 163400 }, { "epoch": 1.82, "learning_rate": 0.0004507401526717589, "loss": 0.0352, "step": 163600 }, { "epoch": 1.82, "learning_rate": 0.00045047028960043946, "loss": 0.0303, "step": 163800 }, { "epoch": 1.82, "learning_rate": 0.0004502002029405296, "loss": 0.0481, "step": 164000 }, { "epoch": 1.82, "learning_rate": 0.00044993124516123255, "loss": 0.0402, "step": 164200 }, { "epoch": 1.83, "learning_rate": 0.00044966071353319173, "loss": 0.0341, "step": 164400 }, { "epoch": 1.83, "learning_rate": 0.0004493899594110033, "loss": 0.0328, "step": 164600 }, { "epoch": 1.83, "learning_rate": 0.00044911898316069216, "loss": 0.0293, "step": 164800 }, { "epoch": 1.83, "learning_rate": 0.0004488491416896699, "loss": 0.039, "step": 165000 }, { "epoch": 1.83, "learning_rate": 0.0004485777233884517, "loss": 0.0447, "step": 165200 }, { "epoch": 1.84, "learning_rate": 0.00044830608405714947, "loss": 0.0305, "step": 165400 }, { "epoch": 1.84, "learning_rate": 0.00044803422406298473, "loss": 0.0432, "step": 165600 }, { "epoch": 1.84, "learning_rate": 0.0004477621437734773, "loss": 0.0504, "step": 165800 }, { "epoch": 1.84, "learning_rate": 0.0004474898435564448, "loss": 0.0424, "step": 166000 }, { "epoch": 1.85, "learning_rate": 0.0004472173237800022, "loss": 0.0464, "step": 166200 }, { "epoch": 1.85, "learning_rate": 0.0004469445848125611, "loss": 0.0236, "step": 166400 }, { "epoch": 1.85, "learning_rate": 0.00044667162702282965, "loss": 0.0314, "step": 166600 }, { "epoch": 1.85, "learning_rate": 0.00044639981720381866, "loss": 0.0404, "step": 166800 }, { "epoch": 1.85, "learning_rate": 0.0004461264239663143, "loss": 0.0525, "step": 167000 }, { "epoch": 1.86, "learning_rate": 0.00044585281301256797, "loss": 0.0358, "step": 167200 }, { "epoch": 1.86, "learning_rate": 0.00044557898471246654, "loss": 0.0365, "step": 167400 }, { "epoch": 1.86, "learning_rate": 0.0004453049394361907, "loss": 0.0372, "step": 167600 }, { "epoch": 1.86, "learning_rate": 0.00044503067755421446, "loss": 0.0436, "step": 167800 }, { "epoch": 1.87, "learning_rate": 0.00044475619943730476, "loss": 0.0383, "step": 168000 }, { "epoch": 1.87, "learning_rate": 0.00044448150545652065, "loss": 0.0435, "step": 168200 }, { "epoch": 1.87, "learning_rate": 0.0004442065959832132, "loss": 0.0453, "step": 168400 }, { "epoch": 1.87, "learning_rate": 0.00044393147138902455, "loss": 0.0376, "step": 168600 }, { "epoch": 1.87, "learning_rate": 0.00044365613204588796, "loss": 0.0468, "step": 168800 }, { "epoch": 1.88, "learning_rate": 0.0004433805783260268, "loss": 0.0366, "step": 169000 }, { "epoch": 1.88, "learning_rate": 0.0004431048106019544, "loss": 0.0503, "step": 169200 }, { "epoch": 1.88, "learning_rate": 0.0004428288292464733, "loss": 0.034, "step": 169400 }, { "epoch": 1.88, "learning_rate": 0.0004425526346326748, "loss": 0.034, "step": 169600 }, { "epoch": 1.89, "learning_rate": 0.00044227622713393847, "loss": 0.0309, "step": 169800 }, { "epoch": 1.89, "learning_rate": 0.00044199960712393185, "loss": 0.0418, "step": 170000 }, { "epoch": 1.89, "learning_rate": 0.00044172277497660966, "loss": 0.055, "step": 170200 }, { "epoch": 1.89, "learning_rate": 0.0004414457310662133, "loss": 0.0288, "step": 170400 }, { "epoch": 1.89, "learning_rate": 0.00044116847576727077, "loss": 0.0413, "step": 170600 }, { "epoch": 1.9, "learning_rate": 0.00044089100945459553, "loss": 0.0362, "step": 170800 }, { "epoch": 1.9, "learning_rate": 0.00044061333250328637, "loss": 0.0318, "step": 171000 }, { "epoch": 1.9, "learning_rate": 0.0004403354452887268, "loss": 0.0294, "step": 171200 }, { "epoch": 1.9, "learning_rate": 0.0004400573481865848, "loss": 0.0362, "step": 171400 }, { "epoch": 1.91, "learning_rate": 0.0004397790415728118, "loss": 0.0363, "step": 171600 }, { "epoch": 1.91, "learning_rate": 0.0004395019189219901, "loss": 0.0308, "step": 171800 }, { "epoch": 1.91, "learning_rate": 0.00043922319545679976, "loss": 0.0445, "step": 172000 }, { "epoch": 1.91, "learning_rate": 0.0004389442636076458, "loss": 0.049, "step": 172200 }, { "epoch": 1.91, "learning_rate": 0.00043866512375160833, "loss": 0.0368, "step": 172400 }, { "epoch": 1.92, "learning_rate": 0.0004383857762660486, "loss": 0.0412, "step": 172600 }, { "epoch": 1.92, "learning_rate": 0.00043810622152860846, "loss": 0.0357, "step": 172800 }, { "epoch": 1.92, "learning_rate": 0.00043782645991721005, "loss": 0.0456, "step": 173000 }, { "epoch": 1.92, "learning_rate": 0.0004375464918100551, "loss": 0.033, "step": 173200 }, { "epoch": 1.93, "learning_rate": 0.00043726631758562464, "loss": 0.0406, "step": 173400 }, { "epoch": 1.93, "learning_rate": 0.00043698593762267815, "loss": 0.0365, "step": 173600 }, { "epoch": 1.93, "learning_rate": 0.0004367053523002534, "loss": 0.0374, "step": 173800 }, { "epoch": 1.93, "learning_rate": 0.0004364245619976656, "loss": 0.0462, "step": 174000 }, { "epoch": 1.93, "learning_rate": 0.00043614356709450726, "loss": 0.0388, "step": 174200 }, { "epoch": 1.94, "learning_rate": 0.00043586236797064746, "loss": 0.039, "step": 174400 }, { "epoch": 1.94, "learning_rate": 0.00043558096500623115, "loss": 0.0306, "step": 174600 }, { "epoch": 1.94, "learning_rate": 0.00043529935858167915, "loss": 0.0333, "step": 174800 }, { "epoch": 1.94, "learning_rate": 0.00043501754907768707, "loss": 0.034, "step": 175000 }, { "epoch": 1.95, "learning_rate": 0.00043473553687522527, "loss": 0.0447, "step": 175200 }, { "epoch": 1.95, "learning_rate": 0.0004344533223555379, "loss": 0.0319, "step": 175400 }, { "epoch": 1.95, "learning_rate": 0.0004341709059001428, "loss": 0.0507, "step": 175600 }, { "epoch": 1.95, "learning_rate": 0.00043388828789083086, "loss": 0.0265, "step": 175800 }, { "epoch": 1.95, "learning_rate": 0.00043360546870966515, "loss": 0.0353, "step": 176000 }, { "epoch": 1.96, "learning_rate": 0.00043332244873898096, "loss": 0.0364, "step": 176200 }, { "epoch": 1.96, "learning_rate": 0.0004330392283613851, "loss": 0.0458, "step": 176400 }, { "epoch": 1.96, "learning_rate": 0.0004327572255586892, "loss": 0.0476, "step": 176600 }, { "epoch": 1.96, "learning_rate": 0.00043247360651342364, "loss": 0.0434, "step": 176800 }, { "epoch": 1.97, "learning_rate": 0.00043219120779531834, "loss": 0.0431, "step": 177000 }, { "epoch": 1.97, "learning_rate": 0.00043190719160839005, "loss": 0.0441, "step": 177200 }, { "epoch": 1.97, "learning_rate": 0.0004316229769277958, "loss": 0.0266, "step": 177400 }, { "epoch": 1.97, "learning_rate": 0.0004313385641377574, "loss": 0.0457, "step": 177600 }, { "epoch": 1.97, "learning_rate": 0.0004310539536227645, "loss": 0.0366, "step": 177800 }, { "epoch": 1.98, "learning_rate": 0.0004307691457675737, "loss": 0.036, "step": 178000 }, { "epoch": 1.98, "learning_rate": 0.00043048414095720887, "loss": 0.043, "step": 178200 }, { "epoch": 1.98, "learning_rate": 0.00043019893957695976, "loss": 0.0472, "step": 178400 }, { "epoch": 1.98, "learning_rate": 0.0004299135420123821, "loss": 0.0298, "step": 178600 }, { "epoch": 1.99, "learning_rate": 0.0004296279486492968, "loss": 0.0349, "step": 178800 }, { "epoch": 1.99, "learning_rate": 0.0004293421598737892, "loss": 0.0334, "step": 179000 }, { "epoch": 1.99, "learning_rate": 0.00042905617607220926, "loss": 0.0583, "step": 179200 }, { "epoch": 1.99, "learning_rate": 0.0004287699976311703, "loss": 0.0377, "step": 179400 }, { "epoch": 1.99, "learning_rate": 0.0004284836249375487, "loss": 0.0328, "step": 179600 }, { "epoch": 2.0, "learning_rate": 0.00042819705837848364, "loss": 0.025, "step": 179800 }, { "epoch": 2.0, "learning_rate": 0.00042791029834137633, "loss": 0.0515, "step": 180000 }, { "epoch": 2.0, "eval_accuracy": 0.9519423623188149, "eval_auc": 0.8535259482590131, "eval_f1": 0.539378614691762, "eval_loss": 0.4177582859992981, "eval_mcc": 0.539618204034957, "eval_precision": 0.42203920774630316, "eval_recall": 0.7470924582868284, "eval_runtime": 4668.3776, "eval_samples_per_second": 24.307, "eval_steps_per_second": 4.861, "step": 180132 }, { "epoch": 2.0, "learning_rate": 0.0004276233452138895, "loss": 0.0324, "step": 180200 }, { "epoch": 2.0, "learning_rate": 0.0004273361993839469, "loss": 0.0406, "step": 180400 }, { "epoch": 2.01, "learning_rate": 0.0004270488612397328, "loss": 0.0277, "step": 180600 }, { "epoch": 2.01, "learning_rate": 0.0004267613311696915, "loss": 0.0353, "step": 180800 }, { "epoch": 2.01, "learning_rate": 0.0004264736095625267, "loss": 0.0335, "step": 181000 }, { "epoch": 2.01, "learning_rate": 0.0004261856968072011, "loss": 0.0364, "step": 181200 }, { "epoch": 2.01, "learning_rate": 0.0004258975932929357, "loss": 0.0426, "step": 181400 }, { "epoch": 2.02, "learning_rate": 0.0004256092994092096, "loss": 0.037, "step": 181600 }, { "epoch": 2.02, "learning_rate": 0.00042532081554575905, "loss": 0.037, "step": 181800 }, { "epoch": 2.02, "learning_rate": 0.0004250321420925771, "loss": 0.0401, "step": 182000 }, { "epoch": 2.02, "learning_rate": 0.00042474472422316436, "loss": 0.0407, "step": 182200 }, { "epoch": 2.03, "learning_rate": 0.0004244556737045969, "loss": 0.0413, "step": 182400 }, { "epoch": 2.03, "learning_rate": 0.0004241664347658587, "loss": 0.0451, "step": 182600 }, { "epoch": 2.03, "learning_rate": 0.00042387700779796374, "loss": 0.0547, "step": 182800 }, { "epoch": 2.03, "learning_rate": 0.00042358739319218, "loss": 0.0399, "step": 183000 }, { "epoch": 2.03, "learning_rate": 0.00042329759134002935, "loss": 0.0412, "step": 183200 }, { "epoch": 2.04, "learning_rate": 0.0004230076026332867, "loss": 0.0334, "step": 183400 }, { "epoch": 2.04, "learning_rate": 0.00042271742746397944, "loss": 0.0356, "step": 183600 }, { "epoch": 2.04, "learning_rate": 0.0004224270662243872, "loss": 0.0338, "step": 183800 }, { "epoch": 2.04, "learning_rate": 0.00042213651930704104, "loss": 0.0401, "step": 184000 }, { "epoch": 2.05, "learning_rate": 0.0004218457871047231, "loss": 0.0317, "step": 184200 }, { "epoch": 2.05, "learning_rate": 0.00042155632505520553, "loss": 0.0571, "step": 184400 }, { "epoch": 2.05, "learning_rate": 0.0004212652243838064, "loss": 0.0311, "step": 184600 }, { "epoch": 2.05, "learning_rate": 0.0004209739396053142, "loss": 0.0351, "step": 184800 }, { "epoch": 2.05, "learning_rate": 0.0004206824711135088, "loss": 0.0351, "step": 185000 }, { "epoch": 2.06, "learning_rate": 0.00042039081930241793, "loss": 0.034, "step": 185200 }, { "epoch": 2.06, "learning_rate": 0.0004200989845663176, "loss": 0.0324, "step": 185400 }, { "epoch": 2.06, "learning_rate": 0.0004198069672997307, "loss": 0.0345, "step": 185600 }, { "epoch": 2.06, "learning_rate": 0.00041951476789742724, "loss": 0.0323, "step": 185800 }, { "epoch": 2.07, "learning_rate": 0.0004192223867544232, "loss": 0.0523, "step": 186000 }, { "epoch": 2.07, "learning_rate": 0.0004189298242659803, "loss": 0.0313, "step": 186200 }, { "epoch": 2.07, "learning_rate": 0.0004186370808276056, "loss": 0.0283, "step": 186400 }, { "epoch": 2.07, "learning_rate": 0.00041834415683505034, "loss": 0.0387, "step": 186600 }, { "epoch": 2.07, "learning_rate": 0.00041805105268431043, "loss": 0.036, "step": 186800 }, { "epoch": 2.08, "learning_rate": 0.0004177577687716249, "loss": 0.0373, "step": 187000 }, { "epoch": 2.08, "learning_rate": 0.00041746430549347604, "loss": 0.0447, "step": 187200 }, { "epoch": 2.08, "learning_rate": 0.00041717066324658837, "loss": 0.0391, "step": 187400 }, { "epoch": 2.08, "learning_rate": 0.00041687684242792863, "loss": 0.0389, "step": 187600 }, { "epoch": 2.09, "learning_rate": 0.0004165828434347047, "loss": 0.0397, "step": 187800 }, { "epoch": 2.09, "learning_rate": 0.00041628866666436554, "loss": 0.0349, "step": 188000 }, { "epoch": 2.09, "learning_rate": 0.00041599431251460045, "loss": 0.0341, "step": 188200 }, { "epoch": 2.09, "learning_rate": 0.00041570125447857746, "loss": 0.0452, "step": 188400 }, { "epoch": 2.09, "learning_rate": 0.0004154065476459124, "loss": 0.0365, "step": 188600 }, { "epoch": 2.1, "learning_rate": 0.00041511166462633294, "loss": 0.0373, "step": 188800 }, { "epoch": 2.1, "learning_rate": 0.00041481660581848303, "loss": 0.0428, "step": 189000 }, { "epoch": 2.1, "learning_rate": 0.0004145213716212442, "loss": 0.0327, "step": 189200 }, { "epoch": 2.1, "learning_rate": 0.00041422596243373535, "loss": 0.0463, "step": 189400 }, { "epoch": 2.11, "learning_rate": 0.00041393037865531177, "loss": 0.0398, "step": 189600 }, { "epoch": 2.11, "learning_rate": 0.00041363462068556465, "loss": 0.0463, "step": 189800 }, { "epoch": 2.11, "learning_rate": 0.0004133386889243209, "loss": 0.0442, "step": 190000 }, { "epoch": 2.11, "learning_rate": 0.0004130425837716421, "loss": 0.0349, "step": 190200 }, { "epoch": 2.11, "learning_rate": 0.00041274630562782445, "loss": 0.0319, "step": 190400 }, { "epoch": 2.12, "learning_rate": 0.000412449854893398, "loss": 0.0408, "step": 190600 }, { "epoch": 2.12, "learning_rate": 0.000412153231969126, "loss": 0.0364, "step": 190800 }, { "epoch": 2.12, "learning_rate": 0.0004118564372560046, "loss": 0.036, "step": 191000 }, { "epoch": 2.12, "learning_rate": 0.0004115594711552622, "loss": 0.0337, "step": 191200 }, { "epoch": 2.13, "learning_rate": 0.00041126233406835866, "loss": 0.0269, "step": 191400 }, { "epoch": 2.13, "learning_rate": 0.00041096502639698535, "loss": 0.0394, "step": 191600 }, { "epoch": 2.13, "learning_rate": 0.00041066903635499747, "loss": 0.0452, "step": 191800 }, { "epoch": 2.13, "learning_rate": 0.0004103713895685811, "loss": 0.0343, "step": 192000 }, { "epoch": 2.13, "learning_rate": 0.00041007357340213746, "loss": 0.034, "step": 192200 }, { "epoch": 2.14, "learning_rate": 0.0004097755882582756, "loss": 0.0326, "step": 192400 }, { "epoch": 2.14, "learning_rate": 0.0004094774345398333, "loss": 0.0267, "step": 192600 }, { "epoch": 2.14, "learning_rate": 0.0004091791126498761, "loss": 0.0375, "step": 192800 }, { "epoch": 2.14, "learning_rate": 0.0004088806229916969, "loss": 0.038, "step": 193000 }, { "epoch": 2.15, "learning_rate": 0.0004085819659688153, "loss": 0.0372, "step": 193200 }, { "epoch": 2.15, "learning_rate": 0.0004082831419849773, "loss": 0.0443, "step": 193400 }, { "epoch": 2.15, "learning_rate": 0.00040798415144415443, "loss": 0.0423, "step": 193600 }, { "epoch": 2.15, "learning_rate": 0.000407686490946648, "loss": 0.0472, "step": 193800 }, { "epoch": 2.15, "learning_rate": 0.00040738716933240616, "loss": 0.0412, "step": 194000 }, { "epoch": 2.16, "learning_rate": 0.0004070876823724196, "loss": 0.0447, "step": 194200 }, { "epoch": 2.16, "learning_rate": 0.00040678803047155615, "loss": 0.0319, "step": 194400 }, { "epoch": 2.16, "learning_rate": 0.00040648821403490654, "loss": 0.0345, "step": 194600 }, { "epoch": 2.16, "learning_rate": 0.00040618823346778437, "loss": 0.041, "step": 194800 }, { "epoch": 2.17, "learning_rate": 0.0004058880891757246, "loss": 0.0414, "step": 195000 }, { "epoch": 2.17, "learning_rate": 0.000405587781564484, "loss": 0.0527, "step": 195200 }, { "epoch": 2.17, "learning_rate": 0.00040528731104003975, "loss": 0.0413, "step": 195400 }, { "epoch": 2.17, "learning_rate": 0.0004049866780085894, "loss": 0.0515, "step": 195600 }, { "epoch": 2.17, "learning_rate": 0.0004046858828765502, "loss": 0.0297, "step": 195800 }, { "epoch": 2.18, "learning_rate": 0.00040438492605055854, "loss": 0.032, "step": 196000 }, { "epoch": 2.18, "learning_rate": 0.00040408681991563507, "loss": 0.0316, "step": 196200 }, { "epoch": 2.18, "learning_rate": 0.0004037855425293061, "loss": 0.0368, "step": 196400 }, { "epoch": 2.18, "learning_rate": 0.00040348410466616914, "loss": 0.0326, "step": 196600 }, { "epoch": 2.19, "learning_rate": 0.0004031825067337296, "loss": 0.0327, "step": 196800 }, { "epoch": 2.19, "learning_rate": 0.00040288074913970913, "loss": 0.0334, "step": 197000 }, { "epoch": 2.19, "learning_rate": 0.00040257883229204524, "loss": 0.0476, "step": 197200 }, { "epoch": 2.19, "learning_rate": 0.0004022767565988907, "loss": 0.0317, "step": 197400 }, { "epoch": 2.19, "learning_rate": 0.00040197452246861295, "loss": 0.0393, "step": 197600 }, { "epoch": 2.2, "learning_rate": 0.00040167213030979393, "loss": 0.0331, "step": 197800 }, { "epoch": 2.2, "learning_rate": 0.0004013695805312289, "loss": 0.0377, "step": 198000 }, { "epoch": 2.2, "learning_rate": 0.0004010668735419263, "loss": 0.0348, "step": 198200 }, { "epoch": 2.2, "learning_rate": 0.00040076400975110705, "loss": 0.0395, "step": 198400 }, { "epoch": 2.21, "learning_rate": 0.0004004609895682043, "loss": 0.0491, "step": 198600 }, { "epoch": 2.21, "learning_rate": 0.00040015781340286215, "loss": 0.0335, "step": 198800 }, { "epoch": 2.21, "learning_rate": 0.000399854481664936, "loss": 0.0373, "step": 199000 }, { "epoch": 2.21, "learning_rate": 0.0003995509947644914, "loss": 0.0302, "step": 199200 }, { "epoch": 2.21, "learning_rate": 0.00039924735311180367, "loss": 0.0394, "step": 199400 }, { "epoch": 2.22, "learning_rate": 0.00039894355711735737, "loss": 0.0379, "step": 199600 }, { "epoch": 2.22, "learning_rate": 0.00039863960719184547, "loss": 0.0343, "step": 199800 }, { "epoch": 2.22, "learning_rate": 0.00039833550374616944, "loss": 0.0347, "step": 200000 }, { "epoch": 2.22, "learning_rate": 0.000398031247191438, "loss": 0.0417, "step": 200200 }, { "epoch": 2.23, "learning_rate": 0.0003977283603643842, "loss": 0.0339, "step": 200400 }, { "epoch": 2.23, "learning_rate": 0.0003974253227681304, "loss": 0.0394, "step": 200600 }, { "epoch": 2.23, "learning_rate": 0.00039712061087165875, "loss": 0.0275, "step": 200800 }, { "epoch": 2.23, "learning_rate": 0.00039681574750851094, "loss": 0.0315, "step": 201000 }, { "epoch": 2.23, "learning_rate": 0.00039651073309082295, "loss": 0.0416, "step": 201200 }, { "epoch": 2.24, "learning_rate": 0.0003962055680309351, "loss": 0.0353, "step": 201400 }, { "epoch": 2.24, "learning_rate": 0.00039590025274139143, "loss": 0.0455, "step": 201600 }, { "epoch": 2.24, "learning_rate": 0.00039559478763493897, "loss": 0.0309, "step": 201800 }, { "epoch": 2.24, "learning_rate": 0.00039528917312452736, "loss": 0.0414, "step": 202000 }, { "epoch": 2.25, "learning_rate": 0.000394983409623308, "loss": 0.0316, "step": 202200 }, { "epoch": 2.25, "learning_rate": 0.00039467749754463414, "loss": 0.0376, "step": 202400 }, { "epoch": 2.25, "learning_rate": 0.00039437143730205944, "loss": 0.0274, "step": 202600 }, { "epoch": 2.25, "learning_rate": 0.00039406522930933804, "loss": 0.0304, "step": 202800 }, { "epoch": 2.25, "learning_rate": 0.00039376040612288223, "loss": 0.0321, "step": 203000 }, { "epoch": 2.26, "learning_rate": 0.00039345390460550806, "loss": 0.0333, "step": 203200 }, { "epoch": 2.26, "learning_rate": 0.0003931472565783736, "loss": 0.0447, "step": 203400 }, { "epoch": 2.26, "learning_rate": 0.00039284046245602756, "loss": 0.0327, "step": 203600 }, { "epoch": 2.26, "learning_rate": 0.00039253352265321636, "loss": 0.0326, "step": 203800 }, { "epoch": 2.27, "learning_rate": 0.00039222643758488296, "loss": 0.0356, "step": 204000 }, { "epoch": 2.27, "learning_rate": 0.0003919192076661671, "loss": 0.0488, "step": 204200 }, { "epoch": 2.27, "learning_rate": 0.0003916118333124041, "loss": 0.0457, "step": 204400 }, { "epoch": 2.27, "learning_rate": 0.0003913043149391246, "loss": 0.0325, "step": 204600 }, { "epoch": 2.27, "learning_rate": 0.00039099665296205406, "loss": 0.0416, "step": 204800 }, { "epoch": 2.28, "learning_rate": 0.00039068884779711185, "loss": 0.0343, "step": 205000 }, { "epoch": 2.28, "learning_rate": 0.00039038089986041113, "loss": 0.0369, "step": 205200 }, { "epoch": 2.28, "learning_rate": 0.00039007280956825785, "loss": 0.0438, "step": 205400 }, { "epoch": 2.28, "learning_rate": 0.00038976457733715054, "loss": 0.0346, "step": 205600 }, { "epoch": 2.28, "learning_rate": 0.0003894562035837796, "loss": 0.0424, "step": 205800 }, { "epoch": 2.29, "learning_rate": 0.0003891476887250268, "loss": 0.0318, "step": 206000 }, { "epoch": 2.29, "learning_rate": 0.0003888390331779644, "loss": 0.0502, "step": 206200 }, { "epoch": 2.29, "learning_rate": 0.0003885302373598554, "loss": 0.0308, "step": 206400 }, { "epoch": 2.29, "learning_rate": 0.0003882213016881519, "loss": 0.0282, "step": 206600 }, { "epoch": 2.3, "learning_rate": 0.0003879122265804952, "loss": 0.0298, "step": 206800 }, { "epoch": 2.3, "learning_rate": 0.00038760301245471525, "loss": 0.0294, "step": 207000 }, { "epoch": 2.3, "learning_rate": 0.0003872936597288298, "loss": 0.0327, "step": 207200 }, { "epoch": 2.3, "learning_rate": 0.00038698416882104394, "loss": 0.0299, "step": 207400 }, { "epoch": 2.3, "learning_rate": 0.0003866745401497498, "loss": 0.0438, "step": 207600 }, { "epoch": 2.31, "learning_rate": 0.00038636477413352545, "loss": 0.0346, "step": 207800 }, { "epoch": 2.31, "learning_rate": 0.00038605642104575763, "loss": 0.0451, "step": 208000 }, { "epoch": 2.31, "learning_rate": 0.00038574638227764326, "loss": 0.0363, "step": 208200 }, { "epoch": 2.31, "learning_rate": 0.0003854362074193489, "loss": 0.0337, "step": 208400 }, { "epoch": 2.32, "learning_rate": 0.00038512589689019124, "loss": 0.0313, "step": 208600 }, { "epoch": 2.32, "learning_rate": 0.0003848154511096703, "loss": 0.0273, "step": 208800 }, { "epoch": 2.32, "learning_rate": 0.0003845048704974686, "loss": 0.0339, "step": 209000 }, { "epoch": 2.32, "learning_rate": 0.0003841941554734517, "loss": 0.0302, "step": 209200 }, { "epoch": 2.32, "learning_rate": 0.00038388330645766595, "loss": 0.0382, "step": 209400 }, { "epoch": 2.33, "learning_rate": 0.0003835723238703396, "loss": 0.0288, "step": 209600 }, { "epoch": 2.33, "learning_rate": 0.000383261208131881, "loss": 0.0321, "step": 209800 }, { "epoch": 2.33, "learning_rate": 0.0003829499596628787, "loss": 0.0306, "step": 210000 }, { "epoch": 2.33, "learning_rate": 0.00038264013611641904, "loss": 0.0515, "step": 210200 }, { "epoch": 2.34, "learning_rate": 0.00038232862410720903, "loss": 0.0313, "step": 210400 }, { "epoch": 2.34, "learning_rate": 0.0003820169806281892, "loss": 0.0375, "step": 210600 }, { "epoch": 2.34, "learning_rate": 0.0003817052061006615, "loss": 0.0357, "step": 210800 }, { "epoch": 2.34, "learning_rate": 0.00038139330094610506, "loss": 0.0361, "step": 211000 }, { "epoch": 2.34, "learning_rate": 0.0003810812655861756, "loss": 0.0413, "step": 211200 }, { "epoch": 2.35, "learning_rate": 0.0003807691004427048, "loss": 0.0341, "step": 211400 }, { "epoch": 2.35, "learning_rate": 0.0003804568059377, "loss": 0.0368, "step": 211600 }, { "epoch": 2.35, "learning_rate": 0.00038014438249334305, "loss": 0.0346, "step": 211800 }, { "epoch": 2.35, "learning_rate": 0.0003798318305319905, "loss": 0.0334, "step": 212000 }, { "epoch": 2.36, "learning_rate": 0.0003795191504761724, "loss": 0.0307, "step": 212200 }, { "epoch": 2.36, "learning_rate": 0.000379206342748592, "loss": 0.0528, "step": 212400 }, { "epoch": 2.36, "learning_rate": 0.00037889340777212514, "loss": 0.0249, "step": 212600 }, { "epoch": 2.36, "learning_rate": 0.0003785819115936106, "loss": 0.0446, "step": 212800 }, { "epoch": 2.36, "learning_rate": 0.00037826872401964623, "loss": 0.0406, "step": 213000 }, { "epoch": 2.37, "learning_rate": 0.00037795541046433556, "loss": 0.0422, "step": 213200 }, { "epoch": 2.37, "learning_rate": 0.00037764197135123833, "loss": 0.045, "step": 213400 }, { "epoch": 2.37, "learning_rate": 0.00037732840710408375, "loss": 0.0342, "step": 213600 }, { "epoch": 2.37, "learning_rate": 0.0003770147181467706, "loss": 0.0318, "step": 213800 }, { "epoch": 2.38, "learning_rate": 0.0003767009049033662, "loss": 0.0322, "step": 214000 }, { "epoch": 2.38, "learning_rate": 0.0003763885377910363, "loss": 0.0313, "step": 214200 }, { "epoch": 2.38, "learning_rate": 0.00037607447786445336, "loss": 0.0457, "step": 214400 }, { "epoch": 2.38, "learning_rate": 0.00037576029492286314, "loss": 0.0376, "step": 214600 }, { "epoch": 2.38, "learning_rate": 0.0003754459893910007, "loss": 0.0431, "step": 214800 }, { "epoch": 2.39, "learning_rate": 0.00037513156169376654, "loss": 0.0434, "step": 215000 }, { "epoch": 2.39, "learning_rate": 0.00037481701225622674, "loss": 0.0592, "step": 215200 }, { "epoch": 2.39, "learning_rate": 0.00037450234150361166, "loss": 0.0461, "step": 215400 }, { "epoch": 2.39, "learning_rate": 0.00037418754986131577, "loss": 0.0446, "step": 215600 }, { "epoch": 2.4, "learning_rate": 0.0003738726377548968, "loss": 0.0396, "step": 215800 }, { "epoch": 2.4, "learning_rate": 0.00037355760561007565, "loss": 0.0411, "step": 216000 }, { "epoch": 2.4, "learning_rate": 0.00037324245385273537, "loss": 0.043, "step": 216200 }, { "epoch": 2.4, "learning_rate": 0.00037292718290892054, "loss": 0.0417, "step": 216400 }, { "epoch": 2.4, "learning_rate": 0.00037261179320483695, "loss": 0.0392, "step": 216600 }, { "epoch": 2.41, "learning_rate": 0.000372296285166851, "loss": 0.0352, "step": 216800 }, { "epoch": 2.41, "learning_rate": 0.0003719806592214891, "loss": 0.0419, "step": 217000 }, { "epoch": 2.41, "learning_rate": 0.000371664915795437, "loss": 0.0312, "step": 217200 }, { "epoch": 2.41, "learning_rate": 0.00037134905531553924, "loss": 0.0312, "step": 217400 }, { "epoch": 2.42, "learning_rate": 0.00037103307820879854, "loss": 0.0347, "step": 217600 }, { "epoch": 2.42, "learning_rate": 0.00037071698490237547, "loss": 0.0434, "step": 217800 }, { "epoch": 2.42, "learning_rate": 0.00037040077582358756, "loss": 0.0314, "step": 218000 }, { "epoch": 2.42, "learning_rate": 0.00037008445139990895, "loss": 0.0359, "step": 218200 }, { "epoch": 2.42, "learning_rate": 0.00036976959454082325, "loss": 0.048, "step": 218400 }, { "epoch": 2.43, "learning_rate": 0.00036945304128179135, "loss": 0.031, "step": 218600 }, { "epoch": 2.43, "learning_rate": 0.00036913637395908413, "loss": 0.0411, "step": 218800 }, { "epoch": 2.43, "learning_rate": 0.00036881959300079505, "loss": 0.0337, "step": 219000 }, { "epoch": 2.43, "learning_rate": 0.0003685026988351713, "loss": 0.0428, "step": 219200 }, { "epoch": 2.44, "learning_rate": 0.00036818569189061304, "loss": 0.037, "step": 219400 }, { "epoch": 2.44, "learning_rate": 0.000367868572595673, "loss": 0.0322, "step": 219600 }, { "epoch": 2.44, "learning_rate": 0.00036755134137905575, "loss": 0.0476, "step": 219800 }, { "epoch": 2.44, "learning_rate": 0.0003672339986696172, "loss": 0.031, "step": 220000 }, { "epoch": 2.44, "learning_rate": 0.0003669165448963639, "loss": 0.0477, "step": 220200 }, { "epoch": 2.45, "learning_rate": 0.00036659898048845257, "loss": 0.0314, "step": 220400 }, { "epoch": 2.45, "learning_rate": 0.0003662813058751894, "loss": 0.039, "step": 220600 }, { "epoch": 2.45, "learning_rate": 0.0003659635214860299, "loss": 0.0382, "step": 220800 }, { "epoch": 2.45, "learning_rate": 0.0003656456277505776, "loss": 0.0337, "step": 221000 }, { "epoch": 2.46, "learning_rate": 0.00036532921538206284, "loss": 0.0472, "step": 221200 }, { "epoch": 2.46, "learning_rate": 0.0003650111047847903, "loss": 0.0326, "step": 221400 }, { "epoch": 2.46, "learning_rate": 0.00036469288612877014, "loss": 0.0307, "step": 221600 }, { "epoch": 2.46, "learning_rate": 0.00036437455984419297, "loss": 0.0319, "step": 221800 }, { "epoch": 2.46, "learning_rate": 0.00036405612636139517, "loss": 0.033, "step": 222000 }, { "epoch": 2.47, "learning_rate": 0.00036373758611085775, "loss": 0.0326, "step": 222200 }, { "epoch": 2.47, "learning_rate": 0.0003634189395232064, "loss": 0.0405, "step": 222400 }, { "epoch": 2.47, "learning_rate": 0.0003631001870292103, "loss": 0.0311, "step": 222600 }, { "epoch": 2.47, "learning_rate": 0.00036278132905978187, "loss": 0.0436, "step": 222800 }, { "epoch": 2.48, "learning_rate": 0.00036246236604597627, "loss": 0.0534, "step": 223000 }, { "epoch": 2.48, "learning_rate": 0.0003621432984189902, "loss": 0.0374, "step": 223200 }, { "epoch": 2.48, "learning_rate": 0.0003618241266101624, "loss": 0.044, "step": 223400 }, { "epoch": 2.48, "learning_rate": 0.00036150485105097217, "loss": 0.037, "step": 223600 }, { "epoch": 2.48, "learning_rate": 0.00036118706932371984, "loss": 0.0545, "step": 223800 }, { "epoch": 2.49, "learning_rate": 0.0003608675880721636, "loss": 0.0369, "step": 224000 }, { "epoch": 2.49, "learning_rate": 0.00036054800436336213, "loss": 0.0513, "step": 224200 }, { "epoch": 2.49, "learning_rate": 0.00036022991731109454, "loss": 0.0339, "step": 224400 }, { "epoch": 2.49, "learning_rate": 0.00035991013049093876, "loss": 0.0425, "step": 224600 }, { "epoch": 2.5, "learning_rate": 0.0003595902425078975, "loss": 0.0498, "step": 224800 }, { "epoch": 2.5, "learning_rate": 0.0003592702537944181, "loss": 0.0323, "step": 225000 }, { "epoch": 2.5, "learning_rate": 0.0003589501647830843, "loss": 0.0365, "step": 225200 }, { "epoch": 2.5, "learning_rate": 0.0003586299759066153, "loss": 0.0373, "step": 225400 }, { "epoch": 2.5, "learning_rate": 0.0003583096875978653, "loss": 0.0474, "step": 225600 }, { "epoch": 2.51, "learning_rate": 0.0003579893002898231, "loss": 0.0323, "step": 225800 }, { "epoch": 2.51, "learning_rate": 0.000357668814415611, "loss": 0.0331, "step": 226000 }, { "epoch": 2.51, "learning_rate": 0.0003573482304084849, "loss": 0.0322, "step": 226200 }, { "epoch": 2.51, "learning_rate": 0.0003570275487018331, "loss": 0.045, "step": 226400 }, { "epoch": 2.52, "learning_rate": 0.0003567067697291761, "loss": 0.0485, "step": 226600 }, { "epoch": 2.52, "learning_rate": 0.00035638589392416594, "loss": 0.028, "step": 226800 }, { "epoch": 2.52, "learning_rate": 0.0003560649217205855, "loss": 0.0559, "step": 227000 }, { "epoch": 2.52, "learning_rate": 0.0003557438535523479, "loss": 0.0237, "step": 227200 }, { "epoch": 2.52, "learning_rate": 0.0003554226898534961, "loss": 0.0347, "step": 227400 }, { "epoch": 2.53, "learning_rate": 0.00035510143105820224, "loss": 0.0367, "step": 227600 }, { "epoch": 2.53, "learning_rate": 0.0003547800776007668, "loss": 0.0328, "step": 227800 }, { "epoch": 2.53, "learning_rate": 0.00035445862991561857, "loss": 0.0342, "step": 228000 }, { "epoch": 2.53, "learning_rate": 0.00035413708843731344, "loss": 0.037, "step": 228200 }, { "epoch": 2.54, "learning_rate": 0.00035381545360053425, "loss": 0.0467, "step": 228400 }, { "epoch": 2.54, "learning_rate": 0.0003534937258400899, "loss": 0.0301, "step": 228600 }, { "epoch": 2.54, "learning_rate": 0.0003531719055909151, "loss": 0.0306, "step": 228800 }, { "epoch": 2.54, "learning_rate": 0.0003528499932880694, "loss": 0.0404, "step": 229000 }, { "epoch": 2.54, "learning_rate": 0.00035252798936673686, "loss": 0.0353, "step": 229200 }, { "epoch": 2.55, "learning_rate": 0.00035220589426222554, "loss": 0.0428, "step": 229400 }, { "epoch": 2.55, "learning_rate": 0.0003518837084099668, "loss": 0.0348, "step": 229600 }, { "epoch": 2.55, "learning_rate": 0.00035156304385026743, "loss": 0.0445, "step": 229800 }, { "epoch": 2.55, "learning_rate": 0.0003512406782575958, "loss": 0.0357, "step": 230000 }, { "epoch": 2.56, "learning_rate": 0.0003509182232220247, "loss": 0.0367, "step": 230200 }, { "epoch": 2.56, "learning_rate": 0.00035059567917947187, "loss": 0.0354, "step": 230400 }, { "epoch": 2.56, "learning_rate": 0.0003502730465659754, "loss": 0.0326, "step": 230600 }, { "epoch": 2.56, "learning_rate": 0.0003499503258176932, "loss": 0.0456, "step": 230800 }, { "epoch": 2.56, "learning_rate": 0.00034962751737090234, "loss": 0.0328, "step": 231000 }, { "epoch": 2.57, "learning_rate": 0.00034930462166199823, "loss": 0.0316, "step": 231200 }, { "epoch": 2.57, "learning_rate": 0.0003489832542554235, "loss": 0.0334, "step": 231400 }, { "epoch": 2.57, "learning_rate": 0.00034866018576280963, "loss": 0.0346, "step": 231600 }, { "epoch": 2.57, "learning_rate": 0.00034833703131579067, "loss": 0.0405, "step": 231800 }, { "epoch": 2.58, "learning_rate": 0.00034801379135123, "loss": 0.0374, "step": 232000 }, { "epoch": 2.58, "learning_rate": 0.00034769046630610647, "loss": 0.0412, "step": 232200 }, { "epoch": 2.58, "learning_rate": 0.00034736705661751396, "loss": 0.0509, "step": 232400 }, { "epoch": 2.58, "learning_rate": 0.0003470435627226611, "loss": 0.0282, "step": 232600 }, { "epoch": 2.58, "learning_rate": 0.00034671998505886976, "loss": 0.0363, "step": 232800 }, { "epoch": 2.59, "learning_rate": 0.00034639632406357557, "loss": 0.0413, "step": 233000 }, { "epoch": 2.59, "learning_rate": 0.00034607258017432666, "loss": 0.0406, "step": 233200 }, { "epoch": 2.59, "learning_rate": 0.0003457487538287831, "loss": 0.038, "step": 233400 }, { "epoch": 2.59, "learning_rate": 0.0003454248454647165, "loss": 0.0458, "step": 233600 }, { "epoch": 2.6, "learning_rate": 0.0003451008555200096, "loss": 0.0419, "step": 233800 }, { "epoch": 2.6, "learning_rate": 0.0003447767844326548, "loss": 0.0435, "step": 234000 }, { "epoch": 2.6, "learning_rate": 0.00034445263264075497, "loss": 0.0377, "step": 234200 }, { "epoch": 2.6, "learning_rate": 0.0003441284005825215, "loss": 0.0435, "step": 234400 }, { "epoch": 2.6, "learning_rate": 0.0003438040886962745, "loss": 0.0332, "step": 234600 }, { "epoch": 2.61, "learning_rate": 0.00034348131957357737, "loss": 0.0272, "step": 234800 }, { "epoch": 2.61, "learning_rate": 0.000343156849740359, "loss": 0.0467, "step": 235000 }, { "epoch": 2.61, "learning_rate": 0.0003428323013925391, "loss": 0.0371, "step": 235200 }, { "epoch": 2.61, "learning_rate": 0.0003425076749688654, "loss": 0.0353, "step": 235400 }, { "epoch": 2.62, "learning_rate": 0.00034218297090819104, "loss": 0.0362, "step": 235600 }, { "epoch": 2.62, "learning_rate": 0.0003418581896494743, "loss": 0.0451, "step": 235800 }, { "epoch": 2.62, "learning_rate": 0.0003415333316317777, "loss": 0.0409, "step": 236000 }, { "epoch": 2.62, "learning_rate": 0.0003412083972942675, "loss": 0.0378, "step": 236200 }, { "epoch": 2.62, "learning_rate": 0.00034088338707621325, "loss": 0.0349, "step": 236400 }, { "epoch": 2.63, "learning_rate": 0.00034055830141698705, "loss": 0.0341, "step": 236600 }, { "epoch": 2.63, "learning_rate": 0.00034023314075606295, "loss": 0.0446, "step": 236800 }, { "epoch": 2.63, "learning_rate": 0.0003399079055330163, "loss": 0.0272, "step": 237000 }, { "epoch": 2.63, "learning_rate": 0.0003395825961875236, "loss": 0.0534, "step": 237200 }, { "epoch": 2.64, "learning_rate": 0.0003392572131593611, "loss": 0.0319, "step": 237400 }, { "epoch": 2.64, "learning_rate": 0.0003389317568884048, "loss": 0.0324, "step": 237600 }, { "epoch": 2.64, "learning_rate": 0.00033860622781463015, "loss": 0.0357, "step": 237800 }, { "epoch": 2.64, "learning_rate": 0.00033828062637811024, "loss": 0.0343, "step": 238000 }, { "epoch": 2.64, "learning_rate": 0.0003379549530190165, "loss": 0.0492, "step": 238200 }, { "epoch": 2.65, "learning_rate": 0.00033762920817761766, "loss": 0.0424, "step": 238400 }, { "epoch": 2.65, "learning_rate": 0.0003373033922942788, "loss": 0.0366, "step": 238600 }, { "epoch": 2.65, "learning_rate": 0.00033697750580946096, "loss": 0.0435, "step": 238800 }, { "epoch": 2.65, "learning_rate": 0.0003366515491637211, "loss": 0.0386, "step": 239000 }, { "epoch": 2.66, "learning_rate": 0.0003363255227977108, "loss": 0.033, "step": 239200 }, { "epoch": 2.66, "learning_rate": 0.0003359994271521755, "loss": 0.0365, "step": 239400 }, { "epoch": 2.66, "learning_rate": 0.0003356748936608829, "loss": 0.0511, "step": 239600 }, { "epoch": 2.66, "learning_rate": 0.0003353486611198012, "loss": 0.0375, "step": 239800 }, { "epoch": 2.66, "learning_rate": 0.0003350223606197862, "loss": 0.0476, "step": 240000 }, { "epoch": 2.67, "learning_rate": 0.00033469599260195433, "loss": 0.0365, "step": 240200 }, { "epoch": 2.67, "learning_rate": 0.00033436955750751313, "loss": 0.0486, "step": 240400 }, { "epoch": 2.67, "learning_rate": 0.000334043055777761, "loss": 0.0383, "step": 240600 }, { "epoch": 2.67, "learning_rate": 0.00033371648785408635, "loss": 0.0411, "step": 240800 }, { "epoch": 2.68, "learning_rate": 0.0003333898541779671, "loss": 0.0384, "step": 241000 }, { "epoch": 2.68, "learning_rate": 0.0003330631551909699, "loss": 0.0375, "step": 241200 }, { "epoch": 2.68, "learning_rate": 0.00033273639133475, "loss": 0.0389, "step": 241400 }, { "epoch": 2.68, "learning_rate": 0.0003324095630510501, "loss": 0.0331, "step": 241600 }, { "epoch": 2.68, "learning_rate": 0.00033208267078169997, "loss": 0.0372, "step": 241800 }, { "epoch": 2.69, "learning_rate": 0.0003317557149686162, "loss": 0.0376, "step": 242000 }, { "epoch": 2.69, "learning_rate": 0.00033142869605380085, "loss": 0.0321, "step": 242200 }, { "epoch": 2.69, "learning_rate": 0.0003311016144793415, "loss": 0.0295, "step": 242400 }, { "epoch": 2.69, "learning_rate": 0.0003307744706874105, "loss": 0.0379, "step": 242600 }, { "epoch": 2.7, "learning_rate": 0.0003304472651202642, "loss": 0.0387, "step": 242800 }, { "epoch": 2.7, "learning_rate": 0.0003301199982202426, "loss": 0.0318, "step": 243000 }, { "epoch": 2.7, "learning_rate": 0.0003297926704297686, "loss": 0.0339, "step": 243200 }, { "epoch": 2.7, "learning_rate": 0.0003294652821913471, "loss": 0.0334, "step": 243400 }, { "epoch": 2.7, "learning_rate": 0.00032914110872557593, "loss": 0.0584, "step": 243600 }, { "epoch": 2.71, "learning_rate": 0.00032881360151253673, "loss": 0.0227, "step": 243800 }, { "epoch": 2.71, "learning_rate": 0.0003284860351751257, "loss": 0.0338, "step": 244000 }, { "epoch": 2.71, "learning_rate": 0.00032815841015617015, "loss": 0.0395, "step": 244200 }, { "epoch": 2.71, "learning_rate": 0.00032783072689857734, "loss": 0.0367, "step": 244400 }, { "epoch": 2.72, "learning_rate": 0.00032750298584533264, "loss": 0.0396, "step": 244600 }, { "epoch": 2.72, "learning_rate": 0.00032717518743950025, "loss": 0.0388, "step": 244800 }, { "epoch": 2.72, "learning_rate": 0.0003268473321242212, "loss": 0.0408, "step": 245000 }, { "epoch": 2.72, "learning_rate": 0.000326519420342714, "loss": 0.0466, "step": 245200 }, { "epoch": 2.72, "learning_rate": 0.0003261930925159188, "loss": 0.0296, "step": 245400 }, { "epoch": 2.73, "learning_rate": 0.0003258650694087094, "loss": 0.0349, "step": 245600 }, { "epoch": 2.73, "learning_rate": 0.00032553699116316475, "loss": 0.041, "step": 245800 }, { "epoch": 2.73, "learning_rate": 0.00032520885822280457, "loss": 0.0411, "step": 246000 }, { "epoch": 2.73, "learning_rate": 0.0003248806710312224, "loss": 0.0274, "step": 246200 }, { "epoch": 2.74, "learning_rate": 0.0003245524300320851, "loss": 0.0302, "step": 246400 }, { "epoch": 2.74, "learning_rate": 0.0003242241356691325, "loss": 0.0317, "step": 246600 }, { "epoch": 2.74, "learning_rate": 0.00032389578838617637, "loss": 0.0348, "step": 246800 }, { "epoch": 2.74, "learning_rate": 0.0003235673886271001, "loss": 0.0376, "step": 247000 }, { "epoch": 2.74, "learning_rate": 0.00032323893683585786, "loss": 0.0356, "step": 247200 }, { "epoch": 2.75, "learning_rate": 0.0003229104334564744, "loss": 0.0264, "step": 247400 }, { "epoch": 2.75, "learning_rate": 0.0003225818789330441, "loss": 0.0341, "step": 247600 }, { "epoch": 2.75, "learning_rate": 0.00032225327370973044, "loss": 0.0411, "step": 247800 }, { "epoch": 2.75, "learning_rate": 0.0003219246182307655, "loss": 0.044, "step": 248000 }, { "epoch": 2.76, "learning_rate": 0.00032159591294044924, "loss": 0.0399, "step": 248200 }, { "epoch": 2.76, "learning_rate": 0.000321267158283149, "loss": 0.0547, "step": 248400 }, { "epoch": 2.76, "learning_rate": 0.0003209383547032989, "loss": 0.0431, "step": 248600 }, { "epoch": 2.76, "learning_rate": 0.00032060950264539934, "loss": 0.042, "step": 248800 }, { "epoch": 2.76, "learning_rate": 0.0003202806025540157, "loss": 0.0343, "step": 249000 }, { "epoch": 2.77, "learning_rate": 0.0003199516548737788, "loss": 0.0419, "step": 249200 }, { "epoch": 2.77, "learning_rate": 0.0003196226600493838, "loss": 0.034, "step": 249400 }, { "epoch": 2.77, "learning_rate": 0.00031929361852558926, "loss": 0.0412, "step": 249600 }, { "epoch": 2.77, "learning_rate": 0.0003189645307472172, "loss": 0.0448, "step": 249800 }, { "epoch": 2.78, "learning_rate": 0.0003186370429403079, "loss": 0.0432, "step": 250000 }, { "epoch": 2.78, "learning_rate": 0.00031830786421321275, "loss": 0.0378, "step": 250200 }, { "epoch": 2.78, "learning_rate": 0.00031797864056415316, "loss": 0.0474, "step": 250400 }, { "epoch": 2.78, "learning_rate": 0.00031764937243819734, "loss": 0.0292, "step": 250600 }, { "epoch": 2.78, "learning_rate": 0.00031732006028047354, "loss": 0.0375, "step": 250800 }, { "epoch": 2.79, "learning_rate": 0.0003169907045361695, "loss": 0.0327, "step": 251000 }, { "epoch": 2.79, "learning_rate": 0.0003166613056505317, "loss": 0.0529, "step": 251200 }, { "epoch": 2.79, "learning_rate": 0.0003163318640688654, "loss": 0.0275, "step": 251400 }, { "epoch": 2.79, "learning_rate": 0.0003160023802365332, "loss": 0.0329, "step": 251600 }, { "epoch": 2.8, "learning_rate": 0.00031567285459895494, "loss": 0.0361, "step": 251800 }, { "epoch": 2.8, "learning_rate": 0.000315344935538739, "loss": 0.0404, "step": 252000 }, { "epoch": 2.8, "learning_rate": 0.00031501532783061626, "loss": 0.055, "step": 252200 }, { "epoch": 2.8, "learning_rate": 0.00031468567965161557, "loss": 0.0295, "step": 252400 }, { "epoch": 2.8, "learning_rate": 0.00031435599144737894, "loss": 0.0306, "step": 252600 }, { "epoch": 2.81, "learning_rate": 0.0003140262636636023, "loss": 0.0364, "step": 252800 }, { "epoch": 2.81, "learning_rate": 0.0003136981456772313, "loss": 0.0397, "step": 253000 }, { "epoch": 2.81, "learning_rate": 0.0003133683402640077, "loss": 0.0398, "step": 253200 }, { "epoch": 2.81, "learning_rate": 0.0003130384966064217, "loss": 0.0501, "step": 253400 }, { "epoch": 2.82, "learning_rate": 0.00031270861515037967, "loss": 0.0415, "step": 253600 }, { "epoch": 2.82, "learning_rate": 0.00031237869634183894, "loss": 0.0487, "step": 253800 }, { "epoch": 2.82, "learning_rate": 0.00031204874062680743, "loss": 0.0338, "step": 254000 }, { "epoch": 2.82, "learning_rate": 0.0003117187484513428, "loss": 0.0319, "step": 254200 }, { "epoch": 2.82, "learning_rate": 0.0003113887202615521, "loss": 0.0379, "step": 254400 }, { "epoch": 2.83, "learning_rate": 0.0003110586565035912, "loss": 0.0336, "step": 254600 }, { "epoch": 2.83, "learning_rate": 0.00031072855762366374, "loss": 0.0352, "step": 254800 }, { "epoch": 2.83, "learning_rate": 0.0003103984240680211, "loss": 0.0351, "step": 255000 }, { "epoch": 2.83, "learning_rate": 0.00031006990720629403, "loss": 0.0361, "step": 255200 }, { "epoch": 2.84, "learning_rate": 0.0003097397058059668, "loss": 0.0369, "step": 255400 }, { "epoch": 2.84, "learning_rate": 0.0003094094710667249, "loss": 0.0409, "step": 255600 }, { "epoch": 2.84, "learning_rate": 0.00030907920343500357, "loss": 0.0378, "step": 255800 }, { "epoch": 2.84, "learning_rate": 0.000308748903357282, "loss": 0.0342, "step": 256000 }, { "epoch": 2.84, "learning_rate": 0.00030841857128008354, "loss": 0.0374, "step": 256200 }, { "epoch": 2.85, "learning_rate": 0.0003080882076499748, "loss": 0.0332, "step": 256400 }, { "epoch": 2.85, "learning_rate": 0.00030775781291356494, "loss": 0.0468, "step": 256600 }, { "epoch": 2.85, "learning_rate": 0.0003074273875175052, "loss": 0.0353, "step": 256800 }, { "epoch": 2.85, "learning_rate": 0.00030709693190848823, "loss": 0.0385, "step": 257000 }, { "epoch": 2.86, "learning_rate": 0.0003067664465332478, "loss": 0.0315, "step": 257200 }, { "epoch": 2.86, "learning_rate": 0.00030643593183855736, "loss": 0.0357, "step": 257400 }, { "epoch": 2.86, "learning_rate": 0.0003061053882712305, "loss": 0.0274, "step": 257600 }, { "epoch": 2.86, "learning_rate": 0.0003057764692080551, "loss": 0.0389, "step": 257800 }, { "epoch": 2.86, "learning_rate": 0.00030544586937483406, "loss": 0.0421, "step": 258000 }, { "epoch": 2.87, "learning_rate": 0.0003051152420074138, "loss": 0.051, "step": 258200 }, { "epoch": 2.87, "learning_rate": 0.00030478458755276014, "loss": 0.0422, "step": 258400 }, { "epoch": 2.87, "learning_rate": 0.00030445390645787555, "loss": 0.0339, "step": 258600 }, { "epoch": 2.87, "learning_rate": 0.0003041231991697982, "loss": 0.0321, "step": 258800 }, { "epoch": 2.88, "learning_rate": 0.00030379246613560206, "loss": 0.0348, "step": 259000 }, { "epoch": 2.88, "learning_rate": 0.00030346170780239557, "loss": 0.0372, "step": 259200 }, { "epoch": 2.88, "learning_rate": 0.00030313092461732165, "loss": 0.0365, "step": 259400 }, { "epoch": 2.88, "learning_rate": 0.0003028001170275566, "loss": 0.0337, "step": 259600 }, { "epoch": 2.88, "learning_rate": 0.00030246928548030984, "loss": 0.044, "step": 259800 }, { "epoch": 2.89, "learning_rate": 0.00030213843042282306, "loss": 0.0406, "step": 260000 }, { "epoch": 2.89, "learning_rate": 0.0003018075523023699, "loss": 0.0301, "step": 260200 }, { "epoch": 2.89, "learning_rate": 0.000301476651566255, "loss": 0.0317, "step": 260400 }, { "epoch": 2.89, "learning_rate": 0.0003011457286618138, "loss": 0.0396, "step": 260600 }, { "epoch": 2.9, "learning_rate": 0.00030081478403641137, "loss": 0.0367, "step": 260800 }, { "epoch": 2.9, "learning_rate": 0.00030048381813744255, "loss": 0.0432, "step": 261000 }, { "epoch": 2.9, "learning_rate": 0.00030015283141233057, "loss": 0.0322, "step": 261200 }, { "epoch": 2.9, "learning_rate": 0.00029982182430852706, "loss": 0.0429, "step": 261400 }, { "epoch": 2.9, "learning_rate": 0.00029949079727351096, "loss": 0.0438, "step": 261600 }, { "epoch": 2.91, "learning_rate": 0.00029915975075478854, "loss": 0.0414, "step": 261800 }, { "epoch": 2.91, "learning_rate": 0.00029882868519989206, "loss": 0.0428, "step": 262000 }, { "epoch": 2.91, "learning_rate": 0.0002984976010563797, "loss": 0.0324, "step": 262200 }, { "epoch": 2.91, "learning_rate": 0.00029816980988300846, "loss": 0.048, "step": 262400 }, { "epoch": 2.92, "learning_rate": 0.000297838690079757, "loss": 0.0369, "step": 262600 }, { "epoch": 2.92, "learning_rate": 0.00029750755302623583, "loss": 0.0635, "step": 262800 }, { "epoch": 2.92, "learning_rate": 0.00029717639917009977, "loss": 0.0297, "step": 263000 }, { "epoch": 2.92, "learning_rate": 0.0002968452289590263, "loss": 0.0329, "step": 263200 }, { "epoch": 2.92, "learning_rate": 0.00029651404284071497, "loss": 0.0375, "step": 263400 }, { "epoch": 2.93, "learning_rate": 0.0002961828412628869, "loss": 0.0356, "step": 263600 }, { "epoch": 2.93, "learning_rate": 0.0002958516246732841, "loss": 0.0355, "step": 263800 }, { "epoch": 2.93, "learning_rate": 0.00029552039351966896, "loss": 0.0448, "step": 264000 }, { "epoch": 2.93, "learning_rate": 0.00029518914824982334, "loss": 0.0486, "step": 264200 }, { "epoch": 2.94, "learning_rate": 0.00029485788931154827, "loss": 0.031, "step": 264400 }, { "epoch": 2.94, "learning_rate": 0.0002945266171526634, "loss": 0.0435, "step": 264600 }, { "epoch": 2.94, "learning_rate": 0.00029419533222100616, "loss": 0.0297, "step": 264800 }, { "epoch": 2.94, "learning_rate": 0.0002938640349644311, "loss": 0.0443, "step": 265000 }, { "epoch": 2.94, "learning_rate": 0.0002935327258308096, "loss": 0.0377, "step": 265200 }, { "epoch": 2.95, "learning_rate": 0.0002932014052680292, "loss": 0.0351, "step": 265400 }, { "epoch": 2.95, "learning_rate": 0.00029287007372399266, "loss": 0.0322, "step": 265600 }, { "epoch": 2.95, "learning_rate": 0.0002925387316466176, "loss": 0.0434, "step": 265800 }, { "epoch": 2.95, "learning_rate": 0.00029220737948383607, "loss": 0.0328, "step": 266000 }, { "epoch": 2.96, "learning_rate": 0.0002918760176835935, "loss": 0.0465, "step": 266200 }, { "epoch": 2.96, "learning_rate": 0.00029154464669384867, "loss": 0.0366, "step": 266400 }, { "epoch": 2.96, "learning_rate": 0.0002912132669625725, "loss": 0.0377, "step": 266600 }, { "epoch": 2.96, "learning_rate": 0.00029088187893774783, "loss": 0.0312, "step": 266800 }, { "epoch": 2.96, "learning_rate": 0.0002905504830673687, "loss": 0.0301, "step": 267000 }, { "epoch": 2.97, "learning_rate": 0.0002902190797994398, "loss": 0.026, "step": 267200 }, { "epoch": 2.97, "learning_rate": 0.00028988766958197576, "loss": 0.0307, "step": 267400 }, { "epoch": 2.97, "learning_rate": 0.00028955625286300067, "loss": 0.0396, "step": 267600 }, { "epoch": 2.97, "learning_rate": 0.0002892264872187265, "loss": 0.0423, "step": 267800 }, { "epoch": 2.98, "learning_rate": 0.0002888950588677487, "loss": 0.0405, "step": 268000 }, { "epoch": 2.98, "learning_rate": 0.000288563625357142, "loss": 0.0391, "step": 268200 }, { "epoch": 2.98, "learning_rate": 0.0002882321871349619, "loss": 0.0308, "step": 268400 }, { "epoch": 2.98, "learning_rate": 0.00028790074464927053, "loss": 0.0306, "step": 268600 }, { "epoch": 2.98, "learning_rate": 0.0002875692983481353, "loss": 0.0408, "step": 268800 }, { "epoch": 2.99, "learning_rate": 0.0002872378486796292, "loss": 0.0339, "step": 269000 }, { "epoch": 2.99, "learning_rate": 0.0002869063960918296, "loss": 0.031, "step": 269200 }, { "epoch": 2.99, "learning_rate": 0.0002865749410328178, "loss": 0.0546, "step": 269400 }, { "epoch": 2.99, "learning_rate": 0.00028624348395067857, "loss": 0.0375, "step": 269600 }, { "epoch": 3.0, "learning_rate": 0.0002859120252934991, "loss": 0.0289, "step": 269800 }, { "epoch": 3.0, "learning_rate": 0.00028558056550936916, "loss": 0.0291, "step": 270000 }, { "epoch": 3.0, "eval_accuracy": 0.9486257590603431, "eval_auc": 0.8901825991549838, "eval_f1": 0.5480253402747288, "eval_loss": 0.29167044162750244, "eval_mcc": 0.5606123600429995, "eval_precision": 0.409794889805426, "eval_recall": 0.8269786150111385, "eval_runtime": 4667.9267, "eval_samples_per_second": 24.31, "eval_steps_per_second": 4.862, "step": 270198 } ], "logging_steps": 200, "max_steps": 540396, "num_train_epochs": 6, "save_steps": 500, "total_flos": 2.701282213665e+17, "trial_name": null, "trial_params": null }