| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.6425043183126764, | |
| "eval_steps": 500, | |
| "global_step": 43600, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00029999755171150014, | |
| "loss": 1.6385, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00029998908859460167, | |
| "loss": 0.8837, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00029997458076451173, | |
| "loss": 0.8445, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002999540288059106, | |
| "loss": 0.8257, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0002999274335470631, | |
| "loss": 0.8054, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00029989479605978546, | |
| "loss": 0.7917, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0002998561176594015, | |
| "loss": 0.7802, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00029981139990469034, | |
| "loss": 0.7826, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00029976064459782297, | |
| "loss": 0.7833, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002997038537842901, | |
| "loss": 0.7782, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0002996410297528193, | |
| "loss": 0.7649, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00029957217503528326, | |
| "loss": 0.7606, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0002994972924065971, | |
| "loss": 0.7673, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00029941638488460725, | |
| "loss": 0.7534, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0002993294557299693, | |
| "loss": 0.7711, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00029923650844601677, | |
| "loss": 0.7499, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00029913754677862004, | |
| "loss": 0.7241, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0002990325747160351, | |
| "loss": 0.7471, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00029892159648874317, | |
| "loss": 0.7424, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00029880461656927996, | |
| "loss": 0.7413, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002986816396720555, | |
| "loss": 0.7299, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000298552670753164, | |
| "loss": 0.7102, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00029841771501018456, | |
| "loss": 0.736, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0002982767778819711, | |
| "loss": 0.716, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00029812986504843366, | |
| "loss": 0.7419, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0002979769824303091, | |
| "loss": 0.714, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00029781813618892303, | |
| "loss": 0.7306, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00029765333272594065, | |
| "loss": 0.7121, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0002974825786831097, | |
| "loss": 0.7107, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00029730588094199214, | |
| "loss": 0.7332, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0002971232466236871, | |
| "loss": 0.7102, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0002969346830885439, | |
| "loss": 0.7205, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00029674019793586516, | |
| "loss": 0.7113, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0002965397990036008, | |
| "loss": 0.7141, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0002963334943680322, | |
| "loss": 0.7039, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0002961212923434465, | |
| "loss": 0.7067, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0002959032014818015, | |
| "loss": 0.7073, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0002956792305723814, | |
| "loss": 0.7113, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00029544938864144225, | |
| "loss": 0.7035, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00029521368495184807, | |
| "loss": 0.6961, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0002949721290026979, | |
| "loss": 0.6884, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0002947247305289429, | |
| "loss": 0.686, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0002944714995009936, | |
| "loss": 0.6998, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00029421244612431877, | |
| "loss": 0.7003, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00029394758083903347, | |
| "loss": 0.6928, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00029367691431947884, | |
| "loss": 0.7097, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0002934004574737915, | |
| "loss": 0.7065, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0002931182214434643, | |
| "loss": 0.6929, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00029283021760289686, | |
| "loss": 0.6902, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00029253645755893777, | |
| "loss": 0.6813, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00029223695315041615, | |
| "loss": 0.682, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0002919317164476651, | |
| "loss": 0.6885, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0002916207597520349, | |
| "loss": 0.6629, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00029130409559539747, | |
| "loss": 0.7018, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0002909817367396412, | |
| "loss": 0.6746, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00029065369617615653, | |
| "loss": 0.699, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00029031998712531273, | |
| "loss": 0.6768, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00028998062303592473, | |
| "loss": 0.682, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00028963561758471135, | |
| "loss": 0.674, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.00028928498467574394, | |
| "loss": 0.6898, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00028892873843988637, | |
| "loss": 0.6947, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.000288566893234225, | |
| "loss": 0.6751, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00028819946364149065, | |
| "loss": 0.6815, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0002878264644694705, | |
| "loss": 0.6678, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0002874479107504114, | |
| "loss": 0.7046, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0002870638177404143, | |
| "loss": 0.6793, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00028667420091881896, | |
| "loss": 0.6718, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0002862790759875807, | |
| "loss": 0.6744, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00028587845887063695, | |
| "loss": 0.6779, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00028547236571326603, | |
| "loss": 0.6737, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00028506081288143617, | |
| "loss": 0.6643, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0002846438169611462, | |
| "loss": 0.678, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00028422139475775673, | |
| "loss": 0.6726, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0002837935632953133, | |
| "loss": 0.6779, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00028336033981586005, | |
| "loss": 0.6765, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00028292174177874487, | |
| "loss": 0.6765, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002824777868599158, | |
| "loss": 0.6804, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002820284929512088, | |
| "loss": 0.6838, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00028157387815962637, | |
| "loss": 0.6774, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00028111396080660815, | |
| "loss": 0.6759, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00028064875942729236, | |
| "loss": 0.6586, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0002801782927697689, | |
| "loss": 0.6711, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.0002797025797943237, | |
| "loss": 0.6582, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0002792216396726747, | |
| "loss": 0.6516, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0002787354917871992, | |
| "loss": 0.6722, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.0002782441557301526, | |
| "loss": 0.6697, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002777476513028789, | |
| "loss": 0.6678, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0002772459985150127, | |
| "loss": 0.6529, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.00027673921758367294, | |
| "loss": 0.669, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00027622732893264776, | |
| "loss": 0.67, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00027571035319157167, | |
| "loss": 0.6703, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0002751883111950942, | |
| "loss": 0.6603, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00027466122398203994, | |
| "loss": 0.6509, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00027412911279456104, | |
| "loss": 0.6677, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002735919990772809, | |
| "loss": 0.6593, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002730499044764299, | |
| "loss": 0.652, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.0002725028508389731, | |
| "loss": 0.658, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00027195086021172994, | |
| "loss": 0.6633, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.0002713939548404853, | |
| "loss": 0.6597, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.0002708321571690937, | |
| "loss": 0.6578, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00027026548983857384, | |
| "loss": 0.6624, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.000269693975686197, | |
| "loss": 0.6546, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0002691176377445662, | |
| "loss": 0.664, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.000268536499240688, | |
| "loss": 0.6626, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00026795058359503675, | |
| "loss": 0.6549, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0002673599144206103, | |
| "loss": 0.6506, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002667645155219785, | |
| "loss": 0.6523, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002661644108943241, | |
| "loss": 0.6721, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00026555962472247537, | |
| "loss": 0.653, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0002649501813799317, | |
| "loss": 0.6623, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00026433610542788116, | |
| "loss": 0.6517, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0002637174216142106, | |
| "loss": 0.6662, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0002630941548725086, | |
| "loss": 0.6713, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0002624663303210602, | |
| "loss": 0.646, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.000261833973261835, | |
| "loss": 0.6539, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0002611971091794672, | |
| "loss": 0.6602, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00026055576374022855, | |
| "loss": 0.6422, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00025990996279099424, | |
| "loss": 0.6511, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00025925973235820096, | |
| "loss": 0.6547, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00025860509864679795, | |
| "loss": 0.6464, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00025794608803919133, | |
| "loss": 0.6591, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00025728272709418044, | |
| "loss": 0.6517, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00025661504254588773, | |
| "loss": 0.6471, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0002559430613026812, | |
| "loss": 0.6601, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00025526681044609004, | |
| "loss": 0.6608, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.0002545863172297133, | |
| "loss": 0.6448, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0002539016090781214, | |
| "loss": 0.6466, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0002532127135857509, | |
| "loss": 0.6367, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00025251965851579245, | |
| "loss": 0.6464, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0002518224717990721, | |
| "loss": 0.6489, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0002511211815329253, | |
| "loss": 0.6455, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00025041581598006475, | |
| "loss": 0.6486, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00024970640356744144, | |
| "loss": 0.6414, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0002489929728850988, | |
| "loss": 0.6455, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00024827555268502075, | |
| "loss": 0.6534, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00024755417187997275, | |
| "loss": 0.6609, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0002468288595423368, | |
| "loss": 0.6415, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00024609964490293954, | |
| "loss": 0.6583, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0002453665573498745, | |
| "loss": 0.6467, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0002446296264273174, | |
| "loss": 0.6433, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00024388888183433577, | |
| "loss": 0.6383, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0002431443534236919, | |
| "loss": 0.636, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00024239607120063995, | |
| "loss": 0.6385, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00024164406532171628, | |
| "loss": 0.6407, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00024088836609352458, | |
| "loss": 0.6344, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00024012900397151418, | |
| "loss": 0.6262, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0002393660095587529, | |
| "loss": 0.6405, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0002385994136046933, | |
| "loss": 0.6487, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0002378292470039341, | |
| "loss": 0.6302, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00023705554079497446, | |
| "loss": 0.6342, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0002362783261589634, | |
| "loss": 0.6386, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.00023549763441844322, | |
| "loss": 0.6415, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00023471349703608696, | |
| "loss": 0.6387, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0002339259456134306, | |
| "loss": 0.6381, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.00023313501188959948, | |
| "loss": 0.6511, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.000232340727740029, | |
| "loss": 0.6413, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00023154312517518024, | |
| "loss": 0.6497, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00023074223633924977, | |
| "loss": 0.6515, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00022993809350887413, | |
| "loss": 0.6363, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00022913072909182936, | |
| "loss": 0.6316, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0002283201756257245, | |
| "loss": 0.633, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00022750646577669083, | |
| "loss": 0.6478, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00022668963233806464, | |
| "loss": 0.6363, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00022586970822906647, | |
| "loss": 0.6303, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0002250467264934738, | |
| "loss": 0.6237, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00022422072029828965, | |
| "loss": 0.6181, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00022339172293240586, | |
| "loss": 0.6164, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00022255976780526145, | |
| "loss": 0.613, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.0002217248884454963, | |
| "loss": 0.6179, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00022088711849959982, | |
| "loss": 0.6066, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0002200464917305549, | |
| "loss": 0.6081, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00021920304201647744, | |
| "loss": 0.6057, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00021835680334925087, | |
| "loss": 0.6165, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0002175078098331562, | |
| "loss": 0.6157, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0002166560956834978, | |
| "loss": 0.6085, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00021580169522522424, | |
| "loss": 0.615, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00021494464289154505, | |
| "loss": 0.6119, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002140849732225431, | |
| "loss": 0.6097, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0002132227208637826, | |
| "loss": 0.6155, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0002123579205649126, | |
| "loss": 0.6066, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00021149060717826694, | |
| "loss": 0.6001, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00021062081565745928, | |
| "loss": 0.6068, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.0002097485810559748, | |
| "loss": 0.6131, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00020887393852575716, | |
| "loss": 0.612, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00020799692331579213, | |
| "loss": 0.6051, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00020711757077068675, | |
| "loss": 0.6145, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00020623591632924515, | |
| "loss": 0.6077, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00020535199552304033, | |
| "loss": 0.6005, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020446584397498178, | |
| "loss": 0.6178, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00020357749739788054, | |
| "loss": 0.6038, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020268699159300927, | |
| "loss": 0.5974, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00020179436244865986, | |
| "loss": 0.6136, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00020089964593869694, | |
| "loss": 0.6098, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00020000287812110793, | |
| "loss": 0.6127, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00019910409513655038, | |
| "loss": 0.6073, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00019820333320689473, | |
| "loss": 0.6008, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00019730062863376524, | |
| "loss": 0.6124, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00019639601779707655, | |
| "loss": 0.6144, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00019548953715356758, | |
| "loss": 0.6123, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0001945812232353326, | |
| "loss": 0.6138, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00019367111264834846, | |
| "loss": 0.6173, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001927592420709998, | |
| "loss": 0.5983, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00019184564825260053, | |
| "loss": 0.6037, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.000190930368011913, | |
| "loss": 0.602, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00019001343823566412, | |
| "loss": 0.6095, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001890948958770587, | |
| "loss": 0.6122, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00018817477795429028, | |
| "loss": 0.61, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00018725312154904925, | |
| "loss": 0.6034, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00018632996380502846, | |
| "loss": 0.6085, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00018540534192642614, | |
| "loss": 0.5977, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00018447929317644672, | |
| "loss": 0.6071, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00018355185487579898, | |
| "loss": 0.6118, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00018262306440119198, | |
| "loss": 0.6078, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00018169295918382883, | |
| "loss": 0.6029, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00018076157670789803, | |
| "loss": 0.5974, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.00017982895450906303, | |
| "loss": 0.6077, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00017889513017294923, | |
| "loss": 0.5953, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00017796014133362946, | |
| "loss": 0.5985, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00017702402567210723, | |
| "loss": 0.5987, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00017608682091479813, | |
| "loss": 0.6017, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.00017514856483200937, | |
| "loss": 0.5985, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.00017420929523641766, | |
| "loss": 0.5999, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.0001732690499815454, | |
| "loss": 0.6052, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00017232786696023492, | |
| "loss": 0.6015, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00017138578410312162, | |
| "loss": 0.6002, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0001704428393771051, | |
| "loss": 0.6011, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00016949907078381927, | |
| "loss": 0.6045, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00016855451635810058, | |
| "loss": 0.607, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.00016760921416645544, | |
| "loss": 0.6062, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.00016666320230552593, | |
| "loss": 0.5966, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.00016571651890055452, | |
| "loss": 0.6026, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0001647692021038477, | |
| "loss": 0.5941, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00016382129009323817, | |
| "loss": 0.5919, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00016287282107054643, | |
| "loss": 0.6035, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00016192383326004106, | |
| "loss": 0.6033, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.00016097436490689838, | |
| "loss": 0.5961, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.00016002445427566107, | |
| "loss": 0.589, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.000159074139648696, | |
| "loss": 0.6, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0001581234593246516, | |
| "loss": 0.5951, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0001571724516169141, | |
| "loss": 0.6027, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00015622115485206385, | |
| "loss": 0.5993, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00015526960736833025, | |
| "loss": 0.5915, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00015431784751404707, | |
| "loss": 0.5977, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00015336591364610686, | |
| "loss": 0.6022, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00015241384412841493, | |
| "loss": 0.6019, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.00015146167733034367, | |
| "loss": 0.5921, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00015050945162518574, | |
| "loss": 0.6011, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0001495572053886079, | |
| "loss": 0.5934, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00014860497699710433, | |
| "loss": 0.5893, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00014765280482645005, | |
| "loss": 0.5951, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00014670072725015437, | |
| "loss": 0.59, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.00014574878263791426, | |
| "loss": 0.6008, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.00014479700935406817, | |
| "loss": 0.5907, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0001438454457560498, | |
| "loss": 0.5977, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.00014289413019284236, | |
| "loss": 0.5999, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.00014194310100343292, | |
| "loss": 0.5979, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.00014099239651526742, | |
| "loss": 0.5889, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0001400420550427061, | |
| "loss": 0.5994, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0001390921148854791, | |
| "loss": 0.5881, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.00013814261432714336, | |
| "loss": 0.5947, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.00013719359163353944, | |
| "loss": 0.5939, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.0001362450850512494, | |
| "loss": 0.5905, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.00013529713280605567, | |
| "loss": 0.5866, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00013434977310140012, | |
| "loss": 0.5942, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0001334030441168447, | |
| "loss": 0.584, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0001324569840065328, | |
| "loss": 0.5932, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.0001315116308976514, | |
| "loss": 0.5896, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.00013056702288889458, | |
| "loss": 0.5899, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.00012962319804892827, | |
| "loss": 0.583, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.00012868019441485568, | |
| "loss": 0.5895, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.00012773804999068473, | |
| "loss": 0.5857, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00012679680274579636, | |
| "loss": 0.5929, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00012585649061341405, | |
| "loss": 0.5899, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.00012491715148907554, | |
| "loss": 0.5856, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.0001239788232291052, | |
| "loss": 0.5907, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00012304154364908856, | |
| "loss": 0.591, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00012210535052234835, | |
| "loss": 0.593, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00012117028157842202, | |
| "loss": 0.5879, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.00012023637450154138, | |
| "loss": 0.5859, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00011930366692911378, | |
| "loss": 0.5866, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00011837219645020536, | |
| "loss": 0.587, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00011744200060402608, | |
| "loss": 0.5844, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.00011651311687841697, | |
| "loss": 0.5848, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.00011558558270833906, | |
| "loss": 0.5925, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011465943547436524, | |
| "loss": 0.5885, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00011373471250117322, | |
| "loss": 0.5904, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0001128114510560416, | |
| "loss": 0.581, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00011188968834734798, | |
| "loss": 0.5822, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00011096946152306923, | |
| "loss": 0.5816, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00011005080766928467, | |
| "loss": 0.5848, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00010913376380868118, | |
| "loss": 0.5783, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00010821836689906128, | |
| "loss": 0.5801, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00010730465383185379, | |
| "loss": 0.5799, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00010639266143062683, | |
| "loss": 0.5884, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00010548242644960404, | |
| "loss": 0.5763, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00010457398557218315, | |
| "loss": 0.5836, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00010366737540945772, | |
| "loss": 0.5814, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00010276263249874166, | |
| "loss": 0.5764, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00010185979330209668, | |
| "loss": 0.5776, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.00010095889420486292, | |
| "loss": 0.5788, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0001000599715141925, | |
| "loss": 0.5809, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 9.916306145758637e-05, | |
| "loss": 0.5828, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.826820018143417e-05, | |
| "loss": 0.5809, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.737542374955779e-05, | |
| "loss": 0.5855, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 9.648476814175755e-05, | |
| "loss": 0.5814, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 9.559626925236263e-05, | |
| "loss": 0.5905, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 9.470996288878409e-05, | |
| "loss": 0.5781, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 9.382588477007196e-05, | |
| "loss": 0.5846, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 9.294407052547586e-05, | |
| "loss": 0.5803, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.206455569300888e-05, | |
| "loss": 0.5849, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.118737571801549e-05, | |
| "loss": 0.5753, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 9.031256595174299e-05, | |
| "loss": 0.5806, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.944016164991682e-05, | |
| "loss": 0.5783, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.857019797131991e-05, | |
| "loss": 0.5853, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 8.77027099763754e-05, | |
| "loss": 0.5874, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.683773262573396e-05, | |
| "loss": 0.5801, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.597530077886474e-05, | |
| "loss": 0.5749, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 8.511544919265039e-05, | |
| "loss": 0.5751, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 8.425821251998646e-05, | |
| "loss": 0.585, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.340362530838499e-05, | |
| "loss": 0.5758, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 8.255172199858192e-05, | |
| "loss": 0.5839, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 8.1702536923149e-05, | |
| "loss": 0.5723, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 8.085610430511064e-05, | |
| "loss": 0.5811, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 8.001245825656439e-05, | |
| "loss": 0.5797, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 7.917163277730609e-05, | |
| "loss": 0.5683, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 7.833366175345985e-05, | |
| "loss": 0.5804, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 7.749857895611223e-05, | |
| "loss": 0.5676, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 7.666641803995134e-05, | |
| "loss": 0.5692, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 7.583721254191065e-05, | |
| "loss": 0.5822, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.50109958798171e-05, | |
| "loss": 0.5727, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.418780135104454e-05, | |
| "loss": 0.5524, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.336766213117173e-05, | |
| "loss": 0.5589, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.255061127264536e-05, | |
| "loss": 0.5523, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.173668170344819e-05, | |
| "loss": 0.5576, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.092590622577162e-05, | |
| "loss": 0.5579, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.011831751469404e-05, | |
| "loss": 0.5557, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.931394811686386e-05, | |
| "loss": 0.5579, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.851283044918787e-05, | |
| "loss": 0.5561, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.771499679752496e-05, | |
| "loss": 0.5587, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.692047931538474e-05, | |
| "loss": 0.557, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.612931002263158e-05, | |
| "loss": 0.5597, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.534152080419484e-05, | |
| "loss": 0.5546, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.455714340878308e-05, | |
| "loss": 0.5592, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.377620944760513e-05, | |
| "loss": 0.5576, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.299875039309576e-05, | |
| "loss": 0.562, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.22247975776475e-05, | |
| "loss": 0.5596, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.1454382192348e-05, | |
| "loss": 0.5537, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.068753528572271e-05, | |
| "loss": 0.5592, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.992428776248381e-05, | |
| "loss": 0.555, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.9164670382284635e-05, | |
| "loss": 0.5572, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.840871375848003e-05, | |
| "loss": 0.5531, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.7656448356892776e-05, | |
| "loss": 0.5583, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.690790449458548e-05, | |
| "loss": 0.5529, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.6163112338638965e-05, | |
| "loss": 0.5566, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.5422101904936474e-05, | |
| "loss": 0.5539, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.468490305695393e-05, | |
| "loss": 0.5507, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.39515455045566e-05, | |
| "loss": 0.557, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.32220588028015e-05, | |
| "loss": 0.5498, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.249647235074647e-05, | |
| "loss": 0.5547, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.177481539026529e-05, | |
| "loss": 0.5542, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.105711700486922e-05, | |
| "loss": 0.5597, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.0343406118534935e-05, | |
| "loss": 0.5583, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.963371149453881e-05, | |
| "loss": 0.5565, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.8928061734297765e-05, | |
| "loss": 0.5534, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 4.822648527621653e-05, | |
| "loss": 0.5549, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.7529010394541746e-05, | |
| "loss": 0.5552, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.6835665198222224e-05, | |
| "loss": 0.5515, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 4.614647762977626e-05, | |
| "loss": 0.5538, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.5461475464165534e-05, | |
| "loss": 0.5562, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.478068630767565e-05, | |
| "loss": 0.5553, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.410413759680383e-05, | |
| "loss": 0.5539, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.343185659715283e-05, | |
| "loss": 0.557, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.276387040233236e-05, | |
| "loss": 0.5542, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.210020593286711e-05, | |
| "loss": 0.5518, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.144088993511181e-05, | |
| "loss": 0.5555, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.078594898017346e-05, | |
| "loss": 0.5565, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.013540946284024e-05, | |
| "loss": 0.5544, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.948929760051797e-05, | |
| "loss": 0.5555, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 3.8847639432173405e-05, | |
| "loss": 0.5563, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 3.821046081728497e-05, | |
| "loss": 0.5525, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 3.757778743480045e-05, | |
| "loss": 0.5541, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 3.694964478210221e-05, | |
| "loss": 0.5564, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 3.63260581739796e-05, | |
| "loss": 0.5551, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.5707052741608636e-05, | |
| "loss": 0.5506, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 3.5092653431539436e-05, | |
| "loss": 0.5483, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.448288500469058e-05, | |
| "loss": 0.556, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.3877772035351326e-05, | |
| "loss": 0.5509, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.327733891019132e-05, | |
| "loss": 0.5574, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 3.268160982727759e-05, | |
| "loss": 0.5525, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 3.209060879509968e-05, | |
| "loss": 0.5502, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 3.150435963160168e-05, | |
| "loss": 0.5523, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 3.0922885963222585e-05, | |
| "loss": 0.5483, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 3.0346211223944077e-05, | |
| "loss": 0.547, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.9774358654346046e-05, | |
| "loss": 0.55, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 2.9207351300670178e-05, | |
| "loss": 0.549, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.864521201389085e-05, | |
| "loss": 0.5536, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 2.8087963448794476e-05, | |
| "loss": 0.546, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 2.7535628063066368e-05, | |
| "loss": 0.551, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.698822811638569e-05, | |
| "loss": 0.5547, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.6445785669528386e-05, | |
| "loss": 0.547, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.590832258347814e-05, | |
| "loss": 0.5556, | |
| "step": 40100 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 2.537586051854522e-05, | |
| "loss": 0.5505, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 2.4848420933493824e-05, | |
| "loss": 0.547, | |
| "step": 40300 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.432602508467691e-05, | |
| "loss": 0.5483, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.3808694025179804e-05, | |
| "loss": 0.5534, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 2.3296448603971657e-05, | |
| "loss": 0.5479, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.2789309465065154e-05, | |
| "loss": 0.5593, | |
| "step": 40700 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.2287297046684737e-05, | |
| "loss": 0.5478, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 2.179043158044263e-05, | |
| "loss": 0.5562, | |
| "step": 40900 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 2.1298733090523722e-05, | |
| "loss": 0.5513, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 2.0812221392878463e-05, | |
| "loss": 0.5507, | |
| "step": 41100 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 2.0330916094424244e-05, | |
| "loss": 0.5465, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 1.985483659225539e-05, | |
| "loss": 0.5504, | |
| "step": 41300 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 1.9384002072861186e-05, | |
| "loss": 0.5444, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 1.89184315113528e-05, | |
| "loss": 0.5514, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 1.8458143670698522e-05, | |
| "loss": 0.5448, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.80031571009676e-05, | |
| "loss": 0.5419, | |
| "step": 41700 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 1.7553490138582786e-05, | |
| "loss": 0.5537, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 1.7109160905580982e-05, | |
| "loss": 0.5493, | |
| "step": 41900 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.6670187308883364e-05, | |
| "loss": 0.5496, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.6236587039573383e-05, | |
| "loss": 0.5491, | |
| "step": 42100 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.5808377572184044e-05, | |
| "loss": 0.5502, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 1.5385576163993417e-05, | |
| "loss": 0.5539, | |
| "step": 42300 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 1.4968199854329322e-05, | |
| "loss": 0.5473, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.4556265463882594e-05, | |
| "loss": 0.5523, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.4149789594029093e-05, | |
| "loss": 0.554, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 1.3748788626160878e-05, | |
| "loss": 0.5487, | |
| "step": 42700 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 1.3353278721025756e-05, | |
| "loss": 0.5558, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 1.2963275818076152e-05, | |
| "loss": 0.5501, | |
| "step": 42900 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 1.2578795634826671e-05, | |
| "loss": 0.5561, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 1.2199853666220678e-05, | |
| "loss": 0.5486, | |
| "step": 43100 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1.18264651840059e-05, | |
| "loss": 0.5514, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 1.145864523611884e-05, | |
| "loss": 0.5537, | |
| "step": 43300 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 1.1096408646078409e-05, | |
| "loss": 0.548, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.073977001238851e-05, | |
| "loss": 0.5509, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.0388743707949648e-05, | |
| "loss": 0.549, | |
| "step": 43600 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 49497, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100, | |
| "total_flos": 3.4286253848387387e+19, | |
| "train_batch_size": 3, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |