| { | |
| "best_metric": 0.36858755350112915, | |
| "best_model_checkpoint": "distilbert-base-multilingual-cased_finetuned_ai4privacy/checkpoint-186561", | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 310935, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.008040265650377089, | |
| "grad_norm": 9.354745864868164, | |
| "learning_rate": 4.0201328251885446e-07, | |
| "loss": 3.2908, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.016080531300754177, | |
| "grad_norm": 2.4111831188201904, | |
| "learning_rate": 8.040265650377089e-07, | |
| "loss": 1.589, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.024120796951131264, | |
| "grad_norm": 9.302245140075684, | |
| "learning_rate": 1.2060398475565633e-06, | |
| "loss": 1.3973, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.032161062601508354, | |
| "grad_norm": 4.2403154373168945, | |
| "learning_rate": 1.6080531300754178e-06, | |
| "loss": 1.3322, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.040201328251885445, | |
| "grad_norm": 10.621835708618164, | |
| "learning_rate": 2.010066412594272e-06, | |
| "loss": 1.2535, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.04824159390226253, | |
| "grad_norm": 15.397185325622559, | |
| "learning_rate": 2.4120796951131266e-06, | |
| "loss": 1.1923, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05628185955263962, | |
| "grad_norm": 5.356049537658691, | |
| "learning_rate": 2.814092977631981e-06, | |
| "loss": 1.1255, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.06432212520301671, | |
| "grad_norm": 12.472413063049316, | |
| "learning_rate": 3.2161062601508357e-06, | |
| "loss": 1.0585, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.0723623908533938, | |
| "grad_norm": 2.5076217651367188, | |
| "learning_rate": 3.61811954266969e-06, | |
| "loss": 1.0072, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.08040265650377089, | |
| "grad_norm": 10.449097633361816, | |
| "learning_rate": 4.020132825188544e-06, | |
| "loss": 1.0171, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.08844292215414798, | |
| "grad_norm": 8.772933006286621, | |
| "learning_rate": 4.422146107707399e-06, | |
| "loss": 0.9755, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.09648318780452506, | |
| "grad_norm": 13.718366622924805, | |
| "learning_rate": 4.824159390226253e-06, | |
| "loss": 0.8844, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.10452345345490215, | |
| "grad_norm": 5.812119483947754, | |
| "learning_rate": 5.226172672745108e-06, | |
| "loss": 0.8733, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.11256371910527924, | |
| "grad_norm": 1.276416540145874, | |
| "learning_rate": 5.628185955263962e-06, | |
| "loss": 0.8734, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.12060398475565633, | |
| "grad_norm": 3.968580961227417, | |
| "learning_rate": 6.030199237782817e-06, | |
| "loss": 0.8541, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.12864425040603342, | |
| "grad_norm": 0.0601482056081295, | |
| "learning_rate": 6.432212520301671e-06, | |
| "loss": 0.8149, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.1366845160564105, | |
| "grad_norm": 9.377354621887207, | |
| "learning_rate": 6.834225802820525e-06, | |
| "loss": 0.8033, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.1447247817067876, | |
| "grad_norm": 7.444766521453857, | |
| "learning_rate": 7.23623908533938e-06, | |
| "loss": 0.7828, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.15276504735716467, | |
| "grad_norm": 7.167333126068115, | |
| "learning_rate": 7.638252367858235e-06, | |
| "loss": 0.7547, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.16080531300754178, | |
| "grad_norm": 7.496470928192139, | |
| "learning_rate": 8.040265650377089e-06, | |
| "loss": 0.7566, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.16884557865791885, | |
| "grad_norm": 12.211381912231445, | |
| "learning_rate": 8.442278932895944e-06, | |
| "loss": 0.7174, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.17688584430829596, | |
| "grad_norm": 6.557511329650879, | |
| "learning_rate": 8.844292215414798e-06, | |
| "loss": 0.698, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.18492610995867304, | |
| "grad_norm": 4.292139053344727, | |
| "learning_rate": 9.246305497933653e-06, | |
| "loss": 0.7122, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.1929663756090501, | |
| "grad_norm": 6.162191390991211, | |
| "learning_rate": 9.648318780452507e-06, | |
| "loss": 0.6993, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.20100664125942722, | |
| "grad_norm": 3.9378864765167236, | |
| "learning_rate": 1.0050332062971362e-05, | |
| "loss": 0.6717, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.2090469069098043, | |
| "grad_norm": 3.709688901901245, | |
| "learning_rate": 1.0452345345490216e-05, | |
| "loss": 0.6516, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.2170871725601814, | |
| "grad_norm": 11.48306655883789, | |
| "learning_rate": 1.0854358628009071e-05, | |
| "loss": 0.6604, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.22512743821055847, | |
| "grad_norm": 5.3870015144348145, | |
| "learning_rate": 1.1256371910527925e-05, | |
| "loss": 0.6391, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.23316770386093558, | |
| "grad_norm": 16.64727210998535, | |
| "learning_rate": 1.165838519304678e-05, | |
| "loss": 0.6573, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.24120796951131265, | |
| "grad_norm": 3.866924524307251, | |
| "learning_rate": 1.2060398475565634e-05, | |
| "loss": 0.6417, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.24924823516168973, | |
| "grad_norm": 1.2701387405395508, | |
| "learning_rate": 1.2462411758084487e-05, | |
| "loss": 0.5996, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.25728850081206683, | |
| "grad_norm": 3.4491727352142334, | |
| "learning_rate": 1.2864425040603343e-05, | |
| "loss": 0.6219, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.2653287664624439, | |
| "grad_norm": 13.956502914428711, | |
| "learning_rate": 1.3266438323122196e-05, | |
| "loss": 0.6224, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.273369032112821, | |
| "grad_norm": 6.554746627807617, | |
| "learning_rate": 1.366845160564105e-05, | |
| "loss": 0.6355, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.2814092977631981, | |
| "grad_norm": 7.591340065002441, | |
| "learning_rate": 1.4070464888159907e-05, | |
| "loss": 0.6195, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.2894495634135752, | |
| "grad_norm": 2.3281121253967285, | |
| "learning_rate": 1.447247817067876e-05, | |
| "loss": 0.5935, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.29748982906395227, | |
| "grad_norm": 1.6392172574996948, | |
| "learning_rate": 1.4874491453197614e-05, | |
| "loss": 0.5783, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.30553009471432935, | |
| "grad_norm": 2.6797776222229004, | |
| "learning_rate": 1.527650473571647e-05, | |
| "loss": 0.5803, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.3135703603647064, | |
| "grad_norm": 4.925036907196045, | |
| "learning_rate": 1.5678518018235323e-05, | |
| "loss": 0.5802, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.32161062601508356, | |
| "grad_norm": 7.5054030418396, | |
| "learning_rate": 1.6080531300754177e-05, | |
| "loss": 0.5807, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.32965089166546063, | |
| "grad_norm": 2.2275919914245605, | |
| "learning_rate": 1.6482544583273034e-05, | |
| "loss": 0.5676, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.3376911573158377, | |
| "grad_norm": 3.462984085083008, | |
| "learning_rate": 1.6884557865791888e-05, | |
| "loss": 0.5433, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.3457314229662148, | |
| "grad_norm": 5.168004035949707, | |
| "learning_rate": 1.728657114831074e-05, | |
| "loss": 0.5799, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.3537716886165919, | |
| "grad_norm": 9.257604598999023, | |
| "learning_rate": 1.7688584430829595e-05, | |
| "loss": 0.5506, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.361811954266969, | |
| "grad_norm": 3.562525510787964, | |
| "learning_rate": 1.8090597713348452e-05, | |
| "loss": 0.5985, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.36985221991734607, | |
| "grad_norm": 6.652304172515869, | |
| "learning_rate": 1.8492610995867306e-05, | |
| "loss": 0.5372, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.37789248556772315, | |
| "grad_norm": 8.097696304321289, | |
| "learning_rate": 1.889462427838616e-05, | |
| "loss": 0.5486, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.3859327512181002, | |
| "grad_norm": 4.553964614868164, | |
| "learning_rate": 1.9296637560905013e-05, | |
| "loss": 0.552, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.39397301686847735, | |
| "grad_norm": 2.6189095973968506, | |
| "learning_rate": 1.969865084342387e-05, | |
| "loss": 0.5565, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.40201328251885443, | |
| "grad_norm": 4.161850929260254, | |
| "learning_rate": 2.0100664125942724e-05, | |
| "loss": 0.5298, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.4100535481692315, | |
| "grad_norm": 2.8359174728393555, | |
| "learning_rate": 2.0502677408461577e-05, | |
| "loss": 0.5577, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.4180938138196086, | |
| "grad_norm": 1.2017419338226318, | |
| "learning_rate": 2.090469069098043e-05, | |
| "loss": 0.5398, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.4261340794699857, | |
| "grad_norm": 6.083527565002441, | |
| "learning_rate": 2.1306703973499288e-05, | |
| "loss": 0.5471, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.4341743451203628, | |
| "grad_norm": 1.8723245859146118, | |
| "learning_rate": 2.1708717256018142e-05, | |
| "loss": 0.4975, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.44221461077073987, | |
| "grad_norm": 2.318981409072876, | |
| "learning_rate": 2.2110730538536996e-05, | |
| "loss": 0.5363, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.45025487642111695, | |
| "grad_norm": 1.1319533586502075, | |
| "learning_rate": 2.251274382105585e-05, | |
| "loss": 0.5189, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.458295142071494, | |
| "grad_norm": 5.3885626792907715, | |
| "learning_rate": 2.2914757103574703e-05, | |
| "loss": 0.5121, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.46633540772187115, | |
| "grad_norm": 4.269131183624268, | |
| "learning_rate": 2.331677038609356e-05, | |
| "loss": 0.5088, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.47437567337224823, | |
| "grad_norm": 7.774847507476807, | |
| "learning_rate": 2.3718783668612414e-05, | |
| "loss": 0.5573, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.4824159390226253, | |
| "grad_norm": 3.4602856636047363, | |
| "learning_rate": 2.4120796951131267e-05, | |
| "loss": 0.5463, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.4904562046730024, | |
| "grad_norm": 1.2964807748794556, | |
| "learning_rate": 2.452281023365012e-05, | |
| "loss": 0.5323, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.49849647032337946, | |
| "grad_norm": 5.7465500831604, | |
| "learning_rate": 2.4924823516168975e-05, | |
| "loss": 0.5289, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.5065367359737566, | |
| "grad_norm": 4.113563537597656, | |
| "learning_rate": 2.532683679868783e-05, | |
| "loss": 0.5177, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.5145770016241337, | |
| "grad_norm": 2.1121294498443604, | |
| "learning_rate": 2.5728850081206685e-05, | |
| "loss": 0.5035, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.5226172672745107, | |
| "grad_norm": 6.446739673614502, | |
| "learning_rate": 2.613086336372554e-05, | |
| "loss": 0.5174, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.5306575329248878, | |
| "grad_norm": 2.8277134895324707, | |
| "learning_rate": 2.6532876646244393e-05, | |
| "loss": 0.5109, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.5386977985752649, | |
| "grad_norm": 5.491194725036621, | |
| "learning_rate": 2.6934889928763246e-05, | |
| "loss": 0.5099, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.546738064225642, | |
| "grad_norm": 4.661314964294434, | |
| "learning_rate": 2.73369032112821e-05, | |
| "loss": 0.528, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.554778329876019, | |
| "grad_norm": 4.397305011749268, | |
| "learning_rate": 2.7738916493800954e-05, | |
| "loss": 0.4957, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.5628185955263962, | |
| "grad_norm": 8.94609260559082, | |
| "learning_rate": 2.8140929776319814e-05, | |
| "loss": 0.5348, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.5708588611767733, | |
| "grad_norm": 1.804666519165039, | |
| "learning_rate": 2.8542943058838668e-05, | |
| "loss": 0.5045, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.5788991268271504, | |
| "grad_norm": 4.391846179962158, | |
| "learning_rate": 2.894495634135752e-05, | |
| "loss": 0.5166, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.5869393924775275, | |
| "grad_norm": 5.855178356170654, | |
| "learning_rate": 2.9346969623876375e-05, | |
| "loss": 0.5196, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.5949796581279045, | |
| "grad_norm": 7.744741916656494, | |
| "learning_rate": 2.974898290639523e-05, | |
| "loss": 0.4924, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.6030199237782816, | |
| "grad_norm": 2.1789309978485107, | |
| "learning_rate": 3.0150996188914082e-05, | |
| "loss": 0.4949, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.6110601894286587, | |
| "grad_norm": 5.0567803382873535, | |
| "learning_rate": 3.055300947143294e-05, | |
| "loss": 0.5164, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.6191004550790358, | |
| "grad_norm": 6.266252517700195, | |
| "learning_rate": 3.095502275395179e-05, | |
| "loss": 0.4989, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.6271407207294128, | |
| "grad_norm": 4.590389251708984, | |
| "learning_rate": 3.135703603647065e-05, | |
| "loss": 0.5296, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.63518098637979, | |
| "grad_norm": 2.7966535091400146, | |
| "learning_rate": 3.1759049318989504e-05, | |
| "loss": 0.4968, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.6432212520301671, | |
| "grad_norm": 7.666275501251221, | |
| "learning_rate": 3.2161062601508354e-05, | |
| "loss": 0.491, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.6512615176805442, | |
| "grad_norm": 10.854148864746094, | |
| "learning_rate": 3.256307588402721e-05, | |
| "loss": 0.4963, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.6593017833309213, | |
| "grad_norm": 8.967985153198242, | |
| "learning_rate": 3.296508916654607e-05, | |
| "loss": 0.5048, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.6673420489812983, | |
| "grad_norm": 2.757068395614624, | |
| "learning_rate": 3.336710244906492e-05, | |
| "loss": 0.4939, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.6753823146316754, | |
| "grad_norm": 8.005558967590332, | |
| "learning_rate": 3.3769115731583775e-05, | |
| "loss": 0.4853, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.6834225802820525, | |
| "grad_norm": 2.626612663269043, | |
| "learning_rate": 3.4171129014102626e-05, | |
| "loss": 0.4924, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.6914628459324296, | |
| "grad_norm": 2.482473611831665, | |
| "learning_rate": 3.457314229662148e-05, | |
| "loss": 0.5245, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.6995031115828066, | |
| "grad_norm": 4.298203468322754, | |
| "learning_rate": 3.497515557914033e-05, | |
| "loss": 0.4821, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.7075433772331838, | |
| "grad_norm": 3.765261650085449, | |
| "learning_rate": 3.537716886165919e-05, | |
| "loss": 0.4826, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.7155836428835609, | |
| "grad_norm": 4.702629566192627, | |
| "learning_rate": 3.577918214417805e-05, | |
| "loss": 0.5011, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.723623908533938, | |
| "grad_norm": 6.423455238342285, | |
| "learning_rate": 3.6181195426696904e-05, | |
| "loss": 0.4706, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.7316641741843151, | |
| "grad_norm": 2.120485782623291, | |
| "learning_rate": 3.6583208709215755e-05, | |
| "loss": 0.4811, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.7397044398346921, | |
| "grad_norm": 3.435938835144043, | |
| "learning_rate": 3.698522199173461e-05, | |
| "loss": 0.4986, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.7477447054850692, | |
| "grad_norm": 6.911807060241699, | |
| "learning_rate": 3.738723527425346e-05, | |
| "loss": 0.4506, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.7557849711354463, | |
| "grad_norm": 4.446883678436279, | |
| "learning_rate": 3.778924855677232e-05, | |
| "loss": 0.4855, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.7638252367858234, | |
| "grad_norm": 2.1728882789611816, | |
| "learning_rate": 3.819126183929117e-05, | |
| "loss": 0.4725, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.7718655024362004, | |
| "grad_norm": 2.65541672706604, | |
| "learning_rate": 3.8593275121810026e-05, | |
| "loss": 0.4665, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.7799057680865776, | |
| "grad_norm": 5.510695457458496, | |
| "learning_rate": 3.899528840432888e-05, | |
| "loss": 0.4569, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.7879460337369547, | |
| "grad_norm": 1.6208312511444092, | |
| "learning_rate": 3.939730168684774e-05, | |
| "loss": 0.4853, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.7959862993873318, | |
| "grad_norm": 2.4537601470947266, | |
| "learning_rate": 3.979931496936659e-05, | |
| "loss": 0.4787, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.8040265650377089, | |
| "grad_norm": 4.070446491241455, | |
| "learning_rate": 4.020132825188545e-05, | |
| "loss": 0.4713, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.8120668306880859, | |
| "grad_norm": 3.117581605911255, | |
| "learning_rate": 4.06033415344043e-05, | |
| "loss": 0.4941, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.820107096338463, | |
| "grad_norm": 2.988654851913452, | |
| "learning_rate": 4.1005354816923155e-05, | |
| "loss": 0.5044, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.8281473619888401, | |
| "grad_norm": 1.2906758785247803, | |
| "learning_rate": 4.1407368099442005e-05, | |
| "loss": 0.4762, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.8361876276392172, | |
| "grad_norm": 1.6807843446731567, | |
| "learning_rate": 4.180938138196086e-05, | |
| "loss": 0.4496, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.8442278932895942, | |
| "grad_norm": 5.928719997406006, | |
| "learning_rate": 4.221139466447971e-05, | |
| "loss": 0.4724, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.8522681589399714, | |
| "grad_norm": 5.874762535095215, | |
| "learning_rate": 4.2613407946998576e-05, | |
| "loss": 0.4756, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.8603084245903485, | |
| "grad_norm": 2.588416814804077, | |
| "learning_rate": 4.301542122951743e-05, | |
| "loss": 0.4774, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.8683486902407256, | |
| "grad_norm": 4.079570770263672, | |
| "learning_rate": 4.3417434512036284e-05, | |
| "loss": 0.4844, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.8763889558911027, | |
| "grad_norm": 3.7888760566711426, | |
| "learning_rate": 4.3819447794555134e-05, | |
| "loss": 0.479, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.8844292215414797, | |
| "grad_norm": 1.7335712909698486, | |
| "learning_rate": 4.422146107707399e-05, | |
| "loss": 0.4803, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.8924694871918568, | |
| "grad_norm": 3.9193661212921143, | |
| "learning_rate": 4.462347435959284e-05, | |
| "loss": 0.474, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.9005097528422339, | |
| "grad_norm": 2.637026071548462, | |
| "learning_rate": 4.50254876421117e-05, | |
| "loss": 0.4741, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.908550018492611, | |
| "grad_norm": 6.095594882965088, | |
| "learning_rate": 4.542750092463055e-05, | |
| "loss": 0.5069, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.916590284142988, | |
| "grad_norm": 2.3618390560150146, | |
| "learning_rate": 4.5829514207149406e-05, | |
| "loss": 0.4665, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.9246305497933651, | |
| "grad_norm": 4.125543594360352, | |
| "learning_rate": 4.6231527489668256e-05, | |
| "loss": 0.4592, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.9326708154437423, | |
| "grad_norm": 1.539061427116394, | |
| "learning_rate": 4.663354077218712e-05, | |
| "loss": 0.4854, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.9407110810941194, | |
| "grad_norm": 3.0301315784454346, | |
| "learning_rate": 4.703555405470597e-05, | |
| "loss": 0.4852, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.9487513467444965, | |
| "grad_norm": 6.304039478302002, | |
| "learning_rate": 4.743756733722483e-05, | |
| "loss": 0.4858, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.9567916123948735, | |
| "grad_norm": 4.986495494842529, | |
| "learning_rate": 4.783958061974368e-05, | |
| "loss": 0.4776, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.9648318780452506, | |
| "grad_norm": 5.5548577308654785, | |
| "learning_rate": 4.8241593902262534e-05, | |
| "loss": 0.4822, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.9728721436956277, | |
| "grad_norm": 15.371631622314453, | |
| "learning_rate": 4.8643607184781385e-05, | |
| "loss": 0.4807, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.9809124093460048, | |
| "grad_norm": 2.943363666534424, | |
| "learning_rate": 4.904562046730024e-05, | |
| "loss": 0.471, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.9889526749963818, | |
| "grad_norm": 3.591721773147583, | |
| "learning_rate": 4.944763374981909e-05, | |
| "loss": 0.44, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.9969929406467589, | |
| "grad_norm": 6.311567306518555, | |
| "learning_rate": 4.984964703233795e-05, | |
| "loss": 0.4774, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_BOD_f1": 0.13530861661702784, | |
| "eval_BUILDING_f1": 0.1842277344260866, | |
| "eval_CARDISSUER_f1": 0.0, | |
| "eval_CITY_f1": 0.125498426023085, | |
| "eval_COUNTRY_f1": 0.23373648524830493, | |
| "eval_DATE_f1": 0.1185302252074279, | |
| "eval_DRIVERLICENSE_f1": 0.09358226862278726, | |
| "eval_EMAIL_f1": 0.12614760939549305, | |
| "eval_GEOCOORD_f1": 0.04995836802664446, | |
| "eval_GIVENNAME1_f1": 0.08928831320040749, | |
| "eval_GIVENNAME2_f1": 0.050629331070569925, | |
| "eval_IDCARD_f1": 0.1040870332255219, | |
| "eval_IP_f1": 0.11216258518143948, | |
| "eval_LASTNAME1_f1": 0.12413162705667276, | |
| "eval_LASTNAME2_f1": 0.046269820639459315, | |
| "eval_LASTNAME3_f1": 0.001999000499750125, | |
| "eval_PASSPORT_f1": 0.10804802134281903, | |
| "eval_PASS_f1": 0.0486428379467885, | |
| "eval_POSTCODE_f1": 0.17260692464358451, | |
| "eval_SECADDRESS_f1": 0.15403976843694941, | |
| "eval_SEX_f1": 0.20438480891338204, | |
| "eval_SOCIALNUMBER_f1": 0.08858949922023913, | |
| "eval_STATE_f1": 0.15877558099202219, | |
| "eval_STREET_f1": 0.12391348883490882, | |
| "eval_TEL_f1": 0.14064037982745642, | |
| "eval_TIME_f1": 0.16669289745261598, | |
| "eval_TITLE_f1": 0.15825641442686944, | |
| "eval_USERNAME_f1": 0.13859120136593164, | |
| "eval_loss": 0.46106186509132385, | |
| "eval_overall_accuracy": 0.8379911879866254, | |
| "eval_overall_f1": 0.12906237540150123, | |
| "eval_overall_precision": 0.17644151174640363, | |
| "eval_overall_recall": 0.10174200110497884, | |
| "eval_runtime": 989.0037, | |
| "eval_samples_per_second": 53.897, | |
| "eval_steps_per_second": 26.948, | |
| "step": 62187 | |
| }, | |
| { | |
| "epoch": 1.005033206297136, | |
| "grad_norm": 1.4062920808792114, | |
| "learning_rate": 4.999980466562704e-05, | |
| "loss": 0.4466, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.0130734719475132, | |
| "grad_norm": 6.279462814331055, | |
| "learning_rate": 4.9998682144362336e-05, | |
| "loss": 0.4643, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.0211137375978903, | |
| "grad_norm": 4.784337043762207, | |
| "learning_rate": 4.999656275836824e-05, | |
| "loss": 0.4599, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.0291540032482673, | |
| "grad_norm": 2.8538565635681152, | |
| "learning_rate": 4.999344659215886e-05, | |
| "loss": 0.4423, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.0371942688986444, | |
| "grad_norm": 2.8800506591796875, | |
| "learning_rate": 4.9989333769996584e-05, | |
| "loss": 0.4526, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.0452345345490215, | |
| "grad_norm": 24.831398010253906, | |
| "learning_rate": 4.998422445588715e-05, | |
| "loss": 0.4537, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.0532748001993986, | |
| "grad_norm": 4.20318078994751, | |
| "learning_rate": 4.997811885357311e-05, | |
| "loss": 0.4654, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.0613150658497756, | |
| "grad_norm": 2.9428658485412598, | |
| "learning_rate": 4.9971017206525673e-05, | |
| "loss": 0.4417, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.0693553315001527, | |
| "grad_norm": 2.4196345806121826, | |
| "learning_rate": 4.996291979793503e-05, | |
| "loss": 0.4563, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.0773955971505298, | |
| "grad_norm": 7.266270637512207, | |
| "learning_rate": 4.995382695069903e-05, | |
| "loss": 0.4543, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.0854358628009069, | |
| "grad_norm": 5.247376918792725, | |
| "learning_rate": 4.994373902741036e-05, | |
| "loss": 0.4647, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.093476128451284, | |
| "grad_norm": 5.3602495193481445, | |
| "learning_rate": 4.9932656430342004e-05, | |
| "loss": 0.4432, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.101516394101661, | |
| "grad_norm": 2.3903250694274902, | |
| "learning_rate": 4.992057960143126e-05, | |
| "loss": 0.4565, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.1095566597520383, | |
| "grad_norm": 1.1927706003189087, | |
| "learning_rate": 4.990750902226213e-05, | |
| "loss": 0.4285, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.1175969254024154, | |
| "grad_norm": 4.542496204376221, | |
| "learning_rate": 4.9893445214046034e-05, | |
| "loss": 0.4548, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.1256371910527925, | |
| "grad_norm": 0.36813923716545105, | |
| "learning_rate": 4.9878388737601136e-05, | |
| "loss": 0.4806, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.1336774567031696, | |
| "grad_norm": 15.352900505065918, | |
| "learning_rate": 4.986234019332988e-05, | |
| "loss": 0.4328, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.1417177223535466, | |
| "grad_norm": 3.338168144226074, | |
| "learning_rate": 4.9845300221195114e-05, | |
| "loss": 0.4483, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.1497579880039237, | |
| "grad_norm": 4.263662815093994, | |
| "learning_rate": 4.982726950069455e-05, | |
| "loss": 0.4812, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.1577982536543008, | |
| "grad_norm": 5.22310733795166, | |
| "learning_rate": 4.980824875083367e-05, | |
| "loss": 0.4573, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.1658385193046779, | |
| "grad_norm": 4.639328479766846, | |
| "learning_rate": 4.9788238730097016e-05, | |
| "loss": 0.447, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.173878784955055, | |
| "grad_norm": 4.110581398010254, | |
| "learning_rate": 4.976724023641803e-05, | |
| "loss": 0.4395, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.181919050605432, | |
| "grad_norm": 3.4644949436187744, | |
| "learning_rate": 4.974525410714717e-05, | |
| "loss": 0.4568, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.189959316255809, | |
| "grad_norm": 5.24867057800293, | |
| "learning_rate": 4.972228121901849e-05, | |
| "loss": 0.4568, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.1979995819061862, | |
| "grad_norm": 6.682961940765381, | |
| "learning_rate": 4.96983224881148e-05, | |
| "loss": 0.4604, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.2060398475565632, | |
| "grad_norm": 2.548774242401123, | |
| "learning_rate": 4.967337886983098e-05, | |
| "loss": 0.4376, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.2140801132069403, | |
| "grad_norm": 6.16270637512207, | |
| "learning_rate": 4.964745135883601e-05, | |
| "loss": 0.4262, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.2221203788573174, | |
| "grad_norm": 3.1907684803009033, | |
| "learning_rate": 4.9620540989033237e-05, | |
| "loss": 0.4643, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.2301606445076945, | |
| "grad_norm": 0.550860583782196, | |
| "learning_rate": 4.9592648833519175e-05, | |
| "loss": 0.4218, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.2382009101580715, | |
| "grad_norm": 0.518520712852478, | |
| "learning_rate": 4.956377600454068e-05, | |
| "loss": 0.4338, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.2462411758084486, | |
| "grad_norm": 2.4898388385772705, | |
| "learning_rate": 4.953392365345066e-05, | |
| "loss": 0.4422, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.2542814414588257, | |
| "grad_norm": 2.029384136199951, | |
| "learning_rate": 4.950309297066209e-05, | |
| "loss": 0.4397, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.2623217071092028, | |
| "grad_norm": 3.3983354568481445, | |
| "learning_rate": 4.94712851856006e-05, | |
| "loss": 0.4624, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.2703619727595798, | |
| "grad_norm": 3.732306718826294, | |
| "learning_rate": 4.943850156665539e-05, | |
| "loss": 0.4338, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.2784022384099571, | |
| "grad_norm": 4.838616371154785, | |
| "learning_rate": 4.940474342112874e-05, | |
| "loss": 0.4407, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.2864425040603342, | |
| "grad_norm": 5.9939374923706055, | |
| "learning_rate": 4.937001209518378e-05, | |
| "loss": 0.4281, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.2944827697107113, | |
| "grad_norm": 2.3598451614379883, | |
| "learning_rate": 4.9334308973790874e-05, | |
| "loss": 0.4491, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.3025230353610884, | |
| "grad_norm": 3.637604236602783, | |
| "learning_rate": 4.929763548067237e-05, | |
| "loss": 0.4191, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.3105633010114655, | |
| "grad_norm": 6.502430438995361, | |
| "learning_rate": 4.925999307824583e-05, | |
| "loss": 0.4084, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.3186035666618425, | |
| "grad_norm": 2.610710620880127, | |
| "learning_rate": 4.9221383267565703e-05, | |
| "loss": 0.4374, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.3266438323122196, | |
| "grad_norm": 4.589448928833008, | |
| "learning_rate": 4.9181807588263504e-05, | |
| "loss": 0.4567, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.3346840979625967, | |
| "grad_norm": 1.6119569540023804, | |
| "learning_rate": 4.914126761848638e-05, | |
| "loss": 0.4339, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.3427243636129738, | |
| "grad_norm": 3.726769208908081, | |
| "learning_rate": 4.9099764974834175e-05, | |
| "loss": 0.4199, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.3507646292633508, | |
| "grad_norm": 1.6461119651794434, | |
| "learning_rate": 4.9057301312295e-05, | |
| "loss": 0.4553, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.358804894913728, | |
| "grad_norm": 3.4956886768341064, | |
| "learning_rate": 4.9013878324179216e-05, | |
| "loss": 0.4426, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.366845160564105, | |
| "grad_norm": 2.0839805603027344, | |
| "learning_rate": 4.8969497742051925e-05, | |
| "loss": 0.4175, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.374885426214482, | |
| "grad_norm": 0.759906530380249, | |
| "learning_rate": 4.892416133566388e-05, | |
| "loss": 0.441, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.3829256918648591, | |
| "grad_norm": 0.981677770614624, | |
| "learning_rate": 4.887787091288096e-05, | |
| "loss": 0.4344, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.3909659575152364, | |
| "grad_norm": 5.431253910064697, | |
| "learning_rate": 4.883062831961208e-05, | |
| "loss": 0.4569, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.3990062231656135, | |
| "grad_norm": 3.129939317703247, | |
| "learning_rate": 4.878243543973553e-05, | |
| "loss": 0.4513, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.4070464888159906, | |
| "grad_norm": 3.264197826385498, | |
| "learning_rate": 4.87332941950239e-05, | |
| "loss": 0.4522, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.4150867544663677, | |
| "grad_norm": 7.076359272003174, | |
| "learning_rate": 4.868320654506745e-05, | |
| "loss": 0.4458, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.4231270201167447, | |
| "grad_norm": 5.3585028648376465, | |
| "learning_rate": 4.8632174487195916e-05, | |
| "loss": 0.4304, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.4311672857671218, | |
| "grad_norm": 2.1781795024871826, | |
| "learning_rate": 4.858020005639891e-05, | |
| "loss": 0.4244, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.439207551417499, | |
| "grad_norm": 1.4065402746200562, | |
| "learning_rate": 4.852728532524477e-05, | |
| "loss": 0.434, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.447247817067876, | |
| "grad_norm": 2.884308099746704, | |
| "learning_rate": 4.847343240379791e-05, | |
| "loss": 0.4242, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.455288082718253, | |
| "grad_norm": 5.535454273223877, | |
| "learning_rate": 4.8418643439534614e-05, | |
| "loss": 0.4404, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.4633283483686301, | |
| "grad_norm": 7.863104343414307, | |
| "learning_rate": 4.8362920617257526e-05, | |
| "loss": 0.4415, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.4713686140190072, | |
| "grad_norm": 2.946786403656006, | |
| "learning_rate": 4.830626615900841e-05, | |
| "loss": 0.4481, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.4794088796693843, | |
| "grad_norm": 0.8720025420188904, | |
| "learning_rate": 4.82486823239796e-05, | |
| "loss": 0.4508, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.4874491453197614, | |
| "grad_norm": 1.1726230382919312, | |
| "learning_rate": 4.819017140842392e-05, | |
| "loss": 0.4291, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.4954894109701384, | |
| "grad_norm": 11.948071479797363, | |
| "learning_rate": 4.8130735745563075e-05, | |
| "loss": 0.4347, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.5035296766205155, | |
| "grad_norm": 1.354583740234375, | |
| "learning_rate": 4.807037770549464e-05, | |
| "loss": 0.4323, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.5115699422708926, | |
| "grad_norm": 3.4977433681488037, | |
| "learning_rate": 4.800909969509755e-05, | |
| "loss": 0.4226, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.5196102079212697, | |
| "grad_norm": 7.07570743560791, | |
| "learning_rate": 4.79469041579361e-05, | |
| "loss": 0.4616, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.5276504735716467, | |
| "grad_norm": 7.3226118087768555, | |
| "learning_rate": 4.7883793574162535e-05, | |
| "loss": 0.4179, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.5356907392220238, | |
| "grad_norm": 5.095849990844727, | |
| "learning_rate": 4.78197704604181e-05, | |
| "loss": 0.4347, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.543731004872401, | |
| "grad_norm": 3.7908096313476562, | |
| "learning_rate": 4.775483736973275e-05, | |
| "loss": 0.4459, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.551771270522778, | |
| "grad_norm": 6.59714937210083, | |
| "learning_rate": 4.7688996891423274e-05, | |
| "loss": 0.4255, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.559811536173155, | |
| "grad_norm": 3.0935490131378174, | |
| "learning_rate": 4.7622251650990125e-05, | |
| "loss": 0.4392, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.5678518018235321, | |
| "grad_norm": 1.1097116470336914, | |
| "learning_rate": 4.755460431001263e-05, | |
| "loss": 0.427, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.5758920674739092, | |
| "grad_norm": 3.3510866165161133, | |
| "learning_rate": 4.748605756604291e-05, | |
| "loss": 0.4465, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.5839323331242863, | |
| "grad_norm": 2.419706106185913, | |
| "learning_rate": 4.741661415249833e-05, | |
| "loss": 0.4277, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.5919725987746633, | |
| "grad_norm": 2.6924312114715576, | |
| "learning_rate": 4.734627683855244e-05, | |
| "loss": 0.4149, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.6000128644250406, | |
| "grad_norm": 1.3521323204040527, | |
| "learning_rate": 4.727504842902459e-05, | |
| "loss": 0.4424, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.6080531300754177, | |
| "grad_norm": 3.8188209533691406, | |
| "learning_rate": 4.720293176426808e-05, | |
| "loss": 0.4276, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.6160933957257948, | |
| "grad_norm": 3.319603204727173, | |
| "learning_rate": 4.712992972005688e-05, | |
| "loss": 0.437, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.6241336613761719, | |
| "grad_norm": 2.5236575603485107, | |
| "learning_rate": 4.7056045207470964e-05, | |
| "loss": 0.4445, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.632173927026549, | |
| "grad_norm": 1.3127907514572144, | |
| "learning_rate": 4.6981281172780245e-05, | |
| "loss": 0.4293, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.640214192676926, | |
| "grad_norm": 1.583178162574768, | |
| "learning_rate": 4.690564059732704e-05, | |
| "loss": 0.4409, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.648254458327303, | |
| "grad_norm": 3.5322415828704834, | |
| "learning_rate": 4.682912649740721e-05, | |
| "loss": 0.4109, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.6562947239776802, | |
| "grad_norm": 1.7454696893692017, | |
| "learning_rate": 4.6751741924149904e-05, | |
| "loss": 0.3912, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.6643349896280573, | |
| "grad_norm": 7.049033164978027, | |
| "learning_rate": 4.667348996339587e-05, | |
| "loss": 0.4189, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.6723752552784346, | |
| "grad_norm": 8.582345008850098, | |
| "learning_rate": 4.659437373557436e-05, | |
| "loss": 0.4216, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.6804155209288116, | |
| "grad_norm": 8.063545227050781, | |
| "learning_rate": 4.6514396395578785e-05, | |
| "loss": 0.4275, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.6884557865791887, | |
| "grad_norm": 3.7764339447021484, | |
| "learning_rate": 4.643356113264082e-05, | |
| "loss": 0.4392, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.6964960522295658, | |
| "grad_norm": 3.022294759750366, | |
| "learning_rate": 4.635187117020331e-05, | |
| "loss": 0.4451, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.7045363178799429, | |
| "grad_norm": 4.214914798736572, | |
| "learning_rate": 4.6269329765791646e-05, | |
| "loss": 0.4314, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.71257658353032, | |
| "grad_norm": 1.1796516180038452, | |
| "learning_rate": 4.6185940210883927e-05, | |
| "loss": 0.4115, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.720616849180697, | |
| "grad_norm": 1.9139032363891602, | |
| "learning_rate": 4.610170583077969e-05, | |
| "loss": 0.4338, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.728657114831074, | |
| "grad_norm": 2.1567490100860596, | |
| "learning_rate": 4.601662998446732e-05, | |
| "loss": 0.3892, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.7366973804814512, | |
| "grad_norm": 2.5963244438171387, | |
| "learning_rate": 4.5930716064490054e-05, | |
| "loss": 0.4195, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.7447376461318282, | |
| "grad_norm": 10.889185905456543, | |
| "learning_rate": 4.584396749681075e-05, | |
| "loss": 0.4195, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.7527779117822053, | |
| "grad_norm": 1.634473204612732, | |
| "learning_rate": 4.575638774067527e-05, | |
| "loss": 0.4055, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.7608181774325824, | |
| "grad_norm": 3.4343137741088867, | |
| "learning_rate": 4.5667980288474494e-05, | |
| "loss": 0.4313, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.7688584430829595, | |
| "grad_norm": 3.3218469619750977, | |
| "learning_rate": 4.5578748665605106e-05, | |
| "loss": 0.4313, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.7768987087333366, | |
| "grad_norm": 0.7914499044418335, | |
| "learning_rate": 4.548869643032899e-05, | |
| "loss": 0.4358, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.7849389743837136, | |
| "grad_norm": 9.380616188049316, | |
| "learning_rate": 4.5397827173631313e-05, | |
| "loss": 0.4278, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.7929792400340907, | |
| "grad_norm": 1.358391284942627, | |
| "learning_rate": 4.530614451907737e-05, | |
| "loss": 0.4007, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.8010195056844678, | |
| "grad_norm": 5.607223987579346, | |
| "learning_rate": 4.521365212266807e-05, | |
| "loss": 0.4083, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.8090597713348449, | |
| "grad_norm": 1.3673557043075562, | |
| "learning_rate": 4.512035367269416e-05, | |
| "loss": 0.4242, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.817100036985222, | |
| "grad_norm": 11.210189819335938, | |
| "learning_rate": 4.5026252889589104e-05, | |
| "loss": 0.4294, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.825140302635599, | |
| "grad_norm": 2.304598808288574, | |
| "learning_rate": 4.493135352578079e-05, | |
| "loss": 0.4239, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.833180568285976, | |
| "grad_norm": 1.7466423511505127, | |
| "learning_rate": 4.4835659365541835e-05, | |
| "loss": 0.4374, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.8412208339363532, | |
| "grad_norm": 2.3758962154388428, | |
| "learning_rate": 4.4739174224838725e-05, | |
| "loss": 0.4061, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.8492610995867302, | |
| "grad_norm": 2.8065474033355713, | |
| "learning_rate": 4.464190195117963e-05, | |
| "loss": 0.42, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.8573013652371073, | |
| "grad_norm": 2.404449462890625, | |
| "learning_rate": 4.454384642346098e-05, | |
| "loss": 0.4309, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.8653416308874844, | |
| "grad_norm": 2.1420180797576904, | |
| "learning_rate": 4.444501155181277e-05, | |
| "loss": 0.3996, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.8733818965378615, | |
| "grad_norm": 7.1883745193481445, | |
| "learning_rate": 4.434540127744265e-05, | |
| "loss": 0.4048, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.8814221621882385, | |
| "grad_norm": 1.8959953784942627, | |
| "learning_rate": 4.42450195724788e-05, | |
| "loss": 0.443, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.8894624278386158, | |
| "grad_norm": 3.61218523979187, | |
| "learning_rate": 4.414387043981144e-05, | |
| "loss": 0.4338, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.897502693488993, | |
| "grad_norm": 2.614368200302124, | |
| "learning_rate": 4.404195791293335e-05, | |
| "loss": 0.4123, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.90554295913937, | |
| "grad_norm": 3.1265552043914795, | |
| "learning_rate": 4.3939286055778864e-05, | |
| "loss": 0.409, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.913583224789747, | |
| "grad_norm": 3.8968493938446045, | |
| "learning_rate": 4.3835858962561936e-05, | |
| "loss": 0.4083, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.9216234904401241, | |
| "grad_norm": 5.376585483551025, | |
| "learning_rate": 4.3731680757612836e-05, | |
| "loss": 0.4065, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.9296637560905012, | |
| "grad_norm": 0.7656725645065308, | |
| "learning_rate": 4.3626755595213675e-05, | |
| "loss": 0.4122, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.9377040217408783, | |
| "grad_norm": 2.6415796279907227, | |
| "learning_rate": 4.3521087659432744e-05, | |
| "loss": 0.395, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.9457442873912554, | |
| "grad_norm": 2.8520469665527344, | |
| "learning_rate": 4.341468116395772e-05, | |
| "loss": 0.4197, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.9537845530416325, | |
| "grad_norm": 4.406233787536621, | |
| "learning_rate": 4.330754035192755e-05, | |
| "loss": 0.3891, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.9618248186920095, | |
| "grad_norm": 7.733519554138184, | |
| "learning_rate": 4.319966949576331e-05, | |
| "loss": 0.4147, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.9698650843423868, | |
| "grad_norm": 3.829367160797119, | |
| "learning_rate": 4.309107289699783e-05, | |
| "loss": 0.4152, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.977905349992764, | |
| "grad_norm": 4.29428243637085, | |
| "learning_rate": 4.298175488610416e-05, | |
| "loss": 0.3881, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.985945615643141, | |
| "grad_norm": 5.015018939971924, | |
| "learning_rate": 4.287171982232287e-05, | |
| "loss": 0.4119, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.993985881293518, | |
| "grad_norm": 2.6516435146331787, | |
| "learning_rate": 4.276097209348821e-05, | |
| "loss": 0.4205, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_BOD_f1": 0.1831212593672987, | |
| "eval_BUILDING_f1": 0.2705975464978235, | |
| "eval_CARDISSUER_f1": 0.0, | |
| "eval_CITY_f1": 0.19231476374333514, | |
| "eval_COUNTRY_f1": 0.2818905232698336, | |
| "eval_DATE_f1": 0.18212234932404936, | |
| "eval_DRIVERLICENSE_f1": 0.1520774243834845, | |
| "eval_EMAIL_f1": 0.18632840575768758, | |
| "eval_GEOCOORD_f1": 0.1197289156626506, | |
| "eval_GIVENNAME1_f1": 0.09968051118210862, | |
| "eval_GIVENNAME2_f1": 0.06621895986256442, | |
| "eval_IDCARD_f1": 0.14727730497929628, | |
| "eval_IP_f1": 0.15122900598284436, | |
| "eval_LASTNAME1_f1": 0.14426433637960553, | |
| "eval_LASTNAME2_f1": 0.09549795361527968, | |
| "eval_LASTNAME3_f1": 0.05270362765229296, | |
| "eval_PASSPORT_f1": 0.1997254632807138, | |
| "eval_PASS_f1": 0.16784006376195537, | |
| "eval_POSTCODE_f1": 0.24691647150663545, | |
| "eval_SECADDRESS_f1": 0.20656975343345915, | |
| "eval_SEX_f1": 0.2640652068955724, | |
| "eval_SOCIALNUMBER_f1": 0.18269502401226845, | |
| "eval_STATE_f1": 0.2265575333757152, | |
| "eval_STREET_f1": 0.16019196306533426, | |
| "eval_TEL_f1": 0.18788235294117647, | |
| "eval_TIME_f1": 0.23721180067762995, | |
| "eval_TITLE_f1": 0.22020825298881605, | |
| "eval_USERNAME_f1": 0.20686748815805, | |
| "eval_loss": 0.4272100329399109, | |
| "eval_overall_accuracy": 0.8541924711412506, | |
| "eval_overall_f1": 0.188736255984349, | |
| "eval_overall_precision": 0.23724229660829085, | |
| "eval_overall_recall": 0.15669810455130245, | |
| "eval_runtime": 651.1261, | |
| "eval_samples_per_second": 81.864, | |
| "eval_steps_per_second": 40.932, | |
| "step": 124374 | |
| }, | |
| { | |
| "epoch": 2.002026146943895, | |
| "grad_norm": 4.718987464904785, | |
| "learning_rate": 4.264951611585317e-05, | |
| "loss": 0.3819, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 2.010066412594272, | |
| "grad_norm": 3.464489459991455, | |
| "learning_rate": 4.253735633391336e-05, | |
| "loss": 0.3827, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 2.0181066782446493, | |
| "grad_norm": 4.967048645019531, | |
| "learning_rate": 4.2424497220229785e-05, | |
| "loss": 0.3857, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 2.0261469438950264, | |
| "grad_norm": 1.6211844682693481, | |
| "learning_rate": 4.2310943275250464e-05, | |
| "loss": 0.3925, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 2.0341872095454034, | |
| "grad_norm": 14.46766185760498, | |
| "learning_rate": 4.219669902713101e-05, | |
| "loss": 0.3816, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 2.0422274751957805, | |
| "grad_norm": 3.7520477771759033, | |
| "learning_rate": 4.2081769031554006e-05, | |
| "loss": 0.3949, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 2.0502677408461576, | |
| "grad_norm": 2.188344717025757, | |
| "learning_rate": 4.1966157871547437e-05, | |
| "loss": 0.3788, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 2.0583080064965347, | |
| "grad_norm": 5.201277256011963, | |
| "learning_rate": 4.184987015730181e-05, | |
| "loss": 0.3833, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 2.0663482721469117, | |
| "grad_norm": 1.3161330223083496, | |
| "learning_rate": 4.1732910525986415e-05, | |
| "loss": 0.3734, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 2.074388537797289, | |
| "grad_norm": 2.9594569206237793, | |
| "learning_rate": 4.161528364156434e-05, | |
| "loss": 0.3745, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 2.082428803447666, | |
| "grad_norm": 5.857306480407715, | |
| "learning_rate": 4.149699419460656e-05, | |
| "loss": 0.4, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 2.090469069098043, | |
| "grad_norm": 1.9004195928573608, | |
| "learning_rate": 4.1378046902104815e-05, | |
| "loss": 0.3832, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 2.09850933474842, | |
| "grad_norm": 1.1724590063095093, | |
| "learning_rate": 4.1258446507283564e-05, | |
| "loss": 0.3715, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 2.106549600398797, | |
| "grad_norm": 3.199882745742798, | |
| "learning_rate": 4.113819777941083e-05, | |
| "loss": 0.3647, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 2.114589866049174, | |
| "grad_norm": 1.9746817350387573, | |
| "learning_rate": 4.1017305513608e-05, | |
| "loss": 0.3894, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 2.1226301316995513, | |
| "grad_norm": 4.971385955810547, | |
| "learning_rate": 4.089577453065864e-05, | |
| "loss": 0.3829, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 2.1306703973499284, | |
| "grad_norm": 0.7727516889572144, | |
| "learning_rate": 4.077360967681622e-05, | |
| "loss": 0.3832, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 2.1387106630003054, | |
| "grad_norm": 2.056898355484009, | |
| "learning_rate": 4.06508158236109e-05, | |
| "loss": 0.3578, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 2.1467509286506825, | |
| "grad_norm": 1.9157487154006958, | |
| "learning_rate": 4.0527397867655265e-05, | |
| "loss": 0.4017, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 2.1547911943010596, | |
| "grad_norm": 0.3201541006565094, | |
| "learning_rate": 4.0403360730449024e-05, | |
| "loss": 0.389, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 2.1628314599514367, | |
| "grad_norm": 2.8520660400390625, | |
| "learning_rate": 4.027870935818281e-05, | |
| "loss": 0.3887, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 2.1708717256018137, | |
| "grad_norm": 6.087955474853516, | |
| "learning_rate": 4.01534487215409e-05, | |
| "loss": 0.3903, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 2.178911991252191, | |
| "grad_norm": 1.2185256481170654, | |
| "learning_rate": 4.0027583815503036e-05, | |
| "loss": 0.3719, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 2.186952256902568, | |
| "grad_norm": 2.539760112762451, | |
| "learning_rate": 3.9901119659145205e-05, | |
| "loss": 0.3725, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 2.194992522552945, | |
| "grad_norm": 1.0864566564559937, | |
| "learning_rate": 3.977406129543954e-05, | |
| "loss": 0.3778, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 2.203032788203322, | |
| "grad_norm": 3.848823308944702, | |
| "learning_rate": 3.964641379105315e-05, | |
| "loss": 0.3731, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 2.211073053853699, | |
| "grad_norm": 3.8714206218719482, | |
| "learning_rate": 3.951818223614619e-05, | |
| "loss": 0.4081, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 2.2191133195040766, | |
| "grad_norm": 6.581301689147949, | |
| "learning_rate": 3.938937174416876e-05, | |
| "loss": 0.3877, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 2.2271535851544533, | |
| "grad_norm": 1.3866485357284546, | |
| "learning_rate": 3.9259987451657074e-05, | |
| "loss": 0.3679, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 2.235193850804831, | |
| "grad_norm": 3.347809314727783, | |
| "learning_rate": 3.913003451802863e-05, | |
| "loss": 0.3855, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 2.243234116455208, | |
| "grad_norm": 2.701221466064453, | |
| "learning_rate": 3.899951812537645e-05, | |
| "loss": 0.3784, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 2.251274382105585, | |
| "grad_norm": 32.630226135253906, | |
| "learning_rate": 3.886844347826241e-05, | |
| "loss": 0.3829, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 2.259314647755962, | |
| "grad_norm": 8.304617881774902, | |
| "learning_rate": 3.873681580350973e-05, | |
| "loss": 0.3748, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 2.267354913406339, | |
| "grad_norm": 1.080276608467102, | |
| "learning_rate": 3.860464034999458e-05, | |
| "loss": 0.3604, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 2.275395179056716, | |
| "grad_norm": 3.447180986404419, | |
| "learning_rate": 3.84719223884367e-05, | |
| "loss": 0.387, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 2.2834354447070933, | |
| "grad_norm": 5.939009189605713, | |
| "learning_rate": 3.833866721118926e-05, | |
| "loss": 0.3879, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 2.2914757103574703, | |
| "grad_norm": 3.8884963989257812, | |
| "learning_rate": 3.820488013202782e-05, | |
| "loss": 0.3578, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 2.2995159760078474, | |
| "grad_norm": 5.440164089202881, | |
| "learning_rate": 3.807056648593844e-05, | |
| "loss": 0.3636, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 2.3075562416582245, | |
| "grad_norm": 1.5944666862487793, | |
| "learning_rate": 3.7935731628904916e-05, | |
| "loss": 0.3725, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 2.3155965073086016, | |
| "grad_norm": 7.6897969245910645, | |
| "learning_rate": 3.780038093769519e-05, | |
| "loss": 0.3875, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 2.3236367729589786, | |
| "grad_norm": 5.982831001281738, | |
| "learning_rate": 3.7664519809647004e-05, | |
| "loss": 0.3817, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 2.3316770386093557, | |
| "grad_norm": 3.50237774848938, | |
| "learning_rate": 3.752815366245261e-05, | |
| "loss": 0.3902, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 2.339717304259733, | |
| "grad_norm": 3.9120254516601562, | |
| "learning_rate": 3.7391287933942756e-05, | |
| "loss": 0.3625, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 2.34775756991011, | |
| "grad_norm": 10.226737022399902, | |
| "learning_rate": 3.7253928081869836e-05, | |
| "loss": 0.3679, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 2.355797835560487, | |
| "grad_norm": 9.554278373718262, | |
| "learning_rate": 3.711607958369026e-05, | |
| "loss": 0.3884, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 2.363838101210864, | |
| "grad_norm": 5.923080921173096, | |
| "learning_rate": 3.697774793634603e-05, | |
| "loss": 0.3753, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 2.371878366861241, | |
| "grad_norm": 1.8321458101272583, | |
| "learning_rate": 3.683893865604554e-05, | |
| "loss": 0.393, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 2.379918632511618, | |
| "grad_norm": 6.580794811248779, | |
| "learning_rate": 3.66996572780436e-05, | |
| "loss": 0.3694, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 2.3879588981619952, | |
| "grad_norm": 1.1640784740447998, | |
| "learning_rate": 3.655990935642071e-05, | |
| "loss": 0.371, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 2.3959991638123723, | |
| "grad_norm": 0.858772873878479, | |
| "learning_rate": 3.641970046386161e-05, | |
| "loss": 0.3701, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 2.4040394294627494, | |
| "grad_norm": 3.2022650241851807, | |
| "learning_rate": 3.627903619143299e-05, | |
| "loss": 0.3733, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 2.4120796951131265, | |
| "grad_norm": 1.4441760778427124, | |
| "learning_rate": 3.6137922148360645e-05, | |
| "loss": 0.3684, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.4201199607635036, | |
| "grad_norm": 2.272792339324951, | |
| "learning_rate": 3.5996363961805666e-05, | |
| "loss": 0.3612, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 2.4281602264138806, | |
| "grad_norm": 5.1610188484191895, | |
| "learning_rate": 3.585436727664019e-05, | |
| "loss": 0.4028, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 2.4362004920642577, | |
| "grad_norm": 2.840369462966919, | |
| "learning_rate": 3.5711937755222145e-05, | |
| "loss": 0.3666, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 2.444240757714635, | |
| "grad_norm": 3.4341518878936768, | |
| "learning_rate": 3.556908107716963e-05, | |
| "loss": 0.3646, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 2.452281023365012, | |
| "grad_norm": 0.8975893259048462, | |
| "learning_rate": 3.5425802939134236e-05, | |
| "loss": 0.3655, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 2.460321289015389, | |
| "grad_norm": 0.6069294214248657, | |
| "learning_rate": 3.528210905457407e-05, | |
| "loss": 0.3761, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 2.468361554665766, | |
| "grad_norm": 1.6750755310058594, | |
| "learning_rate": 3.5138005153525794e-05, | |
| "loss": 0.3675, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 2.476401820316143, | |
| "grad_norm": 3.3722386360168457, | |
| "learning_rate": 3.499349698237618e-05, | |
| "loss": 0.3546, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 2.48444208596652, | |
| "grad_norm": 6.766351222991943, | |
| "learning_rate": 3.4848590303632936e-05, | |
| "loss": 0.379, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 2.4924823516168972, | |
| "grad_norm": 3.2019455432891846, | |
| "learning_rate": 3.470329089569497e-05, | |
| "loss": 0.3441, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 2.5005226172672748, | |
| "grad_norm": 3.39136004447937, | |
| "learning_rate": 3.455760455262189e-05, | |
| "loss": 0.3939, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 2.5085628829176514, | |
| "grad_norm": 2.4117720127105713, | |
| "learning_rate": 3.441153708390304e-05, | |
| "loss": 0.3623, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 2.516603148568029, | |
| "grad_norm": 3.9395382404327393, | |
| "learning_rate": 3.4265094314225746e-05, | |
| "loss": 0.3929, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 2.5246434142184055, | |
| "grad_norm": 5.075194835662842, | |
| "learning_rate": 3.4118282083243144e-05, | |
| "loss": 0.3744, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 2.532683679868783, | |
| "grad_norm": 1.2141026258468628, | |
| "learning_rate": 3.397110624534122e-05, | |
| "loss": 0.3573, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 2.5407239455191597, | |
| "grad_norm": 9.561954498291016, | |
| "learning_rate": 3.382357266940543e-05, | |
| "loss": 0.3653, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 2.548764211169537, | |
| "grad_norm": 3.9251394271850586, | |
| "learning_rate": 3.367568723858662e-05, | |
| "loss": 0.3723, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 2.5568044768199143, | |
| "grad_norm": 2.572277545928955, | |
| "learning_rate": 3.3527455850066474e-05, | |
| "loss": 0.3563, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 2.5648447424702914, | |
| "grad_norm": 6.552402496337891, | |
| "learning_rate": 3.3378884414822286e-05, | |
| "loss": 0.3561, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 2.5728850081206684, | |
| "grad_norm": 4.849091529846191, | |
| "learning_rate": 3.322997885739132e-05, | |
| "loss": 0.3716, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.5809252737710455, | |
| "grad_norm": 2.6486380100250244, | |
| "learning_rate": 3.3080745115634505e-05, | |
| "loss": 0.3871, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 2.5889655394214226, | |
| "grad_norm": 1.2893551588058472, | |
| "learning_rate": 3.293118914049968e-05, | |
| "loss": 0.3596, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 2.5970058050717997, | |
| "grad_norm": 10.066678047180176, | |
| "learning_rate": 3.27813168957843e-05, | |
| "loss": 0.3496, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 2.6050460707221768, | |
| "grad_norm": 3.0136020183563232, | |
| "learning_rate": 3.26311343578976e-05, | |
| "loss": 0.3643, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 2.613086336372554, | |
| "grad_norm": 1.910649299621582, | |
| "learning_rate": 3.248064751562226e-05, | |
| "loss": 0.371, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 2.621126602022931, | |
| "grad_norm": 2.4634552001953125, | |
| "learning_rate": 3.232986236987563e-05, | |
| "loss": 0.3785, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 2.629166867673308, | |
| "grad_norm": 2.0846199989318848, | |
| "learning_rate": 3.217878493347041e-05, | |
| "loss": 0.3573, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 2.637207133323685, | |
| "grad_norm": 4.06114387512207, | |
| "learning_rate": 3.2027421230874885e-05, | |
| "loss": 0.3649, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 2.645247398974062, | |
| "grad_norm": 7.428652763366699, | |
| "learning_rate": 3.1875777297972694e-05, | |
| "loss": 0.3627, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 2.653287664624439, | |
| "grad_norm": 3.985434055328369, | |
| "learning_rate": 3.1723859181822125e-05, | |
| "loss": 0.3766, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 2.6613279302748163, | |
| "grad_norm": 0.9960327744483948, | |
| "learning_rate": 3.157167294041499e-05, | |
| "loss": 0.3545, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 2.6693681959251934, | |
| "grad_norm": 3.332768201828003, | |
| "learning_rate": 3.141922464243505e-05, | |
| "loss": 0.3572, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 2.6774084615755704, | |
| "grad_norm": 10.059576988220215, | |
| "learning_rate": 3.126652036701601e-05, | |
| "loss": 0.3771, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 2.6854487272259475, | |
| "grad_norm": 7.5988311767578125, | |
| "learning_rate": 3.1113566203499124e-05, | |
| "loss": 0.3843, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 2.6934889928763246, | |
| "grad_norm": 5.065057754516602, | |
| "learning_rate": 3.096036825119033e-05, | |
| "loss": 0.3926, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 2.7015292585267017, | |
| "grad_norm": 0.4597207009792328, | |
| "learning_rate": 3.080693261911709e-05, | |
| "loss": 0.3635, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 2.7095695241770787, | |
| "grad_norm": 0.6170072555541992, | |
| "learning_rate": 3.065326542578471e-05, | |
| "loss": 0.3559, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 2.717609789827456, | |
| "grad_norm": 14.608444213867188, | |
| "learning_rate": 3.0499372798932424e-05, | |
| "loss": 0.3613, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 2.725650055477833, | |
| "grad_norm": 5.434160232543945, | |
| "learning_rate": 3.0345260875289e-05, | |
| "loss": 0.3592, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 2.73369032112821, | |
| "grad_norm": 3.654452085494995, | |
| "learning_rate": 3.019093580032803e-05, | |
| "loss": 0.3831, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 2.741730586778587, | |
| "grad_norm": 2.3218584060668945, | |
| "learning_rate": 3.003640372802287e-05, | |
| "loss": 0.3373, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 2.749770852428964, | |
| "grad_norm": 4.1407470703125, | |
| "learning_rate": 2.988167082060127e-05, | |
| "loss": 0.3668, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 2.757811118079341, | |
| "grad_norm": 9.395963668823242, | |
| "learning_rate": 2.9726743248299603e-05, | |
| "loss": 0.3548, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 2.7658513837297183, | |
| "grad_norm": 2.468602180480957, | |
| "learning_rate": 2.957162718911683e-05, | |
| "loss": 0.3374, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 2.7738916493800954, | |
| "grad_norm": 4.715238094329834, | |
| "learning_rate": 2.9416328828568152e-05, | |
| "loss": 0.3507, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 2.781931915030473, | |
| "grad_norm": 1.999306082725525, | |
| "learning_rate": 2.926085435943834e-05, | |
| "loss": 0.3465, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.7899721806808495, | |
| "grad_norm": 1.5468521118164062, | |
| "learning_rate": 2.9105209981534798e-05, | |
| "loss": 0.3627, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.798012446331227, | |
| "grad_norm": 2.314556360244751, | |
| "learning_rate": 2.894940190144033e-05, | |
| "loss": 0.3795, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.8060527119816037, | |
| "grad_norm": 5.513248920440674, | |
| "learning_rate": 2.8793436332265638e-05, | |
| "loss": 0.3477, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.814092977631981, | |
| "grad_norm": 5.006555080413818, | |
| "learning_rate": 2.863731949340157e-05, | |
| "loss": 0.3763, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.822133243282358, | |
| "grad_norm": 0.8381386995315552, | |
| "learning_rate": 2.84810576102711e-05, | |
| "loss": 0.3399, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.8301735089327353, | |
| "grad_norm": 5.990314960479736, | |
| "learning_rate": 2.832465691408111e-05, | |
| "loss": 0.3748, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.838213774583112, | |
| "grad_norm": 0.01276963297277689, | |
| "learning_rate": 2.8168123641573856e-05, | |
| "loss": 0.3532, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.8462540402334895, | |
| "grad_norm": 4.30054235458374, | |
| "learning_rate": 2.801146403477832e-05, | |
| "loss": 0.3627, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.8542943058838666, | |
| "grad_norm": 2.3280787467956543, | |
| "learning_rate": 2.7854684340761283e-05, | |
| "loss": 0.3805, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.8623345715342436, | |
| "grad_norm": 3.336151123046875, | |
| "learning_rate": 2.7697790811378187e-05, | |
| "loss": 0.3491, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.8703748371846207, | |
| "grad_norm": 1.300593614578247, | |
| "learning_rate": 2.7540789703023857e-05, | |
| "loss": 0.3849, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.878415102834998, | |
| "grad_norm": 5.360976219177246, | |
| "learning_rate": 2.738368727638303e-05, | |
| "loss": 0.3253, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.886455368485375, | |
| "grad_norm": 11.331759452819824, | |
| "learning_rate": 2.7226489796180648e-05, | |
| "loss": 0.3817, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.894495634135752, | |
| "grad_norm": 7.809913158416748, | |
| "learning_rate": 2.7069203530932113e-05, | |
| "loss": 0.336, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.902535899786129, | |
| "grad_norm": 5.816734313964844, | |
| "learning_rate": 2.6911834752693255e-05, | |
| "loss": 0.3355, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.910576165436506, | |
| "grad_norm": 1.9233486652374268, | |
| "learning_rate": 2.6754389736810243e-05, | |
| "loss": 0.3883, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.918616431086883, | |
| "grad_norm": 1.3360695838928223, | |
| "learning_rate": 2.6596874761669382e-05, | |
| "loss": 0.3717, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.9266566967372603, | |
| "grad_norm": 6.034420967102051, | |
| "learning_rate": 2.6439296108446694e-05, | |
| "loss": 0.3531, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.9346969623876373, | |
| "grad_norm": 1.9546772241592407, | |
| "learning_rate": 2.6281660060857478e-05, | |
| "loss": 0.3433, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.9427372280380144, | |
| "grad_norm": 2.3421380519866943, | |
| "learning_rate": 2.612397290490573e-05, | |
| "loss": 0.3594, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.9507774936883915, | |
| "grad_norm": 3.1251261234283447, | |
| "learning_rate": 2.5966240928633494e-05, | |
| "loss": 0.3795, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.9588177593387686, | |
| "grad_norm": 5.281341075897217, | |
| "learning_rate": 2.5808470421870072e-05, | |
| "loss": 0.3424, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.9668580249891456, | |
| "grad_norm": 1.9180567264556885, | |
| "learning_rate": 2.5650667675981262e-05, | |
| "loss": 0.3471, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.9748982906395227, | |
| "grad_norm": 13.388057708740234, | |
| "learning_rate": 2.5492838983618428e-05, | |
| "loss": 0.3667, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.9829385562899, | |
| "grad_norm": 1.5807502269744873, | |
| "learning_rate": 2.533499063846762e-05, | |
| "loss": 0.3576, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.990978821940277, | |
| "grad_norm": 6.836522102355957, | |
| "learning_rate": 2.5177128934998545e-05, | |
| "loss": 0.3533, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.999019087590654, | |
| "grad_norm": 3.711378335952759, | |
| "learning_rate": 2.501926016821362e-05, | |
| "loss": 0.3367, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_BOD_f1": 0.2374555224747658, | |
| "eval_BUILDING_f1": 0.28714220808723306, | |
| "eval_CARDISSUER_f1": 0.0, | |
| "eval_CITY_f1": 0.25397850725233423, | |
| "eval_COUNTRY_f1": 0.3055481815768516, | |
| "eval_DATE_f1": 0.2341297523543774, | |
| "eval_DRIVERLICENSE_f1": 0.2233434378369581, | |
| "eval_EMAIL_f1": 0.2653814157051398, | |
| "eval_GEOCOORD_f1": 0.16032388663967612, | |
| "eval_GIVENNAME1_f1": 0.21614488277411203, | |
| "eval_GIVENNAME2_f1": 0.15067155067155066, | |
| "eval_IDCARD_f1": 0.24722656398917311, | |
| "eval_IP_f1": 0.18513513513513513, | |
| "eval_LASTNAME1_f1": 0.22957798483959949, | |
| "eval_LASTNAME2_f1": 0.13049962714392246, | |
| "eval_LASTNAME3_f1": 0.12451593684837652, | |
| "eval_PASSPORT_f1": 0.27920583363806495, | |
| "eval_PASS_f1": 0.19796215429403202, | |
| "eval_POSTCODE_f1": 0.2794223307946544, | |
| "eval_SECADDRESS_f1": 0.2486166906692943, | |
| "eval_SEX_f1": 0.29331175212440025, | |
| "eval_SOCIALNUMBER_f1": 0.22580711200278417, | |
| "eval_STATE_f1": 0.29213965980304385, | |
| "eval_STREET_f1": 0.21770362646275057, | |
| "eval_TEL_f1": 0.24091208570855654, | |
| "eval_TIME_f1": 0.28926309929719574, | |
| "eval_TITLE_f1": 0.2814272766943037, | |
| "eval_USERNAME_f1": 0.23682905559070463, | |
| "eval_loss": 0.36858755350112915, | |
| "eval_overall_accuracy": 0.8688154462267319, | |
| "eval_overall_f1": 0.24593334983551945, | |
| "eval_overall_precision": 0.2884671977802535, | |
| "eval_overall_recall": 0.2143307557001095, | |
| "eval_runtime": 651.3527, | |
| "eval_samples_per_second": 81.836, | |
| "eval_steps_per_second": 40.918, | |
| "step": 186561 | |
| }, | |
| { | |
| "epoch": 3.007059353241031, | |
| "grad_norm": 1.4970557689666748, | |
| "learning_rate": 2.4861390633396914e-05, | |
| "loss": 0.297, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 3.015099618891408, | |
| "grad_norm": 4.103343486785889, | |
| "learning_rate": 2.4703526625863127e-05, | |
| "loss": 0.3009, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 3.023139884541785, | |
| "grad_norm": 3.1448333263397217, | |
| "learning_rate": 2.4545674440706536e-05, | |
| "loss": 0.3032, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 3.0311801501921622, | |
| "grad_norm": 0.6826614737510681, | |
| "learning_rate": 2.4387840372550003e-05, | |
| "loss": 0.3175, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 3.0392204158425393, | |
| "grad_norm": 7.4535651206970215, | |
| "learning_rate": 2.4230030715293922e-05, | |
| "loss": 0.3245, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 3.0472606814929164, | |
| "grad_norm": 2.087162971496582, | |
| "learning_rate": 2.4072251761865274e-05, | |
| "loss": 0.316, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 3.0553009471432935, | |
| "grad_norm": 12.662734985351562, | |
| "learning_rate": 2.391450980396668e-05, | |
| "loss": 0.3173, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 3.0633412127936706, | |
| "grad_norm": 2.1878128051757812, | |
| "learning_rate": 2.375681113182547e-05, | |
| "loss": 0.3471, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 3.0713814784440476, | |
| "grad_norm": 3.3643546104431152, | |
| "learning_rate": 2.3599162033942926e-05, | |
| "loss": 0.3155, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 3.0794217440944247, | |
| "grad_norm": 4.645682334899902, | |
| "learning_rate": 2.344156879684343e-05, | |
| "loss": 0.2922, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 3.087462009744802, | |
| "grad_norm": 7.068358421325684, | |
| "learning_rate": 2.3284037704823854e-05, | |
| "loss": 0.302, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 3.095502275395179, | |
| "grad_norm": 4.4994611740112305, | |
| "learning_rate": 2.3126575039702906e-05, | |
| "loss": 0.3305, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 3.103542541045556, | |
| "grad_norm": 2.6543383598327637, | |
| "learning_rate": 2.2969187080570673e-05, | |
| "loss": 0.314, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 3.1115828066959335, | |
| "grad_norm": 3.4015653133392334, | |
| "learning_rate": 2.28118801035382e-05, | |
| "loss": 0.3051, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 3.1196230723463105, | |
| "grad_norm": 0.8205671310424805, | |
| "learning_rate": 2.265466038148724e-05, | |
| "loss": 0.3187, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 3.1276633379966876, | |
| "grad_norm": 6.238074779510498, | |
| "learning_rate": 2.2497534183820118e-05, | |
| "loss": 0.3108, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 3.1357036036470647, | |
| "grad_norm": 6.068066120147705, | |
| "learning_rate": 2.2340507776209697e-05, | |
| "loss": 0.3361, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 3.1437438692974418, | |
| "grad_norm": 1.5126768350601196, | |
| "learning_rate": 2.2183587420349553e-05, | |
| "loss": 0.3268, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 3.151784134947819, | |
| "grad_norm": 5.775200366973877, | |
| "learning_rate": 2.2026779373704258e-05, | |
| "loss": 0.3172, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 3.159824400598196, | |
| "grad_norm": 2.7656421661376953, | |
| "learning_rate": 2.187008988925989e-05, | |
| "loss": 0.3239, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 3.167864666248573, | |
| "grad_norm": 8.297327995300293, | |
| "learning_rate": 2.1713525215274623e-05, | |
| "loss": 0.3073, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 3.17590493189895, | |
| "grad_norm": 5.742626667022705, | |
| "learning_rate": 2.1557091595029637e-05, | |
| "loss": 0.3092, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 3.183945197549327, | |
| "grad_norm": 8.81966495513916, | |
| "learning_rate": 2.1400795266580093e-05, | |
| "loss": 0.3186, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 3.191985463199704, | |
| "grad_norm": 10.023892402648926, | |
| "learning_rate": 2.1244642462506435e-05, | |
| "loss": 0.3089, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 3.2000257288500813, | |
| "grad_norm": 4.242392539978027, | |
| "learning_rate": 2.1088639409665808e-05, | |
| "loss": 0.2912, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 3.2080659945004584, | |
| "grad_norm": 2.3702950477600098, | |
| "learning_rate": 2.0932792328943794e-05, | |
| "loss": 0.3217, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 3.2161062601508354, | |
| "grad_norm": 3.393665313720703, | |
| "learning_rate": 2.07771074350063e-05, | |
| "loss": 0.3073, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 3.2241465258012125, | |
| "grad_norm": 2.3688278198242188, | |
| "learning_rate": 2.0621590936051782e-05, | |
| "loss": 0.3398, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 3.2321867914515896, | |
| "grad_norm": 0.7889087796211243, | |
| "learning_rate": 2.0466249033563648e-05, | |
| "loss": 0.3192, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 3.2402270571019667, | |
| "grad_norm": 5.064262390136719, | |
| "learning_rate": 2.0311087922062984e-05, | |
| "loss": 0.3092, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 3.2482673227523438, | |
| "grad_norm": 5.569192886352539, | |
| "learning_rate": 2.0156113788861524e-05, | |
| "loss": 0.3158, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 3.256307588402721, | |
| "grad_norm": 5.586752414703369, | |
| "learning_rate": 2.0001332813814933e-05, | |
| "loss": 0.318, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 3.264347854053098, | |
| "grad_norm": 0.3673694133758545, | |
| "learning_rate": 1.9846751169076352e-05, | |
| "loss": 0.3108, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 3.272388119703475, | |
| "grad_norm": 7.468791484832764, | |
| "learning_rate": 1.9692375018850315e-05, | |
| "loss": 0.2924, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 3.280428385353852, | |
| "grad_norm": 3.7493669986724854, | |
| "learning_rate": 1.953821051914689e-05, | |
| "loss": 0.3202, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 3.288468651004229, | |
| "grad_norm": 9.650253295898438, | |
| "learning_rate": 1.938426381753624e-05, | |
| "loss": 0.3082, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 3.296508916654606, | |
| "grad_norm": 0.8365656733512878, | |
| "learning_rate": 1.9230541052903442e-05, | |
| "loss": 0.3106, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 3.3045491823049833, | |
| "grad_norm": 2.7463884353637695, | |
| "learning_rate": 1.9077048355203732e-05, | |
| "loss": 0.3053, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 3.3125894479553604, | |
| "grad_norm": 0.542148768901825, | |
| "learning_rate": 1.8923791845218015e-05, | |
| "loss": 0.3137, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 3.3206297136057374, | |
| "grad_norm": 0.447301983833313, | |
| "learning_rate": 1.8770777634308826e-05, | |
| "loss": 0.3117, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 3.3286699792561145, | |
| "grad_norm": 0.1598307341337204, | |
| "learning_rate": 1.8618011824176604e-05, | |
| "loss": 0.3164, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 3.3367102449064916, | |
| "grad_norm": 2.622774362564087, | |
| "learning_rate": 1.84655005066164e-05, | |
| "loss": 0.3163, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 3.3447505105568687, | |
| "grad_norm": 7.138498783111572, | |
| "learning_rate": 1.8313249763274928e-05, | |
| "loss": 0.3128, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 3.3527907762072457, | |
| "grad_norm": 7.845987319946289, | |
| "learning_rate": 1.816126566540808e-05, | |
| "loss": 0.3173, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 3.360831041857623, | |
| "grad_norm": 5.490318775177002, | |
| "learning_rate": 1.800955427363879e-05, | |
| "loss": 0.3176, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 3.368871307508, | |
| "grad_norm": 12.626228332519531, | |
| "learning_rate": 1.7858121637715397e-05, | |
| "loss": 0.336, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 3.376911573158377, | |
| "grad_norm": 4.305318355560303, | |
| "learning_rate": 1.770697379627036e-05, | |
| "loss": 0.2958, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 3.384951838808754, | |
| "grad_norm": 1.9120597839355469, | |
| "learning_rate": 1.755611677657949e-05, | |
| "loss": 0.3152, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 3.3929921044591316, | |
| "grad_norm": 7.113236904144287, | |
| "learning_rate": 1.740555659432158e-05, | |
| "loss": 0.303, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 3.401032370109508, | |
| "grad_norm": 6.3328680992126465, | |
| "learning_rate": 1.7255299253338537e-05, | |
| "loss": 0.2964, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 3.4090726357598857, | |
| "grad_norm": 0.872886061668396, | |
| "learning_rate": 1.7105350745395936e-05, | |
| "loss": 0.3331, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 3.4171129014102624, | |
| "grad_norm": 0.8954824805259705, | |
| "learning_rate": 1.6955717049944153e-05, | |
| "loss": 0.2972, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 3.42515316706064, | |
| "grad_norm": 1.4798979759216309, | |
| "learning_rate": 1.680640413387986e-05, | |
| "loss": 0.3012, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 3.433193432711017, | |
| "grad_norm": 1.585610270500183, | |
| "learning_rate": 1.6657417951308098e-05, | |
| "loss": 0.3144, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 3.441233698361394, | |
| "grad_norm": 3.92868971824646, | |
| "learning_rate": 1.6508764443304876e-05, | |
| "loss": 0.3114, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 3.449273964011771, | |
| "grad_norm": 5.895984172821045, | |
| "learning_rate": 1.636044953768023e-05, | |
| "loss": 0.3074, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 3.457314229662148, | |
| "grad_norm": 1.7379788160324097, | |
| "learning_rate": 1.6212479148741866e-05, | |
| "loss": 0.3042, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 3.4653544953125253, | |
| "grad_norm": 9.651080131530762, | |
| "learning_rate": 1.606485917705929e-05, | |
| "loss": 0.304, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 3.4733947609629023, | |
| "grad_norm": 4.100153923034668, | |
| "learning_rate": 1.591759550922854e-05, | |
| "loss": 0.32, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 3.4814350266132794, | |
| "grad_norm": 2.7346136569976807, | |
| "learning_rate": 1.5770694017637423e-05, | |
| "loss": 0.3246, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 3.4894752922636565, | |
| "grad_norm": 2.6546661853790283, | |
| "learning_rate": 1.562416056023137e-05, | |
| "loss": 0.3376, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 3.4975155579140336, | |
| "grad_norm": 2.415144681930542, | |
| "learning_rate": 1.5478000980279812e-05, | |
| "loss": 0.3064, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 3.5055558235644106, | |
| "grad_norm": 6.306326389312744, | |
| "learning_rate": 1.5332221106143202e-05, | |
| "loss": 0.2905, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 3.5135960892147877, | |
| "grad_norm": 4.871018409729004, | |
| "learning_rate": 1.5186826751040553e-05, | |
| "loss": 0.2948, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 3.521636354865165, | |
| "grad_norm": 13.619241714477539, | |
| "learning_rate": 1.5041823712817685e-05, | |
| "loss": 0.3048, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 3.529676620515542, | |
| "grad_norm": 19.28036880493164, | |
| "learning_rate": 1.4897217773715985e-05, | |
| "loss": 0.2881, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 3.537716886165919, | |
| "grad_norm": 2.432436943054199, | |
| "learning_rate": 1.4753014700141826e-05, | |
| "loss": 0.2993, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 3.545757151816296, | |
| "grad_norm": 1.5334402322769165, | |
| "learning_rate": 1.4609220242436666e-05, | |
| "loss": 0.2971, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 3.553797417466673, | |
| "grad_norm": 1.596092700958252, | |
| "learning_rate": 1.4465840134647687e-05, | |
| "loss": 0.2993, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 3.56183768311705, | |
| "grad_norm": 1.7750720977783203, | |
| "learning_rate": 1.4322880094299177e-05, | |
| "loss": 0.2994, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 3.5698779487674273, | |
| "grad_norm": 4.999639511108398, | |
| "learning_rate": 1.4180345822164526e-05, | |
| "loss": 0.2961, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 3.5779182144178043, | |
| "grad_norm": 1.6016736030578613, | |
| "learning_rate": 1.4038243002038898e-05, | |
| "loss": 0.2791, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 3.5859584800681814, | |
| "grad_norm": 1.8162273168563843, | |
| "learning_rate": 1.3896577300512584e-05, | |
| "loss": 0.3175, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 3.5939987457185585, | |
| "grad_norm": 8.3301420211792, | |
| "learning_rate": 1.3755354366745005e-05, | |
| "loss": 0.3059, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 3.6020390113689356, | |
| "grad_norm": 4.8512187004089355, | |
| "learning_rate": 1.3614579832239503e-05, | |
| "loss": 0.3, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 3.6100792770193126, | |
| "grad_norm": 2.8132553100585938, | |
| "learning_rate": 1.3474259310618715e-05, | |
| "loss": 0.3056, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 3.6181195426696897, | |
| "grad_norm": 3.5456409454345703, | |
| "learning_rate": 1.3334398397400778e-05, | |
| "loss": 0.2956, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 3.626159808320067, | |
| "grad_norm": 4.033205032348633, | |
| "learning_rate": 1.3195002669776113e-05, | |
| "loss": 0.3004, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 3.634200073970444, | |
| "grad_norm": 2.705427646636963, | |
| "learning_rate": 1.3056077686385135e-05, | |
| "loss": 0.3259, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 3.642240339620821, | |
| "grad_norm": 6.071278095245361, | |
| "learning_rate": 1.2917628987096502e-05, | |
| "loss": 0.2781, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 3.650280605271198, | |
| "grad_norm": 1.0004699230194092, | |
| "learning_rate": 1.2779662092786266e-05, | |
| "loss": 0.3164, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 3.658320870921575, | |
| "grad_norm": 5.386746883392334, | |
| "learning_rate": 1.264218250511765e-05, | |
| "loss": 0.3081, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 3.666361136571952, | |
| "grad_norm": 2.796396493911743, | |
| "learning_rate": 1.2505195706321732e-05, | |
| "loss": 0.3024, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 3.6744014022223297, | |
| "grad_norm": 3.1216211318969727, | |
| "learning_rate": 1.2368707158978795e-05, | |
| "loss": 0.3061, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 3.6824416678727063, | |
| "grad_norm": 6.516884803771973, | |
| "learning_rate": 1.223272230580051e-05, | |
| "loss": 0.3046, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 3.690481933523084, | |
| "grad_norm": 1.585449457168579, | |
| "learning_rate": 1.209724656941286e-05, | |
| "loss": 0.2909, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 3.6985221991734605, | |
| "grad_norm": 2.5586652755737305, | |
| "learning_rate": 1.1962285352139968e-05, | |
| "loss": 0.3033, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 3.706562464823838, | |
| "grad_norm": 13.890535354614258, | |
| "learning_rate": 1.1827844035788622e-05, | |
| "loss": 0.2895, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 3.7146027304742146, | |
| "grad_norm": 7.9294233322143555, | |
| "learning_rate": 1.1693927981433687e-05, | |
| "loss": 0.3093, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 3.722642996124592, | |
| "grad_norm": 1.08167564868927, | |
| "learning_rate": 1.1560542529204312e-05, | |
| "loss": 0.3001, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 3.730683261774969, | |
| "grad_norm": 2.191358804702759, | |
| "learning_rate": 1.1427692998071e-05, | |
| "loss": 0.2963, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 3.7387235274253463, | |
| "grad_norm": 5.05871057510376, | |
| "learning_rate": 1.1295384685633487e-05, | |
| "loss": 0.305, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 3.7467637930757234, | |
| "grad_norm": 4.525569915771484, | |
| "learning_rate": 1.116362286790948e-05, | |
| "loss": 0.2826, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 3.7548040587261005, | |
| "grad_norm": 1.5160119533538818, | |
| "learning_rate": 1.1032412799124314e-05, | |
| "loss": 0.3061, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 3.7628443243764775, | |
| "grad_norm": 0.28604656457901, | |
| "learning_rate": 1.0901759711501388e-05, | |
| "loss": 0.3063, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 3.7708845900268546, | |
| "grad_norm": 1.981669306755066, | |
| "learning_rate": 1.0771668815053548e-05, | |
| "loss": 0.284, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 3.7789248556772317, | |
| "grad_norm": 3.0476274490356445, | |
| "learning_rate": 1.064214529737529e-05, | |
| "loss": 0.3071, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 3.7869651213276088, | |
| "grad_norm": 0.9368652105331421, | |
| "learning_rate": 1.0513194323435938e-05, | |
| "loss": 0.3066, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 3.795005386977986, | |
| "grad_norm": 4.352938652038574, | |
| "learning_rate": 1.0384821035373673e-05, | |
| "loss": 0.2915, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 3.803045652628363, | |
| "grad_norm": 2.2861974239349365, | |
| "learning_rate": 1.0257030552290473e-05, | |
| "loss": 0.2969, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 3.81108591827874, | |
| "grad_norm": 5.970806121826172, | |
| "learning_rate": 1.0129827970047959e-05, | |
| "loss": 0.3166, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 3.819126183929117, | |
| "grad_norm": 2.2724108695983887, | |
| "learning_rate": 1.0003218361064237e-05, | |
| "loss": 0.3052, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 3.827166449579494, | |
| "grad_norm": 7.409987449645996, | |
| "learning_rate": 9.877206774111593e-06, | |
| "loss": 0.2881, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 3.835206715229871, | |
| "grad_norm": 2.1269917488098145, | |
| "learning_rate": 9.751798234115183e-06, | |
| "loss": 0.3008, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 3.8432469808802483, | |
| "grad_norm": 1.2044695615768433, | |
| "learning_rate": 9.626997741952618e-06, | |
| "loss": 0.2894, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 3.8512872465306254, | |
| "grad_norm": 2.6249988079071045, | |
| "learning_rate": 9.502810274254598e-06, | |
| "loss": 0.285, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 3.8593275121810025, | |
| "grad_norm": 2.9487357139587402, | |
| "learning_rate": 9.379240783206427e-06, | |
| "loss": 0.3121, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 3.8673677778313795, | |
| "grad_norm": 5.342014312744141, | |
| "learning_rate": 9.256294196350565e-06, | |
| "loss": 0.2873, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 3.8754080434817566, | |
| "grad_norm": 0.5511460304260254, | |
| "learning_rate": 9.133975416390068e-06, | |
| "loss": 0.2886, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 3.8834483091321337, | |
| "grad_norm": 1.6954889297485352, | |
| "learning_rate": 9.01228932099317e-06, | |
| "loss": 0.2971, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 3.8914885747825108, | |
| "grad_norm": 3.035940647125244, | |
| "learning_rate": 8.89124076259873e-06, | |
| "loss": 0.2984, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 3.899528840432888, | |
| "grad_norm": 6.392285346984863, | |
| "learning_rate": 8.770834568222737e-06, | |
| "loss": 0.283, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 3.907569106083265, | |
| "grad_norm": 1.361126184463501, | |
| "learning_rate": 8.651075539265819e-06, | |
| "loss": 0.2877, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 3.915609371733642, | |
| "grad_norm": 3.15057110786438, | |
| "learning_rate": 8.5319684513218e-06, | |
| "loss": 0.2965, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 3.923649637384019, | |
| "grad_norm": 3.342039108276367, | |
| "learning_rate": 8.413518053987257e-06, | |
| "loss": 0.2769, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 3.931689903034396, | |
| "grad_norm": 4.928660869598389, | |
| "learning_rate": 8.295729070672115e-06, | |
| "loss": 0.2982, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 3.939730168684773, | |
| "grad_norm": 5.679259777069092, | |
| "learning_rate": 8.17860619841128e-06, | |
| "loss": 0.2961, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 3.9477704343351503, | |
| "grad_norm": 0.06428790092468262, | |
| "learning_rate": 8.062154107677374e-06, | |
| "loss": 0.2967, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 3.9558106999855274, | |
| "grad_norm": 4.07528018951416, | |
| "learning_rate": 7.946377442194464e-06, | |
| "loss": 0.2951, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 3.9638509656359044, | |
| "grad_norm": 5.545145034790039, | |
| "learning_rate": 7.831280818752903e-06, | |
| "loss": 0.3032, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 3.971891231286282, | |
| "grad_norm": 1.64756441116333, | |
| "learning_rate": 7.716868827025189e-06, | |
| "loss": 0.293, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 3.9799314969366586, | |
| "grad_norm": 1.6225277185440063, | |
| "learning_rate": 7.603146029382999e-06, | |
| "loss": 0.302, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 3.987971762587036, | |
| "grad_norm": 2.7999932765960693, | |
| "learning_rate": 7.490116960715221e-06, | |
| "loss": 0.2857, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 3.9960120282374127, | |
| "grad_norm": 2.5309460163116455, | |
| "learning_rate": 7.377786128247138e-06, | |
| "loss": 0.301, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_BOD_f1": 0.256508739091329, | |
| "eval_BUILDING_f1": 0.3272183720614214, | |
| "eval_CARDISSUER_f1": 0.14285714285714288, | |
| "eval_CITY_f1": 0.26337623012869044, | |
| "eval_COUNTRY_f1": 0.3355216881594373, | |
| "eval_DATE_f1": 0.2707292707292707, | |
| "eval_DRIVERLICENSE_f1": 0.259093115836489, | |
| "eval_EMAIL_f1": 0.3031833379771524, | |
| "eval_GEOCOORD_f1": 0.2153357171235598, | |
| "eval_GIVENNAME1_f1": 0.24575660940232483, | |
| "eval_GIVENNAME2_f1": 0.18468561942311681, | |
| "eval_IDCARD_f1": 0.27567022627137633, | |
| "eval_IP_f1": 0.22522886695993344, | |
| "eval_LASTNAME1_f1": 0.25935347769435996, | |
| "eval_LASTNAME2_f1": 0.16800833412254948, | |
| "eval_LASTNAME3_f1": 0.15509693558474047, | |
| "eval_PASSPORT_f1": 0.30801445229101654, | |
| "eval_PASS_f1": 0.24096695383824096, | |
| "eval_POSTCODE_f1": 0.2944592790387183, | |
| "eval_SECADDRESS_f1": 0.24883936861652736, | |
| "eval_SEX_f1": 0.3139299481405475, | |
| "eval_SOCIALNUMBER_f1": 0.25224933783537884, | |
| "eval_STATE_f1": 0.3007278020378457, | |
| "eval_STREET_f1": 0.2447499709943149, | |
| "eval_TEL_f1": 0.2584240464297058, | |
| "eval_TIME_f1": 0.3106610099904202, | |
| "eval_TITLE_f1": 0.2933262288530704, | |
| "eval_USERNAME_f1": 0.2879884225759769, | |
| "eval_loss": 0.3734145760536194, | |
| "eval_overall_accuracy": 0.8737132897986565, | |
| "eval_overall_f1": 0.2747378011131309, | |
| "eval_overall_precision": 0.30726104561181544, | |
| "eval_overall_recall": 0.2484406327663321, | |
| "eval_runtime": 653.1149, | |
| "eval_samples_per_second": 81.615, | |
| "eval_steps_per_second": 40.808, | |
| "step": 248748 | |
| }, | |
| { | |
| "epoch": 4.00405229388779, | |
| "grad_norm": 2.237973928451538, | |
| "learning_rate": 7.266158011360649e-06, | |
| "loss": 0.271, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 4.012092559538167, | |
| "grad_norm": 5.230581283569336, | |
| "learning_rate": 7.155237061415729e-06, | |
| "loss": 0.2592, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 4.020132825188544, | |
| "grad_norm": 4.403975486755371, | |
| "learning_rate": 7.045027701572842e-06, | |
| "loss": 0.2493, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 4.028173090838921, | |
| "grad_norm": 1.0564706325531006, | |
| "learning_rate": 6.935534326616613e-06, | |
| "loss": 0.2775, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 4.036213356489299, | |
| "grad_norm": 2.5550620555877686, | |
| "learning_rate": 6.826761302780535e-06, | |
| "loss": 0.2545, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 4.044253622139675, | |
| "grad_norm": 6.118185997009277, | |
| "learning_rate": 6.718712967572896e-06, | |
| "loss": 0.2569, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 4.052293887790053, | |
| "grad_norm": 2.3821053504943848, | |
| "learning_rate": 6.6113936296038045e-06, | |
| "loss": 0.2684, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 4.060334153440429, | |
| "grad_norm": 3.7676379680633545, | |
| "learning_rate": 6.504807568413371e-06, | |
| "loss": 0.2513, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 4.068374419090807, | |
| "grad_norm": 2.8824093341827393, | |
| "learning_rate": 6.398959034301033e-06, | |
| "loss": 0.2488, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 4.0764146847411835, | |
| "grad_norm": 12.843091011047363, | |
| "learning_rate": 6.293852248156113e-06, | |
| "loss": 0.2583, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 4.084454950391561, | |
| "grad_norm": 0.8622458577156067, | |
| "learning_rate": 6.189491401289465e-06, | |
| "loss": 0.2494, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 4.092495216041938, | |
| "grad_norm": 7.87520170211792, | |
| "learning_rate": 6.0858806552663735e-06, | |
| "loss": 0.2528, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 4.100535481692315, | |
| "grad_norm": 4.272862434387207, | |
| "learning_rate": 5.983024141740545e-06, | |
| "loss": 0.2567, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 4.108575747342692, | |
| "grad_norm": 9.08167552947998, | |
| "learning_rate": 5.880925962289422e-06, | |
| "loss": 0.2337, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 4.116616012993069, | |
| "grad_norm": 3.927924394607544, | |
| "learning_rate": 5.779590188250583e-06, | |
| "loss": 0.2525, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 4.124656278643446, | |
| "grad_norm": 6.682066440582275, | |
| "learning_rate": 5.6790208605594085e-06, | |
| "loss": 0.2348, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 4.1326965442938235, | |
| "grad_norm": 8.70799732208252, | |
| "learning_rate": 5.579221989587915e-06, | |
| "loss": 0.255, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 4.140736809944201, | |
| "grad_norm": 3.948021650314331, | |
| "learning_rate": 5.48019755498487e-06, | |
| "loss": 0.2546, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 4.148777075594578, | |
| "grad_norm": 3.1053335666656494, | |
| "learning_rate": 5.381951505517082e-06, | |
| "loss": 0.2639, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 4.156817341244954, | |
| "grad_norm": 3.1850690841674805, | |
| "learning_rate": 5.284487758911935e-06, | |
| "loss": 0.2405, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 4.164857606895332, | |
| "grad_norm": 6.587006568908691, | |
| "learning_rate": 5.187810201701149e-06, | |
| "loss": 0.2577, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 4.172897872545709, | |
| "grad_norm": 6.057389259338379, | |
| "learning_rate": 5.091922689065825e-06, | |
| "loss": 0.2393, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 4.180938138196086, | |
| "grad_norm": 6.784351825714111, | |
| "learning_rate": 4.996829044682708e-06, | |
| "loss": 0.279, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 4.1889784038464635, | |
| "grad_norm": 7.198575496673584, | |
| "learning_rate": 4.902533060571693e-06, | |
| "loss": 0.2661, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 4.19701866949684, | |
| "grad_norm": 6.301877498626709, | |
| "learning_rate": 4.809038496944612e-06, | |
| "loss": 0.2556, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 4.205058935147218, | |
| "grad_norm": 1.3413364887237549, | |
| "learning_rate": 4.716349082055319e-06, | |
| "loss": 0.2643, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 4.213099200797594, | |
| "grad_norm": 4.0268235206604, | |
| "learning_rate": 4.624468512050994e-06, | |
| "loss": 0.2515, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 4.221139466447972, | |
| "grad_norm": 2.4277729988098145, | |
| "learning_rate": 4.5334004508247655e-06, | |
| "loss": 0.255, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 4.229179732098348, | |
| "grad_norm": 4.360696792602539, | |
| "learning_rate": 4.4431485298695785e-06, | |
| "loss": 0.2344, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 4.237219997748726, | |
| "grad_norm": 6.219848155975342, | |
| "learning_rate": 4.35371634813343e-06, | |
| "loss": 0.2368, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 4.245260263399103, | |
| "grad_norm": 3.6934189796447754, | |
| "learning_rate": 4.265107471875812e-06, | |
| "loss": 0.263, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 4.25330052904948, | |
| "grad_norm": 0.771640419960022, | |
| "learning_rate": 4.1773254345255335e-06, | |
| "loss": 0.2455, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 4.261340794699857, | |
| "grad_norm": 3.0738141536712646, | |
| "learning_rate": 4.090373736539782e-06, | |
| "loss": 0.2422, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 4.269381060350234, | |
| "grad_norm": 10.216859817504883, | |
| "learning_rate": 4.004255845264579e-06, | |
| "loss": 0.2495, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 4.277421326000611, | |
| "grad_norm": 6.357425212860107, | |
| "learning_rate": 3.918975194796484e-06, | |
| "loss": 0.2514, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 4.285461591650988, | |
| "grad_norm": 14.743285179138184, | |
| "learning_rate": 3.834535185845672e-06, | |
| "loss": 0.2425, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 4.293501857301365, | |
| "grad_norm": 9.270341873168945, | |
| "learning_rate": 3.7509391856002966e-06, | |
| "loss": 0.2433, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 4.3015421229517425, | |
| "grad_norm": 3.2252187728881836, | |
| "learning_rate": 3.6681905275922466e-06, | |
| "loss": 0.2472, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 4.309582388602119, | |
| "grad_norm": 2.994213342666626, | |
| "learning_rate": 3.5862925115642293e-06, | |
| "loss": 0.2333, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 4.317622654252497, | |
| "grad_norm": 1.6329305171966553, | |
| "learning_rate": 3.505248403338124e-06, | |
| "loss": 0.2674, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 4.325662919902873, | |
| "grad_norm": 4.605139255523682, | |
| "learning_rate": 3.4250614346848174e-06, | |
| "loss": 0.2403, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 4.333703185553251, | |
| "grad_norm": 4.4717817306518555, | |
| "learning_rate": 3.3457348031953022e-06, | |
| "loss": 0.2375, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 4.3417434512036275, | |
| "grad_norm": 6.33212947845459, | |
| "learning_rate": 3.2672716721531717e-06, | |
| "loss": 0.2613, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 4.349783716854005, | |
| "grad_norm": 5.012121200561523, | |
| "learning_rate": 3.189675170408468e-06, | |
| "loss": 0.2661, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 4.357823982504382, | |
| "grad_norm": 7.388403415679932, | |
| "learning_rate": 3.1129483922529372e-06, | |
| "loss": 0.2554, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 4.365864248154759, | |
| "grad_norm": 9.56619930267334, | |
| "learning_rate": 3.037094397296622e-06, | |
| "loss": 0.2532, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 4.373904513805136, | |
| "grad_norm": 4.4190874099731445, | |
| "learning_rate": 2.9621162103458665e-06, | |
| "loss": 0.2471, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 4.381944779455513, | |
| "grad_norm": 0.8777914047241211, | |
| "learning_rate": 2.8880168212826715e-06, | |
| "loss": 0.2437, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 4.38998504510589, | |
| "grad_norm": 1.1738107204437256, | |
| "learning_rate": 2.8147991849454964e-06, | |
| "loss": 0.2501, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 4.3980253107562675, | |
| "grad_norm": 2.609962224960327, | |
| "learning_rate": 2.742466221011422e-06, | |
| "loss": 0.2451, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 4.406065576406644, | |
| "grad_norm": 0.6662173271179199, | |
| "learning_rate": 2.6710208138797267e-06, | |
| "loss": 0.2511, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 4.414105842057022, | |
| "grad_norm": 17.864526748657227, | |
| "learning_rate": 2.600465812556835e-06, | |
| "loss": 0.2399, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 4.422146107707398, | |
| "grad_norm": 7.632662296295166, | |
| "learning_rate": 2.5308040305427575e-06, | |
| "loss": 0.2614, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 4.430186373357776, | |
| "grad_norm": 0.22275756299495697, | |
| "learning_rate": 2.46203824571887e-06, | |
| "loss": 0.2499, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 4.438226639008153, | |
| "grad_norm": 2.5097222328186035, | |
| "learning_rate": 2.3941712002371443e-06, | |
| "loss": 0.2635, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 4.44626690465853, | |
| "grad_norm": 3.5931217670440674, | |
| "learning_rate": 2.3272056004107893e-06, | |
| "loss": 0.2697, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 4.4543071703089065, | |
| "grad_norm": 4.865581512451172, | |
| "learning_rate": 2.261144116606359e-06, | |
| "loss": 0.2457, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 4.462347435959284, | |
| "grad_norm": 1.557501196861267, | |
| "learning_rate": 2.195989383137245e-06, | |
| "loss": 0.2694, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 4.470387701609662, | |
| "grad_norm": 2.3111207485198975, | |
| "learning_rate": 2.1317439981586416e-06, | |
| "loss": 0.231, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 4.478427967260038, | |
| "grad_norm": 6.507102012634277, | |
| "learning_rate": 2.0684105235639237e-06, | |
| "loss": 0.2664, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 4.486468232910416, | |
| "grad_norm": 9.655716896057129, | |
| "learning_rate": 2.0059914848825024e-06, | |
| "loss": 0.251, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 4.494508498560792, | |
| "grad_norm": 4.831887245178223, | |
| "learning_rate": 1.9444893711791147e-06, | |
| "loss": 0.2507, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 4.50254876421117, | |
| "grad_norm": 6.776580810546875, | |
| "learning_rate": 1.8839066349545631e-06, | |
| "loss": 0.2496, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 4.5105890298615465, | |
| "grad_norm": 3.9420840740203857, | |
| "learning_rate": 1.8242456920479073e-06, | |
| "loss": 0.2558, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 4.518629295511924, | |
| "grad_norm": 3.8227767944335938, | |
| "learning_rate": 1.765508921540146e-06, | |
| "loss": 0.2302, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 4.526669561162301, | |
| "grad_norm": 3.5808634757995605, | |
| "learning_rate": 1.7076986656593492e-06, | |
| "loss": 0.2601, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 4.534709826812678, | |
| "grad_norm": 3.8970723152160645, | |
| "learning_rate": 1.6508172296872405e-06, | |
| "loss": 0.2454, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 4.542750092463055, | |
| "grad_norm": 3.7558376789093018, | |
| "learning_rate": 1.5948668818672713e-06, | |
| "loss": 0.2542, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 4.550790358113432, | |
| "grad_norm": 1.801005244255066, | |
| "learning_rate": 1.539849853314193e-06, | |
| "loss": 0.2469, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 4.558830623763809, | |
| "grad_norm": 7.793933391571045, | |
| "learning_rate": 1.485768337925067e-06, | |
| "loss": 0.2502, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 4.5668708894141865, | |
| "grad_norm": 6.472097396850586, | |
| "learning_rate": 1.4326244922917814e-06, | |
| "loss": 0.2339, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 4.574911155064563, | |
| "grad_norm": 1.343406319618225, | |
| "learning_rate": 1.3804204356150652e-06, | |
| "loss": 0.2553, | |
| "step": 284500 | |
| }, | |
| { | |
| "epoch": 4.582951420714941, | |
| "grad_norm": 2.0633301734924316, | |
| "learning_rate": 1.3291582496199633e-06, | |
| "loss": 0.2578, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 4.590991686365317, | |
| "grad_norm": 10.775948524475098, | |
| "learning_rate": 1.2788399784728372e-06, | |
| "loss": 0.2668, | |
| "step": 285500 | |
| }, | |
| { | |
| "epoch": 4.599031952015695, | |
| "grad_norm": 0.4255613386631012, | |
| "learning_rate": 1.2294676286998541e-06, | |
| "loss": 0.2315, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 4.6070722176660714, | |
| "grad_norm": 3.3115196228027344, | |
| "learning_rate": 1.181043169106963e-06, | |
| "loss": 0.2331, | |
| "step": 286500 | |
| }, | |
| { | |
| "epoch": 4.615112483316449, | |
| "grad_norm": 1.838809847831726, | |
| "learning_rate": 1.1335685307013816e-06, | |
| "loss": 0.2614, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 4.623152748966826, | |
| "grad_norm": 8.416891098022461, | |
| "learning_rate": 1.0870456066146145e-06, | |
| "loss": 0.2337, | |
| "step": 287500 | |
| }, | |
| { | |
| "epoch": 4.631193014617203, | |
| "grad_norm": 1.2128727436065674, | |
| "learning_rate": 1.0414762520269377e-06, | |
| "loss": 0.2532, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 4.63923328026758, | |
| "grad_norm": 4.361663341522217, | |
| "learning_rate": 9.968622840934361e-07, | |
| "loss": 0.2503, | |
| "step": 288500 | |
| }, | |
| { | |
| "epoch": 4.647273545917957, | |
| "grad_norm": 1.9845813512802124, | |
| "learning_rate": 9.532054818715302e-07, | |
| "loss": 0.262, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 4.655313811568334, | |
| "grad_norm": 0.5194038152694702, | |
| "learning_rate": 9.105075862500451e-07, | |
| "loss": 0.2495, | |
| "step": 289500 | |
| }, | |
| { | |
| "epoch": 4.663354077218711, | |
| "grad_norm": 5.471443176269531, | |
| "learning_rate": 8.687702998797842e-07, | |
| "loss": 0.2462, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 4.671394342869088, | |
| "grad_norm": 0.4706054925918579, | |
| "learning_rate": 8.279952871056263e-07, | |
| "loss": 0.2402, | |
| "step": 290500 | |
| }, | |
| { | |
| "epoch": 4.679434608519466, | |
| "grad_norm": 10.552009582519531, | |
| "learning_rate": 7.881841739001638e-07, | |
| "loss": 0.2435, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 4.687474874169842, | |
| "grad_norm": 0.00723261758685112, | |
| "learning_rate": 7.493385477988724e-07, | |
| "loss": 0.2633, | |
| "step": 291500 | |
| }, | |
| { | |
| "epoch": 4.69551513982022, | |
| "grad_norm": 7.638974189758301, | |
| "learning_rate": 7.114599578367881e-07, | |
| "loss": 0.2448, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 4.703555405470596, | |
| "grad_norm": 8.284395217895508, | |
| "learning_rate": 6.74549914486744e-07, | |
| "loss": 0.2316, | |
| "step": 292500 | |
| }, | |
| { | |
| "epoch": 4.711595671120974, | |
| "grad_norm": 7.808977127075195, | |
| "learning_rate": 6.386098895991455e-07, | |
| "loss": 0.2706, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 4.7196359367713505, | |
| "grad_norm": 2.4569876194000244, | |
| "learning_rate": 6.036413163432702e-07, | |
| "loss": 0.2456, | |
| "step": 293500 | |
| }, | |
| { | |
| "epoch": 4.727676202421728, | |
| "grad_norm": 1.582727074623108, | |
| "learning_rate": 5.69645589150114e-07, | |
| "loss": 0.2483, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 4.735716468072106, | |
| "grad_norm": 4.948225498199463, | |
| "learning_rate": 5.366240636567959e-07, | |
| "loss": 0.2414, | |
| "step": 294500 | |
| }, | |
| { | |
| "epoch": 4.743756733722482, | |
| "grad_norm": 3.3736865520477295, | |
| "learning_rate": 5.045780566524882e-07, | |
| "loss": 0.2608, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 4.751796999372859, | |
| "grad_norm": 26.526477813720703, | |
| "learning_rate": 4.735088460259246e-07, | |
| "loss": 0.2418, | |
| "step": 295500 | |
| }, | |
| { | |
| "epoch": 4.759837265023236, | |
| "grad_norm": 0.4031164050102234, | |
| "learning_rate": 4.434176707144189e-07, | |
| "loss": 0.2514, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 4.767877530673614, | |
| "grad_norm": 2.257049798965454, | |
| "learning_rate": 4.143057306544823e-07, | |
| "loss": 0.2678, | |
| "step": 296500 | |
| }, | |
| { | |
| "epoch": 4.7759177963239905, | |
| "grad_norm": 5.4704437255859375, | |
| "learning_rate": 3.8617418673395353e-07, | |
| "loss": 0.2447, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 4.783958061974368, | |
| "grad_norm": 4.099658966064453, | |
| "learning_rate": 3.590241607457329e-07, | |
| "loss": 0.2455, | |
| "step": 297500 | |
| }, | |
| { | |
| "epoch": 4.791998327624745, | |
| "grad_norm": 1.8913758993148804, | |
| "learning_rate": 3.3285673534301784e-07, | |
| "loss": 0.2395, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 4.800038593275122, | |
| "grad_norm": 1.6071025133132935, | |
| "learning_rate": 3.0767295399615206e-07, | |
| "loss": 0.225, | |
| "step": 298500 | |
| }, | |
| { | |
| "epoch": 4.808078858925499, | |
| "grad_norm": 7.276731014251709, | |
| "learning_rate": 2.834738209510107e-07, | |
| "loss": 0.259, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 4.816119124575876, | |
| "grad_norm": 6.0964179039001465, | |
| "learning_rate": 2.602603011889498e-07, | |
| "loss": 0.272, | |
| "step": 299500 | |
| }, | |
| { | |
| "epoch": 4.824159390226253, | |
| "grad_norm": 0.2317555695772171, | |
| "learning_rate": 2.3803332038832836e-07, | |
| "loss": 0.244, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 4.8321996558766305, | |
| "grad_norm": 2.796255111694336, | |
| "learning_rate": 2.1679376488759894e-07, | |
| "loss": 0.2649, | |
| "step": 300500 | |
| }, | |
| { | |
| "epoch": 4.840239921527007, | |
| "grad_norm": 1.2079377174377441, | |
| "learning_rate": 1.9654248164995836e-07, | |
| "loss": 0.2552, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 4.848280187177385, | |
| "grad_norm": 0.5868381857872009, | |
| "learning_rate": 1.772802782295746e-07, | |
| "loss": 0.2527, | |
| "step": 301500 | |
| }, | |
| { | |
| "epoch": 4.856320452827761, | |
| "grad_norm": 0.9448522925376892, | |
| "learning_rate": 1.5900792273938758e-07, | |
| "loss": 0.2487, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 4.864360718478139, | |
| "grad_norm": 4.876669406890869, | |
| "learning_rate": 1.4172614382047534e-07, | |
| "loss": 0.254, | |
| "step": 302500 | |
| }, | |
| { | |
| "epoch": 4.872400984128515, | |
| "grad_norm": 2.3135619163513184, | |
| "learning_rate": 1.2543563061299668e-07, | |
| "loss": 0.2623, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 4.880441249778893, | |
| "grad_norm": 0.9300447702407837, | |
| "learning_rate": 1.1013703272871878e-07, | |
| "loss": 0.2359, | |
| "step": 303500 | |
| }, | |
| { | |
| "epoch": 4.88848151542927, | |
| "grad_norm": 2.540415048599243, | |
| "learning_rate": 9.583096022511006e-08, | |
| "loss": 0.2412, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 4.896521781079647, | |
| "grad_norm": 1.0760257244110107, | |
| "learning_rate": 8.25179835809986e-08, | |
| "loss": 0.2562, | |
| "step": 304500 | |
| }, | |
| { | |
| "epoch": 4.904562046730024, | |
| "grad_norm": 5.539297103881836, | |
| "learning_rate": 7.019863367385138e-08, | |
| "loss": 0.2516, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 4.912602312380401, | |
| "grad_norm": 1.6742088794708252, | |
| "learning_rate": 5.887340175857736e-08, | |
| "loss": 0.238, | |
| "step": 305500 | |
| }, | |
| { | |
| "epoch": 4.920642578030778, | |
| "grad_norm": 1.7447808980941772, | |
| "learning_rate": 4.854273944795429e-08, | |
| "loss": 0.2592, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 4.928682843681155, | |
| "grad_norm": 1.5578666925430298, | |
| "learning_rate": 3.920705869460972e-08, | |
| "loss": 0.2456, | |
| "step": 306500 | |
| }, | |
| { | |
| "epoch": 4.936723109331532, | |
| "grad_norm": 1.9053585529327393, | |
| "learning_rate": 3.0866731774606414e-08, | |
| "loss": 0.2533, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 4.9447633749819095, | |
| "grad_norm": 4.792218208312988, | |
| "learning_rate": 2.352209127258753e-08, | |
| "loss": 0.2638, | |
| "step": 307500 | |
| }, | |
| { | |
| "epoch": 4.952803640632286, | |
| "grad_norm": 6.906048774719238, | |
| "learning_rate": 1.7173430068509466e-08, | |
| "loss": 0.2232, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 4.960843906282664, | |
| "grad_norm": 8.483711242675781, | |
| "learning_rate": 1.1821001325978965e-08, | |
| "loss": 0.261, | |
| "step": 308500 | |
| }, | |
| { | |
| "epoch": 4.96888417193304, | |
| "grad_norm": 1.2564901113510132, | |
| "learning_rate": 7.465018482141761e-09, | |
| "loss": 0.2672, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 4.976924437583418, | |
| "grad_norm": 1.0019638538360596, | |
| "learning_rate": 4.10565523918105e-09, | |
| "loss": 0.2683, | |
| "step": 309500 | |
| }, | |
| { | |
| "epoch": 4.9849647032337945, | |
| "grad_norm": 1.2965469360351562, | |
| "learning_rate": 1.7430455573896886e-09, | |
| "loss": 0.2355, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 4.993004968884172, | |
| "grad_norm": 2.0296106338500977, | |
| "learning_rate": 3.772836498217025e-10, | |
| "loss": 0.2451, | |
| "step": 310500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_BOD_f1": 0.2719637633292441, | |
| "eval_BUILDING_f1": 0.3313212175990187, | |
| "eval_CARDISSUER_f1": 0.0, | |
| "eval_CITY_f1": 0.27732444210464746, | |
| "eval_COUNTRY_f1": 0.3469542761658818, | |
| "eval_DATE_f1": 0.28033489891770474, | |
| "eval_DRIVERLICENSE_f1": 0.27320312939719704, | |
| "eval_EMAIL_f1": 0.3109130943370261, | |
| "eval_GEOCOORD_f1": 0.2202404032570764, | |
| "eval_GIVENNAME1_f1": 0.2554230207861493, | |
| "eval_GIVENNAME2_f1": 0.19454619454619454, | |
| "eval_IDCARD_f1": 0.28987177640109496, | |
| "eval_IP_f1": 0.2381632935318796, | |
| "eval_LASTNAME1_f1": 0.2539076583741596, | |
| "eval_LASTNAME2_f1": 0.1799779168200221, | |
| "eval_LASTNAME3_f1": 0.16510997288339863, | |
| "eval_PASSPORT_f1": 0.3155746070626659, | |
| "eval_PASS_f1": 0.25141341082088375, | |
| "eval_POSTCODE_f1": 0.29816828288764813, | |
| "eval_SECADDRESS_f1": 0.2719869706840391, | |
| "eval_SEX_f1": 0.3364393027273203, | |
| "eval_SOCIALNUMBER_f1": 0.26947459323663914, | |
| "eval_STATE_f1": 0.3196135654624132, | |
| "eval_STREET_f1": 0.25608687045782313, | |
| "eval_TEL_f1": 0.2732484256304539, | |
| "eval_TIME_f1": 0.31686995934146506, | |
| "eval_TITLE_f1": 0.30542304795380365, | |
| "eval_USERNAME_f1": 0.30195064919081904, | |
| "eval_loss": 0.38949722051620483, | |
| "eval_overall_accuracy": 0.8744043928876162, | |
| "eval_overall_f1": 0.28616190289433646, | |
| "eval_overall_precision": 0.3091229151279809, | |
| "eval_overall_recall": 0.2663760403102531, | |
| "eval_runtime": 652.7121, | |
| "eval_samples_per_second": 81.665, | |
| "eval_steps_per_second": 40.833, | |
| "step": 310935 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 310935, | |
| "total_flos": 4.090026161830714e+16, | |
| "train_loss": 0.39710686157782077, | |
| "train_runtime": 12879.2811, | |
| "train_samples_per_second": 48.284, | |
| "train_steps_per_second": 24.142 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_BOD_f1": 0.2374555224747658, | |
| "eval_BUILDING_f1": 0.28714220808723306, | |
| "eval_CARDISSUER_f1": 0.0, | |
| "eval_CITY_f1": 0.25397850725233423, | |
| "eval_COUNTRY_f1": 0.3055481815768516, | |
| "eval_DATE_f1": 0.2341297523543774, | |
| "eval_DRIVERLICENSE_f1": 0.2233434378369581, | |
| "eval_EMAIL_f1": 0.2653814157051398, | |
| "eval_GEOCOORD_f1": 0.16032388663967612, | |
| "eval_GIVENNAME1_f1": 0.21614488277411203, | |
| "eval_GIVENNAME2_f1": 0.15067155067155066, | |
| "eval_IDCARD_f1": 0.24722656398917311, | |
| "eval_IP_f1": 0.18513513513513513, | |
| "eval_LASTNAME1_f1": 0.22957798483959949, | |
| "eval_LASTNAME2_f1": 0.13049962714392246, | |
| "eval_LASTNAME3_f1": 0.12451593684837652, | |
| "eval_PASSPORT_f1": 0.27920583363806495, | |
| "eval_PASS_f1": 0.19796215429403202, | |
| "eval_POSTCODE_f1": 0.2794223307946544, | |
| "eval_SECADDRESS_f1": 0.2486166906692943, | |
| "eval_SEX_f1": 0.29331175212440025, | |
| "eval_SOCIALNUMBER_f1": 0.22580711200278417, | |
| "eval_STATE_f1": 0.29213965980304385, | |
| "eval_STREET_f1": 0.21770362646275057, | |
| "eval_TEL_f1": 0.24091208570855654, | |
| "eval_TIME_f1": 0.28926309929719574, | |
| "eval_TITLE_f1": 0.2814272766943037, | |
| "eval_USERNAME_f1": 0.23682905559070463, | |
| "eval_loss": 0.36858755350112915, | |
| "eval_overall_accuracy": 0.8688154462267319, | |
| "eval_overall_f1": 0.24593334983551945, | |
| "eval_overall_precision": 0.2884671977802535, | |
| "eval_overall_recall": 0.2143307557001095, | |
| "eval_runtime": 654.0487, | |
| "eval_samples_per_second": 81.499, | |
| "eval_steps_per_second": 40.749, | |
| "step": 310935 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 310935, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "total_flos": 4.090026161830714e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |