Invalid JSON:Unexpected token 'I', ..."ad_norm": Infinity,
"... is not valid JSON
| { | |
| "best_global_step": 344910, | |
| "best_metric": 0.9034274548462476, | |
| "best_model_checkpoint": "T:\\laupodteam\\AIOS\\Bram\\language_modeling\\Models\\language_models\\MutliClinAI\\XLMR\\multilabel_3ldense_10epochs_40splits/fold_0\\checkpoint-344910", | |
| "epoch": 10.0, | |
| "eval_steps": 500, | |
| "global_step": 344910, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.007422226087964976, | |
| "grad_norm": 0.7518407702445984, | |
| "learning_rate": 9.992606766982692e-06, | |
| "loss": 0.5412131547927856, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.014844452175929952, | |
| "grad_norm": 0.475884348154068, | |
| "learning_rate": 9.985184540894727e-06, | |
| "loss": 0.47313085198402405, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.02226667826389493, | |
| "grad_norm": 0.20832128822803497, | |
| "learning_rate": 9.977762314806761e-06, | |
| "loss": 0.47467106580734253, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.029688904351859904, | |
| "grad_norm": 0.8777472376823425, | |
| "learning_rate": 9.970340088718798e-06, | |
| "loss": 0.4484080374240875, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.037111130439824884, | |
| "grad_norm": 0.5118074417114258, | |
| "learning_rate": 9.962917862630833e-06, | |
| "loss": 0.401199609041214, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.04453335652778986, | |
| "grad_norm": 1.3781318664550781, | |
| "learning_rate": 9.955495636542867e-06, | |
| "loss": 0.38673847913742065, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 0.051955582615754836, | |
| "grad_norm": 1.5291341543197632, | |
| "learning_rate": 9.948073410454902e-06, | |
| "loss": 0.37920689582824707, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 0.05937780870371981, | |
| "grad_norm": 2.174276113510132, | |
| "learning_rate": 9.940651184366937e-06, | |
| "loss": 0.37507209181785583, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 0.06680003479168478, | |
| "grad_norm": 0.9687999486923218, | |
| "learning_rate": 9.933228958278971e-06, | |
| "loss": 0.36908626556396484, | |
| "step": 2304 | |
| }, | |
| { | |
| "epoch": 0.07422226087964977, | |
| "grad_norm": 1.1717259883880615, | |
| "learning_rate": 9.925806732191006e-06, | |
| "loss": 0.36402738094329834, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.08164448696761474, | |
| "grad_norm": 1.0056860446929932, | |
| "learning_rate": 9.918384506103043e-06, | |
| "loss": 0.3641241788864136, | |
| "step": 2816 | |
| }, | |
| { | |
| "epoch": 0.08906671305557971, | |
| "grad_norm": 1.2744220495224, | |
| "learning_rate": 9.910962280015077e-06, | |
| "loss": 0.3602203130722046, | |
| "step": 3072 | |
| }, | |
| { | |
| "epoch": 0.0964889391435447, | |
| "grad_norm": 3.087216854095459, | |
| "learning_rate": 9.903540053927112e-06, | |
| "loss": 0.35854455828666687, | |
| "step": 3328 | |
| }, | |
| { | |
| "epoch": 0.10391116523150967, | |
| "grad_norm": 1.2993056774139404, | |
| "learning_rate": 9.896117827839147e-06, | |
| "loss": 0.3533003032207489, | |
| "step": 3584 | |
| }, | |
| { | |
| "epoch": 0.11133339131947464, | |
| "grad_norm": 3.572425365447998, | |
| "learning_rate": 9.888695601751183e-06, | |
| "loss": 0.34851548075675964, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.11875561740743962, | |
| "grad_norm": 0.9699921011924744, | |
| "learning_rate": 9.881273375663218e-06, | |
| "loss": 0.34326642751693726, | |
| "step": 4096 | |
| }, | |
| { | |
| "epoch": 0.1261778434954046, | |
| "grad_norm": 0.7386257648468018, | |
| "learning_rate": 9.873851149575252e-06, | |
| "loss": 0.34405675530433655, | |
| "step": 4352 | |
| }, | |
| { | |
| "epoch": 0.13360006958336956, | |
| "grad_norm": 1.7287063598632812, | |
| "learning_rate": 9.866428923487287e-06, | |
| "loss": 0.339326947927475, | |
| "step": 4608 | |
| }, | |
| { | |
| "epoch": 0.14102229567133456, | |
| "grad_norm": 2.6771416664123535, | |
| "learning_rate": 9.859006697399322e-06, | |
| "loss": 0.33704981207847595, | |
| "step": 4864 | |
| }, | |
| { | |
| "epoch": 0.14844452175929954, | |
| "grad_norm": 2.183525800704956, | |
| "learning_rate": 9.851584471311357e-06, | |
| "loss": 0.3340759575366974, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.1558667478472645, | |
| "grad_norm": 1.3297194242477417, | |
| "learning_rate": 9.844162245223391e-06, | |
| "loss": 0.33215370774269104, | |
| "step": 5376 | |
| }, | |
| { | |
| "epoch": 0.16328897393522948, | |
| "grad_norm": 1.9320464134216309, | |
| "learning_rate": 9.836740019135428e-06, | |
| "loss": 0.3280167579650879, | |
| "step": 5632 | |
| }, | |
| { | |
| "epoch": 0.17071120002319445, | |
| "grad_norm": 1.6145955324172974, | |
| "learning_rate": 9.829317793047462e-06, | |
| "loss": 0.32415950298309326, | |
| "step": 5888 | |
| }, | |
| { | |
| "epoch": 0.17813342611115943, | |
| "grad_norm": 1.8750313520431519, | |
| "learning_rate": 9.821895566959497e-06, | |
| "loss": 0.32567980885505676, | |
| "step": 6144 | |
| }, | |
| { | |
| "epoch": 0.1855556521991244, | |
| "grad_norm": 1.6577664613723755, | |
| "learning_rate": 9.814473340871532e-06, | |
| "loss": 0.3212590217590332, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.1929778782870894, | |
| "grad_norm": 0.6327639222145081, | |
| "learning_rate": 9.807051114783568e-06, | |
| "loss": 0.3169021010398865, | |
| "step": 6656 | |
| }, | |
| { | |
| "epoch": 0.20040010437505437, | |
| "grad_norm": 1.0281022787094116, | |
| "learning_rate": 9.799628888695603e-06, | |
| "loss": 0.31635186076164246, | |
| "step": 6912 | |
| }, | |
| { | |
| "epoch": 0.20782233046301934, | |
| "grad_norm": 1.4524767398834229, | |
| "learning_rate": 9.792206662607638e-06, | |
| "loss": 0.30951330065727234, | |
| "step": 7168 | |
| }, | |
| { | |
| "epoch": 0.21524455655098432, | |
| "grad_norm": 1.2955704927444458, | |
| "learning_rate": 9.784784436519674e-06, | |
| "loss": 0.3031282424926758, | |
| "step": 7424 | |
| }, | |
| { | |
| "epoch": 0.2226667826389493, | |
| "grad_norm": 3.3441994190216064, | |
| "learning_rate": 9.777362210431709e-06, | |
| "loss": 0.3032401502132416, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.23008900872691426, | |
| "grad_norm": 1.418245553970337, | |
| "learning_rate": 9.769939984343743e-06, | |
| "loss": 0.2950679361820221, | |
| "step": 7936 | |
| }, | |
| { | |
| "epoch": 0.23751123481487924, | |
| "grad_norm": 1.6895215511322021, | |
| "learning_rate": 9.762517758255778e-06, | |
| "loss": 0.28757432103157043, | |
| "step": 8192 | |
| }, | |
| { | |
| "epoch": 0.2449334609028442, | |
| "grad_norm": 1.1834076642990112, | |
| "learning_rate": 9.755095532167813e-06, | |
| "loss": 0.27872321009635925, | |
| "step": 8448 | |
| }, | |
| { | |
| "epoch": 0.2523556869908092, | |
| "grad_norm": 1.086266279220581, | |
| "learning_rate": 9.747702299150504e-06, | |
| "loss": 0.27419590950012207, | |
| "step": 8704 | |
| }, | |
| { | |
| "epoch": 0.2597779130787742, | |
| "grad_norm": 0.9157214164733887, | |
| "learning_rate": 9.740280073062539e-06, | |
| "loss": 0.271694153547287, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 0.2672001391667391, | |
| "grad_norm": 3.2456769943237305, | |
| "learning_rate": 9.732857846974573e-06, | |
| "loss": 0.2656308114528656, | |
| "step": 9216 | |
| }, | |
| { | |
| "epoch": 0.2746223652547041, | |
| "grad_norm": 1.0417124032974243, | |
| "learning_rate": 9.725435620886608e-06, | |
| "loss": 0.2627953886985779, | |
| "step": 9472 | |
| }, | |
| { | |
| "epoch": 0.2820445913426691, | |
| "grad_norm": 1.687821626663208, | |
| "learning_rate": 9.718013394798645e-06, | |
| "loss": 0.2567230463027954, | |
| "step": 9728 | |
| }, | |
| { | |
| "epoch": 0.28946681743063407, | |
| "grad_norm": 1.360467553138733, | |
| "learning_rate": 9.71059116871068e-06, | |
| "loss": 0.2528173625469208, | |
| "step": 9984 | |
| }, | |
| { | |
| "epoch": 0.29688904351859907, | |
| "grad_norm": 1.376082420349121, | |
| "learning_rate": 9.703168942622714e-06, | |
| "loss": 0.2498796284198761, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.304311269606564, | |
| "grad_norm": 1.0839852094650269, | |
| "learning_rate": 9.695746716534749e-06, | |
| "loss": 0.24654100835323334, | |
| "step": 10496 | |
| }, | |
| { | |
| "epoch": 0.311733495694529, | |
| "grad_norm": 1.2203876972198486, | |
| "learning_rate": 9.68835348351744e-06, | |
| "loss": 0.24533048272132874, | |
| "step": 10752 | |
| }, | |
| { | |
| "epoch": 0.31915572178249396, | |
| "grad_norm": 0.7553607821464539, | |
| "learning_rate": 9.680931257429476e-06, | |
| "loss": 0.24138209223747253, | |
| "step": 11008 | |
| }, | |
| { | |
| "epoch": 0.32657794787045896, | |
| "grad_norm": 1.5078504085540771, | |
| "learning_rate": 9.673509031341511e-06, | |
| "loss": 0.2399892508983612, | |
| "step": 11264 | |
| }, | |
| { | |
| "epoch": 0.33400017395842396, | |
| "grad_norm": 0.7138952016830444, | |
| "learning_rate": 9.666086805253546e-06, | |
| "loss": 0.23641596734523773, | |
| "step": 11520 | |
| }, | |
| { | |
| "epoch": 0.3414224000463889, | |
| "grad_norm": 1.7338546514511108, | |
| "learning_rate": 9.65866457916558e-06, | |
| "loss": 0.23360668122768402, | |
| "step": 11776 | |
| }, | |
| { | |
| "epoch": 0.3488446261343539, | |
| "grad_norm": 1.3493647575378418, | |
| "learning_rate": 9.651242353077615e-06, | |
| "loss": 0.23118990659713745, | |
| "step": 12032 | |
| }, | |
| { | |
| "epoch": 0.35626685222231885, | |
| "grad_norm": 1.0640701055526733, | |
| "learning_rate": 9.64382012698965e-06, | |
| "loss": 0.2264977991580963, | |
| "step": 12288 | |
| }, | |
| { | |
| "epoch": 0.36368907831028385, | |
| "grad_norm": 2.106741189956665, | |
| "learning_rate": 9.636397900901685e-06, | |
| "loss": 0.2229805737733841, | |
| "step": 12544 | |
| }, | |
| { | |
| "epoch": 0.3711113043982488, | |
| "grad_norm": 1.5093495845794678, | |
| "learning_rate": 9.629004667884376e-06, | |
| "loss": 0.22241438925266266, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.3785335304862138, | |
| "grad_norm": 0.7995802760124207, | |
| "learning_rate": 9.62158244179641e-06, | |
| "loss": 0.22048331797122955, | |
| "step": 13056 | |
| }, | |
| { | |
| "epoch": 0.3859557565741788, | |
| "grad_norm": 1.934958815574646, | |
| "learning_rate": 9.614160215708447e-06, | |
| "loss": 0.2158215045928955, | |
| "step": 13312 | |
| }, | |
| { | |
| "epoch": 0.39337798266214374, | |
| "grad_norm": 1.0992568731307983, | |
| "learning_rate": 9.606737989620482e-06, | |
| "loss": 0.215406134724617, | |
| "step": 13568 | |
| }, | |
| { | |
| "epoch": 0.40080020875010874, | |
| "grad_norm": 0.8470885753631592, | |
| "learning_rate": 9.599315763532516e-06, | |
| "loss": 0.2131994664669037, | |
| "step": 13824 | |
| }, | |
| { | |
| "epoch": 0.4082224348380737, | |
| "grad_norm": 1.1812682151794434, | |
| "learning_rate": 9.591893537444551e-06, | |
| "loss": 0.2090103030204773, | |
| "step": 14080 | |
| }, | |
| { | |
| "epoch": 0.4156446609260387, | |
| "grad_norm": 2.277434825897217, | |
| "learning_rate": 9.584471311356587e-06, | |
| "loss": 0.20873476564884186, | |
| "step": 14336 | |
| }, | |
| { | |
| "epoch": 0.42306688701400363, | |
| "grad_norm": 0.7682158350944519, | |
| "learning_rate": 9.577049085268622e-06, | |
| "loss": 0.20455653965473175, | |
| "step": 14592 | |
| }, | |
| { | |
| "epoch": 0.43048911310196863, | |
| "grad_norm": 0.9997185468673706, | |
| "learning_rate": 9.569655852251313e-06, | |
| "loss": 0.2045592963695526, | |
| "step": 14848 | |
| }, | |
| { | |
| "epoch": 0.4379113391899336, | |
| "grad_norm": 0.9479852914810181, | |
| "learning_rate": 9.562233626163348e-06, | |
| "loss": 0.2046046257019043, | |
| "step": 15104 | |
| }, | |
| { | |
| "epoch": 0.4453335652778986, | |
| "grad_norm": 1.4512850046157837, | |
| "learning_rate": 9.554811400075383e-06, | |
| "loss": 0.20438788831233978, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.4527557913658636, | |
| "grad_norm": 1.3652234077453613, | |
| "learning_rate": 9.547389173987418e-06, | |
| "loss": 0.1992088109254837, | |
| "step": 15616 | |
| }, | |
| { | |
| "epoch": 0.4601780174538285, | |
| "grad_norm": 2.078918695449829, | |
| "learning_rate": 9.539966947899452e-06, | |
| "loss": 0.19647617638111115, | |
| "step": 15872 | |
| }, | |
| { | |
| "epoch": 0.4676002435417935, | |
| "grad_norm": 2.0355148315429688, | |
| "learning_rate": 9.532544721811487e-06, | |
| "loss": 0.19802288711071014, | |
| "step": 16128 | |
| }, | |
| { | |
| "epoch": 0.47502246962975847, | |
| "grad_norm": 0.8739069104194641, | |
| "learning_rate": 9.525122495723523e-06, | |
| "loss": 0.1988070160150528, | |
| "step": 16384 | |
| }, | |
| { | |
| "epoch": 0.48244469571772347, | |
| "grad_norm": 0.9023852944374084, | |
| "learning_rate": 9.517700269635558e-06, | |
| "loss": 0.19897446036338806, | |
| "step": 16640 | |
| }, | |
| { | |
| "epoch": 0.4898669218056884, | |
| "grad_norm": 2.5597965717315674, | |
| "learning_rate": 9.510278043547593e-06, | |
| "loss": 0.19377563893795013, | |
| "step": 16896 | |
| }, | |
| { | |
| "epoch": 0.4972891478936534, | |
| "grad_norm": 1.7188621759414673, | |
| "learning_rate": 9.502884810530284e-06, | |
| "loss": 0.19135312736034393, | |
| "step": 17152 | |
| }, | |
| { | |
| "epoch": 0.5047113739816184, | |
| "grad_norm": 0.9326071739196777, | |
| "learning_rate": 9.495462584442319e-06, | |
| "loss": 0.19048641622066498, | |
| "step": 17408 | |
| }, | |
| { | |
| "epoch": 0.5121336000695834, | |
| "grad_norm": 1.098150610923767, | |
| "learning_rate": 9.488040358354355e-06, | |
| "loss": 0.18813419342041016, | |
| "step": 17664 | |
| }, | |
| { | |
| "epoch": 0.5195558261575484, | |
| "grad_norm": 2.041393995285034, | |
| "learning_rate": 9.48061813226639e-06, | |
| "loss": 0.1874156892299652, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.5269780522455133, | |
| "grad_norm": 1.4699207544326782, | |
| "learning_rate": 9.473195906178424e-06, | |
| "loss": 0.18520450592041016, | |
| "step": 18176 | |
| }, | |
| { | |
| "epoch": 0.5344002783334783, | |
| "grad_norm": 1.014280915260315, | |
| "learning_rate": 9.46577368009046e-06, | |
| "loss": 0.18690264225006104, | |
| "step": 18432 | |
| }, | |
| { | |
| "epoch": 0.5418225044214433, | |
| "grad_norm": 0.9460366368293762, | |
| "learning_rate": 9.458351454002494e-06, | |
| "loss": 0.18374530971050262, | |
| "step": 18688 | |
| }, | |
| { | |
| "epoch": 0.5492447305094083, | |
| "grad_norm": 1.106070637702942, | |
| "learning_rate": 9.450929227914529e-06, | |
| "loss": 0.18379539251327515, | |
| "step": 18944 | |
| }, | |
| { | |
| "epoch": 0.5566669565973732, | |
| "grad_norm": 2.079352378845215, | |
| "learning_rate": 9.443507001826563e-06, | |
| "loss": 0.18233495950698853, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.5640891826853383, | |
| "grad_norm": 2.3012380599975586, | |
| "learning_rate": 9.436113768809255e-06, | |
| "loss": 0.1818738728761673, | |
| "step": 19456 | |
| }, | |
| { | |
| "epoch": 0.5715114087733032, | |
| "grad_norm": 1.2030616998672485, | |
| "learning_rate": 9.42869154272129e-06, | |
| "loss": 0.18131691217422485, | |
| "step": 19712 | |
| }, | |
| { | |
| "epoch": 0.5789336348612681, | |
| "grad_norm": 0.8204110860824585, | |
| "learning_rate": 9.421269316633326e-06, | |
| "loss": 0.1797848343849182, | |
| "step": 19968 | |
| }, | |
| { | |
| "epoch": 0.5863558609492331, | |
| "grad_norm": 0.9082216620445251, | |
| "learning_rate": 9.41384709054536e-06, | |
| "loss": 0.17948311567306519, | |
| "step": 20224 | |
| }, | |
| { | |
| "epoch": 0.5937780870371981, | |
| "grad_norm": 1.4475148916244507, | |
| "learning_rate": 9.406424864457395e-06, | |
| "loss": 0.17923176288604736, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 0.6012003131251631, | |
| "grad_norm": 0.8240416049957275, | |
| "learning_rate": 9.39900263836943e-06, | |
| "loss": 0.17499788105487823, | |
| "step": 20736 | |
| }, | |
| { | |
| "epoch": 0.608622539213128, | |
| "grad_norm": 1.1081963777542114, | |
| "learning_rate": 9.391580412281466e-06, | |
| "loss": 0.17557330429553986, | |
| "step": 20992 | |
| }, | |
| { | |
| "epoch": 0.6160447653010931, | |
| "grad_norm": 1.480975866317749, | |
| "learning_rate": 9.384158186193501e-06, | |
| "loss": 0.17561376094818115, | |
| "step": 21248 | |
| }, | |
| { | |
| "epoch": 0.623466991389058, | |
| "grad_norm": 2.3068583011627197, | |
| "learning_rate": 9.376764953176192e-06, | |
| "loss": 0.17281724512577057, | |
| "step": 21504 | |
| }, | |
| { | |
| "epoch": 0.630889217477023, | |
| "grad_norm": 2.04187273979187, | |
| "learning_rate": 9.369342727088227e-06, | |
| "loss": 0.17577329277992249, | |
| "step": 21760 | |
| }, | |
| { | |
| "epoch": 0.6383114435649879, | |
| "grad_norm": 1.9467560052871704, | |
| "learning_rate": 9.361920501000262e-06, | |
| "loss": 0.17350482940673828, | |
| "step": 22016 | |
| }, | |
| { | |
| "epoch": 0.645733669652953, | |
| "grad_norm": 0.799000084400177, | |
| "learning_rate": 9.354498274912296e-06, | |
| "loss": 0.17236289381980896, | |
| "step": 22272 | |
| }, | |
| { | |
| "epoch": 0.6531558957409179, | |
| "grad_norm": 0.8497450351715088, | |
| "learning_rate": 9.347076048824331e-06, | |
| "loss": 0.1705632209777832, | |
| "step": 22528 | |
| }, | |
| { | |
| "epoch": 0.6605781218288829, | |
| "grad_norm": 1.250152349472046, | |
| "learning_rate": 9.339653822736366e-06, | |
| "loss": 0.17060533165931702, | |
| "step": 22784 | |
| }, | |
| { | |
| "epoch": 0.6680003479168479, | |
| "grad_norm": 1.1720199584960938, | |
| "learning_rate": 9.332231596648402e-06, | |
| "loss": 0.17060840129852295, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 0.6754225740048129, | |
| "grad_norm": 0.8918880224227905, | |
| "learning_rate": 9.324809370560437e-06, | |
| "loss": 0.16725586354732513, | |
| "step": 23296 | |
| }, | |
| { | |
| "epoch": 0.6828448000927778, | |
| "grad_norm": 2.859825611114502, | |
| "learning_rate": 9.317416137543128e-06, | |
| "loss": 0.16767054796218872, | |
| "step": 23552 | |
| }, | |
| { | |
| "epoch": 0.6902670261807428, | |
| "grad_norm": 1.0053714513778687, | |
| "learning_rate": 9.309993911455163e-06, | |
| "loss": 0.16494125127792358, | |
| "step": 23808 | |
| }, | |
| { | |
| "epoch": 0.6976892522687078, | |
| "grad_norm": 1.7721636295318604, | |
| "learning_rate": 9.302571685367197e-06, | |
| "loss": 0.166941300034523, | |
| "step": 24064 | |
| }, | |
| { | |
| "epoch": 0.7051114783566728, | |
| "grad_norm": 2.37420392036438, | |
| "learning_rate": 9.295149459279234e-06, | |
| "loss": 0.1632576286792755, | |
| "step": 24320 | |
| }, | |
| { | |
| "epoch": 0.7125337044446377, | |
| "grad_norm": 1.2416902780532837, | |
| "learning_rate": 9.287727233191269e-06, | |
| "loss": 0.16529348492622375, | |
| "step": 24576 | |
| }, | |
| { | |
| "epoch": 0.7199559305326028, | |
| "grad_norm": 1.5650802850723267, | |
| "learning_rate": 9.280305007103303e-06, | |
| "loss": 0.1623227745294571, | |
| "step": 24832 | |
| }, | |
| { | |
| "epoch": 0.7273781566205677, | |
| "grad_norm": 0.8756684064865112, | |
| "learning_rate": 9.272882781015338e-06, | |
| "loss": 0.1634257286787033, | |
| "step": 25088 | |
| }, | |
| { | |
| "epoch": 0.7348003827085327, | |
| "grad_norm": 0.762071967124939, | |
| "learning_rate": 9.265460554927374e-06, | |
| "loss": 0.16134406626224518, | |
| "step": 25344 | |
| }, | |
| { | |
| "epoch": 0.7422226087964976, | |
| "grad_norm": 1.2213808298110962, | |
| "learning_rate": 9.258038328839409e-06, | |
| "loss": 0.160996675491333, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.7496448348844627, | |
| "grad_norm": 1.087438941001892, | |
| "learning_rate": 9.250645095822099e-06, | |
| "loss": 0.1597295105457306, | |
| "step": 25856 | |
| }, | |
| { | |
| "epoch": 0.7570670609724276, | |
| "grad_norm": 1.2111320495605469, | |
| "learning_rate": 9.243222869734135e-06, | |
| "loss": 0.16009105741977692, | |
| "step": 26112 | |
| }, | |
| { | |
| "epoch": 0.7644892870603925, | |
| "grad_norm": 1.8586764335632324, | |
| "learning_rate": 9.23580064364617e-06, | |
| "loss": 0.160374253988266, | |
| "step": 26368 | |
| }, | |
| { | |
| "epoch": 0.7719115131483576, | |
| "grad_norm": 1.5776656866073608, | |
| "learning_rate": 9.228378417558204e-06, | |
| "loss": 0.1620863527059555, | |
| "step": 26624 | |
| }, | |
| { | |
| "epoch": 0.7793337392363225, | |
| "grad_norm": 0.9666963219642639, | |
| "learning_rate": 9.220956191470239e-06, | |
| "loss": 0.1561657339334488, | |
| "step": 26880 | |
| }, | |
| { | |
| "epoch": 0.7867559653242875, | |
| "grad_norm": 1.9027835130691528, | |
| "learning_rate": 9.213533965382274e-06, | |
| "loss": 0.15570582449436188, | |
| "step": 27136 | |
| }, | |
| { | |
| "epoch": 0.7941781914122524, | |
| "grad_norm": 0.8957810997962952, | |
| "learning_rate": 9.206111739294309e-06, | |
| "loss": 0.1572333723306656, | |
| "step": 27392 | |
| }, | |
| { | |
| "epoch": 0.8016004175002175, | |
| "grad_norm": 2.0321531295776367, | |
| "learning_rate": 9.198689513206345e-06, | |
| "loss": 0.15403559803962708, | |
| "step": 27648 | |
| }, | |
| { | |
| "epoch": 0.8090226435881824, | |
| "grad_norm": 1.0518842935562134, | |
| "learning_rate": 9.19126728711838e-06, | |
| "loss": 0.15894752740859985, | |
| "step": 27904 | |
| }, | |
| { | |
| "epoch": 0.8164448696761474, | |
| "grad_norm": 0.8605303764343262, | |
| "learning_rate": 9.183845061030414e-06, | |
| "loss": 0.15891948342323303, | |
| "step": 28160 | |
| }, | |
| { | |
| "epoch": 0.8238670957641123, | |
| "grad_norm": 0.9070032835006714, | |
| "learning_rate": 9.17642283494245e-06, | |
| "loss": 0.15218831598758698, | |
| "step": 28416 | |
| }, | |
| { | |
| "epoch": 0.8312893218520774, | |
| "grad_norm": 1.3426278829574585, | |
| "learning_rate": 9.169000608854485e-06, | |
| "loss": 0.15455909073352814, | |
| "step": 28672 | |
| }, | |
| { | |
| "epoch": 0.8387115479400423, | |
| "grad_norm": 1.940657615661621, | |
| "learning_rate": 9.16157838276652e-06, | |
| "loss": 0.15511828660964966, | |
| "step": 28928 | |
| }, | |
| { | |
| "epoch": 0.8461337740280073, | |
| "grad_norm": 1.2703078985214233, | |
| "learning_rate": 9.154185149749211e-06, | |
| "loss": 0.1547073870897293, | |
| "step": 29184 | |
| }, | |
| { | |
| "epoch": 0.8535560001159723, | |
| "grad_norm": 1.8077315092086792, | |
| "learning_rate": 9.146762923661246e-06, | |
| "loss": 0.15338259935379028, | |
| "step": 29440 | |
| }, | |
| { | |
| "epoch": 0.8609782262039373, | |
| "grad_norm": 2.1672475337982178, | |
| "learning_rate": 9.13934069757328e-06, | |
| "loss": 0.15254677832126617, | |
| "step": 29696 | |
| }, | |
| { | |
| "epoch": 0.8684004522919022, | |
| "grad_norm": 1.1730506420135498, | |
| "learning_rate": 9.131918471485316e-06, | |
| "loss": 0.15243162214756012, | |
| "step": 29952 | |
| }, | |
| { | |
| "epoch": 0.8758226783798672, | |
| "grad_norm": 2.8621113300323486, | |
| "learning_rate": 9.12449624539735e-06, | |
| "loss": 0.15081022679805756, | |
| "step": 30208 | |
| }, | |
| { | |
| "epoch": 0.8832449044678322, | |
| "grad_norm": 0.8935790061950684, | |
| "learning_rate": 9.117074019309385e-06, | |
| "loss": 0.1498076468706131, | |
| "step": 30464 | |
| }, | |
| { | |
| "epoch": 0.8906671305557972, | |
| "grad_norm": 1.3302828073501587, | |
| "learning_rate": 9.109651793221421e-06, | |
| "loss": 0.15235257148742676, | |
| "step": 30720 | |
| }, | |
| { | |
| "epoch": 0.8980893566437621, | |
| "grad_norm": 1.384374737739563, | |
| "learning_rate": 9.102229567133456e-06, | |
| "loss": 0.14965111017227173, | |
| "step": 30976 | |
| }, | |
| { | |
| "epoch": 0.9055115827317272, | |
| "grad_norm": 1.2355798482894897, | |
| "learning_rate": 9.094836334116147e-06, | |
| "loss": 0.14785613119602203, | |
| "step": 31232 | |
| }, | |
| { | |
| "epoch": 0.9129338088196921, | |
| "grad_norm": 1.1680980920791626, | |
| "learning_rate": 9.087414108028182e-06, | |
| "loss": 0.14858569204807281, | |
| "step": 31488 | |
| }, | |
| { | |
| "epoch": 0.920356034907657, | |
| "grad_norm": 0.706349790096283, | |
| "learning_rate": 9.079991881940217e-06, | |
| "loss": 0.14675690233707428, | |
| "step": 31744 | |
| }, | |
| { | |
| "epoch": 0.927778260995622, | |
| "grad_norm": 1.825018286705017, | |
| "learning_rate": 9.072569655852253e-06, | |
| "loss": 0.14775897562503815, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.935200487083587, | |
| "grad_norm": 1.093258023262024, | |
| "learning_rate": 9.065147429764288e-06, | |
| "loss": 0.14692749083042145, | |
| "step": 32256 | |
| }, | |
| { | |
| "epoch": 0.942622713171552, | |
| "grad_norm": 2.0166993141174316, | |
| "learning_rate": 9.057725203676323e-06, | |
| "loss": 0.1459830105304718, | |
| "step": 32512 | |
| }, | |
| { | |
| "epoch": 0.9500449392595169, | |
| "grad_norm": 1.315467119216919, | |
| "learning_rate": 9.050302977588357e-06, | |
| "loss": 0.14377714693546295, | |
| "step": 32768 | |
| }, | |
| { | |
| "epoch": 0.957467165347482, | |
| "grad_norm": 1.1817374229431152, | |
| "learning_rate": 9.042880751500392e-06, | |
| "loss": 0.14771975576877594, | |
| "step": 33024 | |
| }, | |
| { | |
| "epoch": 0.9648893914354469, | |
| "grad_norm": 2.038198709487915, | |
| "learning_rate": 9.035487518483083e-06, | |
| "loss": 0.14584626257419586, | |
| "step": 33280 | |
| }, | |
| { | |
| "epoch": 0.9723116175234119, | |
| "grad_norm": 2.3715620040893555, | |
| "learning_rate": 9.028065292395118e-06, | |
| "loss": 0.1448163092136383, | |
| "step": 33536 | |
| }, | |
| { | |
| "epoch": 0.9797338436113768, | |
| "grad_norm": 1.5495351552963257, | |
| "learning_rate": 9.020643066307153e-06, | |
| "loss": 0.14230488240718842, | |
| "step": 33792 | |
| }, | |
| { | |
| "epoch": 0.9871560696993419, | |
| "grad_norm": 0.9654718041419983, | |
| "learning_rate": 9.013220840219187e-06, | |
| "loss": 0.14520296454429626, | |
| "step": 34048 | |
| }, | |
| { | |
| "epoch": 0.9945782957873068, | |
| "grad_norm": 2.4293212890625, | |
| "learning_rate": 9.005798614131224e-06, | |
| "loss": 0.1460076868534088, | |
| "step": 34304 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_f1_B-DISEASE": 0.7824379295833918, | |
| "eval_f1_B-PROCEDURE": 0.7919397979406018, | |
| "eval_f1_B-SYMPTOM": 0.7630874921557618, | |
| "eval_f1_I-DISEASE": 0.8230338254097408, | |
| "eval_f1_I-PROCEDURE": 0.8305460246767706, | |
| "eval_f1_I-SYMPTOM": 0.7860246977324555, | |
| "eval_f1_O": 0.9140864034920458, | |
| "eval_f1_macro": 0.8130223101415383, | |
| "eval_f1_micro": 0.8665886842202686, | |
| "eval_loss": 0.07880550622940063, | |
| "eval_precision_B-DISEASE": 0.8287768929038393, | |
| "eval_precision_B-PROCEDURE": 0.8298576822538484, | |
| "eval_precision_B-SYMPTOM": 0.7723818226392387, | |
| "eval_precision_I-DISEASE": 0.8314419915453265, | |
| "eval_precision_I-PROCEDURE": 0.8824175558881754, | |
| "eval_precision_I-SYMPTOM": 0.7535474645253547, | |
| "eval_precision_O": 0.9323725077547601, | |
| "eval_precision_macro": 0.8329708453586491, | |
| "eval_precision_micro": 0.8786030281318118, | |
| "eval_rauc_macro": 0.8844591022652377, | |
| "eval_rauc_micro": 0.9172522355553351, | |
| "eval_recall_B-DISEASE": 0.7410064296721398, | |
| "eval_recall_B-PROCEDURE": 0.7573355951970737, | |
| "eval_recall_B-SYMPTOM": 0.7540141856316087, | |
| "eval_recall_I-DISEASE": 0.8147940161104718, | |
| "eval_recall_I-PROCEDURE": 0.7844342862048245, | |
| "eval_recall_I-SYMPTOM": 0.8214275008243652, | |
| "eval_recall_O": 0.8965037728104465, | |
| "eval_recall_macro": 0.7956451123501329, | |
| "eval_recall_micro": 0.8548984850293558, | |
| "eval_roc_auc_B-DISEASE": 0.8688497493020165, | |
| "eval_roc_auc_B-PROCEDURE": 0.8769866980021385, | |
| "eval_roc_auc_B-SYMPTOM": 0.8740622985534965, | |
| "eval_roc_auc_I-DISEASE": 0.8959156116543142, | |
| "eval_roc_auc_I-PROCEDURE": 0.8861047787575571, | |
| "eval_roc_auc_I-SYMPTOM": 0.8870230407945396, | |
| "eval_roc_auc_O": 0.9022715387926017, | |
| "eval_runtime": 50.1911, | |
| "eval_samples_per_second": 281.922, | |
| "eval_steps_per_second": 35.245, | |
| "step": 34491 | |
| }, | |
| { | |
| "epoch": 1.0020005218752719, | |
| "grad_norm": 1.3597321510314941, | |
| "learning_rate": 8.998376388043258e-06, | |
| "loss": 0.14205749332904816, | |
| "step": 34560 | |
| }, | |
| { | |
| "epoch": 1.0094227479632367, | |
| "grad_norm": 1.5216364860534668, | |
| "learning_rate": 8.990954161955293e-06, | |
| "loss": 0.13985899090766907, | |
| "step": 34816 | |
| }, | |
| { | |
| "epoch": 1.0168449740512018, | |
| "grad_norm": 1.1807845830917358, | |
| "learning_rate": 8.98353193586733e-06, | |
| "loss": 0.1413014978170395, | |
| "step": 35072 | |
| }, | |
| { | |
| "epoch": 1.0242672001391668, | |
| "grad_norm": 1.9225717782974243, | |
| "learning_rate": 8.976109709779364e-06, | |
| "loss": 0.13812309503555298, | |
| "step": 35328 | |
| }, | |
| { | |
| "epoch": 1.0316894262271317, | |
| "grad_norm": 1.0795660018920898, | |
| "learning_rate": 8.968687483691399e-06, | |
| "loss": 0.14011716842651367, | |
| "step": 35584 | |
| }, | |
| { | |
| "epoch": 1.0391116523150967, | |
| "grad_norm": 1.509298324584961, | |
| "learning_rate": 8.96129425067409e-06, | |
| "loss": 0.13760092854499817, | |
| "step": 35840 | |
| }, | |
| { | |
| "epoch": 1.0465338784030616, | |
| "grad_norm": 1.0444724559783936, | |
| "learning_rate": 8.953872024586125e-06, | |
| "loss": 0.13855981826782227, | |
| "step": 36096 | |
| }, | |
| { | |
| "epoch": 1.0539561044910266, | |
| "grad_norm": 1.9751235246658325, | |
| "learning_rate": 8.94644979849816e-06, | |
| "loss": 0.13731317222118378, | |
| "step": 36352 | |
| }, | |
| { | |
| "epoch": 1.0613783305789917, | |
| "grad_norm": 1.20009183883667, | |
| "learning_rate": 8.939027572410194e-06, | |
| "loss": 0.1373283565044403, | |
| "step": 36608 | |
| }, | |
| { | |
| "epoch": 1.0688005566669565, | |
| "grad_norm": 1.3423534631729126, | |
| "learning_rate": 8.931605346322229e-06, | |
| "loss": 0.1386987268924713, | |
| "step": 36864 | |
| }, | |
| { | |
| "epoch": 1.0762227827549216, | |
| "grad_norm": 1.004169225692749, | |
| "learning_rate": 8.924183120234264e-06, | |
| "loss": 0.13769270479679108, | |
| "step": 37120 | |
| }, | |
| { | |
| "epoch": 1.0836450088428866, | |
| "grad_norm": 1.5145548582077026, | |
| "learning_rate": 8.9167608941463e-06, | |
| "loss": 0.13567906618118286, | |
| "step": 37376 | |
| }, | |
| { | |
| "epoch": 1.0910672349308514, | |
| "grad_norm": 1.0763458013534546, | |
| "learning_rate": 8.909338668058335e-06, | |
| "loss": 0.13864187896251678, | |
| "step": 37632 | |
| }, | |
| { | |
| "epoch": 1.0984894610188165, | |
| "grad_norm": 1.1319243907928467, | |
| "learning_rate": 8.90191644197037e-06, | |
| "loss": 0.13519945740699768, | |
| "step": 37888 | |
| }, | |
| { | |
| "epoch": 1.1059116871067816, | |
| "grad_norm": 0.9084325432777405, | |
| "learning_rate": 8.894494215882404e-06, | |
| "loss": 0.1363523304462433, | |
| "step": 38144 | |
| }, | |
| { | |
| "epoch": 1.1133339131947464, | |
| "grad_norm": 0.9511646032333374, | |
| "learning_rate": 8.887100982865095e-06, | |
| "loss": 0.13671842217445374, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 1.1207561392827114, | |
| "grad_norm": 1.5679905414581299, | |
| "learning_rate": 8.879678756777132e-06, | |
| "loss": 0.13565286993980408, | |
| "step": 38656 | |
| }, | |
| { | |
| "epoch": 1.1281783653706765, | |
| "grad_norm": 1.9631683826446533, | |
| "learning_rate": 8.872256530689167e-06, | |
| "loss": 0.13194496929645538, | |
| "step": 38912 | |
| }, | |
| { | |
| "epoch": 1.1356005914586413, | |
| "grad_norm": 1.5896919965744019, | |
| "learning_rate": 8.864834304601201e-06, | |
| "loss": 0.13439622521400452, | |
| "step": 39168 | |
| }, | |
| { | |
| "epoch": 1.1430228175466064, | |
| "grad_norm": 1.235543131828308, | |
| "learning_rate": 8.857412078513236e-06, | |
| "loss": 0.13403554260730743, | |
| "step": 39424 | |
| }, | |
| { | |
| "epoch": 1.1504450436345715, | |
| "grad_norm": 2.1350796222686768, | |
| "learning_rate": 8.84998985242527e-06, | |
| "loss": 0.13712982833385468, | |
| "step": 39680 | |
| }, | |
| { | |
| "epoch": 1.1578672697225363, | |
| "grad_norm": 0.9020894169807434, | |
| "learning_rate": 8.842567626337305e-06, | |
| "loss": 0.13317421078681946, | |
| "step": 39936 | |
| }, | |
| { | |
| "epoch": 1.1652894958105013, | |
| "grad_norm": 3.322113513946533, | |
| "learning_rate": 8.83514540024934e-06, | |
| "loss": 0.13471046090126038, | |
| "step": 40192 | |
| }, | |
| { | |
| "epoch": 1.1727117218984662, | |
| "grad_norm": 1.5997190475463867, | |
| "learning_rate": 8.827723174161376e-06, | |
| "loss": 0.13153131306171417, | |
| "step": 40448 | |
| }, | |
| { | |
| "epoch": 1.1801339479864312, | |
| "grad_norm": 0.8981184959411621, | |
| "learning_rate": 8.820300948073411e-06, | |
| "loss": 0.13265979290008545, | |
| "step": 40704 | |
| }, | |
| { | |
| "epoch": 1.1875561740743963, | |
| "grad_norm": 0.800658106803894, | |
| "learning_rate": 8.812878721985446e-06, | |
| "loss": 0.13258570432662964, | |
| "step": 40960 | |
| }, | |
| { | |
| "epoch": 1.1949784001623611, | |
| "grad_norm": 1.0833916664123535, | |
| "learning_rate": 8.805485488968137e-06, | |
| "loss": 0.13603292405605316, | |
| "step": 41216 | |
| }, | |
| { | |
| "epoch": 1.2024006262503262, | |
| "grad_norm": 1.1857402324676514, | |
| "learning_rate": 8.798063262880172e-06, | |
| "loss": 0.1306399703025818, | |
| "step": 41472 | |
| }, | |
| { | |
| "epoch": 1.2098228523382912, | |
| "grad_norm": 1.9018610715866089, | |
| "learning_rate": 8.790641036792208e-06, | |
| "loss": 0.1320612132549286, | |
| "step": 41728 | |
| }, | |
| { | |
| "epoch": 1.217245078426256, | |
| "grad_norm": 0.7998947501182556, | |
| "learning_rate": 8.783218810704243e-06, | |
| "loss": 0.1327388733625412, | |
| "step": 41984 | |
| }, | |
| { | |
| "epoch": 1.2246673045142211, | |
| "grad_norm": 0.7305934429168701, | |
| "learning_rate": 8.775796584616278e-06, | |
| "loss": 0.1329335719347, | |
| "step": 42240 | |
| }, | |
| { | |
| "epoch": 1.232089530602186, | |
| "grad_norm": 1.332330584526062, | |
| "learning_rate": 8.768374358528312e-06, | |
| "loss": 0.13091133534908295, | |
| "step": 42496 | |
| }, | |
| { | |
| "epoch": 1.239511756690151, | |
| "grad_norm": 0.902974545955658, | |
| "learning_rate": 8.760952132440347e-06, | |
| "loss": 0.13031251728534698, | |
| "step": 42752 | |
| }, | |
| { | |
| "epoch": 1.246933982778116, | |
| "grad_norm": 1.322767972946167, | |
| "learning_rate": 8.753529906352383e-06, | |
| "loss": 0.13118214905261993, | |
| "step": 43008 | |
| }, | |
| { | |
| "epoch": 1.254356208866081, | |
| "grad_norm": 0.9607165455818176, | |
| "learning_rate": 8.746107680264418e-06, | |
| "loss": 0.12728999555110931, | |
| "step": 43264 | |
| }, | |
| { | |
| "epoch": 1.261778434954046, | |
| "grad_norm": 2.0725040435791016, | |
| "learning_rate": 8.738685454176453e-06, | |
| "loss": 0.12813667953014374, | |
| "step": 43520 | |
| }, | |
| { | |
| "epoch": 1.269200661042011, | |
| "grad_norm": 0.9913116097450256, | |
| "learning_rate": 8.731263228088488e-06, | |
| "loss": 0.12957149744033813, | |
| "step": 43776 | |
| }, | |
| { | |
| "epoch": 1.2766228871299758, | |
| "grad_norm": 1.021332025527954, | |
| "learning_rate": 8.723841002000522e-06, | |
| "loss": 0.13045457005500793, | |
| "step": 44032 | |
| }, | |
| { | |
| "epoch": 1.284045113217941, | |
| "grad_norm": 1.0940579175949097, | |
| "learning_rate": 8.716418775912557e-06, | |
| "loss": 0.12959586083889008, | |
| "step": 44288 | |
| }, | |
| { | |
| "epoch": 1.291467339305906, | |
| "grad_norm": 1.2633270025253296, | |
| "learning_rate": 8.708996549824592e-06, | |
| "loss": 0.1291976124048233, | |
| "step": 44544 | |
| }, | |
| { | |
| "epoch": 1.2988895653938708, | |
| "grad_norm": 1.3075908422470093, | |
| "learning_rate": 8.701574323736628e-06, | |
| "loss": 0.12747886776924133, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 1.3063117914818358, | |
| "grad_norm": 0.7914457321166992, | |
| "learning_rate": 8.69418109071932e-06, | |
| "loss": 0.12809380888938904, | |
| "step": 45056 | |
| }, | |
| { | |
| "epoch": 1.313734017569801, | |
| "grad_norm": 2.0950191020965576, | |
| "learning_rate": 8.686758864631354e-06, | |
| "loss": 0.12761422991752625, | |
| "step": 45312 | |
| }, | |
| { | |
| "epoch": 1.3211562436577657, | |
| "grad_norm": 1.1979840993881226, | |
| "learning_rate": 8.679336638543389e-06, | |
| "loss": 0.12491865456104279, | |
| "step": 45568 | |
| }, | |
| { | |
| "epoch": 1.3285784697457308, | |
| "grad_norm": 1.2227606773376465, | |
| "learning_rate": 8.671914412455423e-06, | |
| "loss": 0.1275608092546463, | |
| "step": 45824 | |
| }, | |
| { | |
| "epoch": 1.3360006958336959, | |
| "grad_norm": 1.1692591905593872, | |
| "learning_rate": 8.66449218636746e-06, | |
| "loss": 0.12718796730041504, | |
| "step": 46080 | |
| }, | |
| { | |
| "epoch": 1.3434229219216607, | |
| "grad_norm": 2.8268024921417236, | |
| "learning_rate": 8.657069960279495e-06, | |
| "loss": 0.12772701680660248, | |
| "step": 46336 | |
| }, | |
| { | |
| "epoch": 1.3508451480096257, | |
| "grad_norm": 1.737999677658081, | |
| "learning_rate": 8.64964773419153e-06, | |
| "loss": 0.12752339243888855, | |
| "step": 46592 | |
| }, | |
| { | |
| "epoch": 1.3582673740975908, | |
| "grad_norm": 0.8608792424201965, | |
| "learning_rate": 8.642225508103564e-06, | |
| "loss": 0.12575802206993103, | |
| "step": 46848 | |
| }, | |
| { | |
| "epoch": 1.3656896001855556, | |
| "grad_norm": 0.7649343609809875, | |
| "learning_rate": 8.634861268156912e-06, | |
| "loss": 0.12526509165763855, | |
| "step": 47104 | |
| }, | |
| { | |
| "epoch": 1.3731118262735207, | |
| "grad_norm": 1.2016552686691284, | |
| "learning_rate": 8.627439042068947e-06, | |
| "loss": 0.12652921676635742, | |
| "step": 47360 | |
| }, | |
| { | |
| "epoch": 1.3805340523614857, | |
| "grad_norm": 1.7116305828094482, | |
| "learning_rate": 8.620016815980981e-06, | |
| "loss": 0.12610307335853577, | |
| "step": 47616 | |
| }, | |
| { | |
| "epoch": 1.3879562784494506, | |
| "grad_norm": 1.7277467250823975, | |
| "learning_rate": 8.612594589893016e-06, | |
| "loss": 0.12416792660951614, | |
| "step": 47872 | |
| }, | |
| { | |
| "epoch": 1.3953785045374156, | |
| "grad_norm": 1.5151106119155884, | |
| "learning_rate": 8.60517236380505e-06, | |
| "loss": 0.12358472496271133, | |
| "step": 48128 | |
| }, | |
| { | |
| "epoch": 1.4028007306253805, | |
| "grad_norm": 1.2170330286026, | |
| "learning_rate": 8.597750137717085e-06, | |
| "loss": 0.12389586120843887, | |
| "step": 48384 | |
| }, | |
| { | |
| "epoch": 1.4102229567133455, | |
| "grad_norm": 1.6019481420516968, | |
| "learning_rate": 8.590327911629122e-06, | |
| "loss": 0.12313312292098999, | |
| "step": 48640 | |
| }, | |
| { | |
| "epoch": 1.4176451828013104, | |
| "grad_norm": 2.814764976501465, | |
| "learning_rate": 8.582905685541156e-06, | |
| "loss": 0.12451320886611938, | |
| "step": 48896 | |
| }, | |
| { | |
| "epoch": 1.4250674088892754, | |
| "grad_norm": 1.2372478246688843, | |
| "learning_rate": 8.575483459453191e-06, | |
| "loss": 0.12642629444599152, | |
| "step": 49152 | |
| }, | |
| { | |
| "epoch": 1.4324896349772405, | |
| "grad_norm": 1.5136011838912964, | |
| "learning_rate": 8.568061233365228e-06, | |
| "loss": 0.12238305807113647, | |
| "step": 49408 | |
| }, | |
| { | |
| "epoch": 1.4399118610652053, | |
| "grad_norm": 1.046945571899414, | |
| "learning_rate": 8.560639007277262e-06, | |
| "loss": 0.12444665282964706, | |
| "step": 49664 | |
| }, | |
| { | |
| "epoch": 1.4473340871531704, | |
| "grad_norm": 0.8409854769706726, | |
| "learning_rate": 8.553216781189297e-06, | |
| "loss": 0.1218654066324234, | |
| "step": 49920 | |
| }, | |
| { | |
| "epoch": 1.4547563132411354, | |
| "grad_norm": 1.408853530883789, | |
| "learning_rate": 8.545794555101332e-06, | |
| "loss": 0.12094538658857346, | |
| "step": 50176 | |
| }, | |
| { | |
| "epoch": 1.4621785393291002, | |
| "grad_norm": 1.2416728734970093, | |
| "learning_rate": 8.538372329013366e-06, | |
| "loss": 0.12048879265785217, | |
| "step": 50432 | |
| }, | |
| { | |
| "epoch": 1.4696007654170653, | |
| "grad_norm": 0.9202849268913269, | |
| "learning_rate": 8.530950102925401e-06, | |
| "loss": 0.12463593482971191, | |
| "step": 50688 | |
| }, | |
| { | |
| "epoch": 1.4770229915050304, | |
| "grad_norm": 1.6941767930984497, | |
| "learning_rate": 8.523527876837436e-06, | |
| "loss": 0.11946967244148254, | |
| "step": 50944 | |
| }, | |
| { | |
| "epoch": 1.4844452175929952, | |
| "grad_norm": 0.7557116150856018, | |
| "learning_rate": 8.516134643820127e-06, | |
| "loss": 0.12142129242420197, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 1.4918674436809602, | |
| "grad_norm": 1.440682053565979, | |
| "learning_rate": 8.508712417732162e-06, | |
| "loss": 0.11962714046239853, | |
| "step": 51456 | |
| }, | |
| { | |
| "epoch": 1.4992896697689253, | |
| "grad_norm": 1.6416200399398804, | |
| "learning_rate": 8.501290191644198e-06, | |
| "loss": 0.12134124338626862, | |
| "step": 51712 | |
| }, | |
| { | |
| "epoch": 1.5067118958568901, | |
| "grad_norm": 2.0762887001037598, | |
| "learning_rate": 8.493867965556233e-06, | |
| "loss": 0.11776599287986755, | |
| "step": 51968 | |
| }, | |
| { | |
| "epoch": 1.5141341219448552, | |
| "grad_norm": 1.653464674949646, | |
| "learning_rate": 8.486445739468268e-06, | |
| "loss": 0.11823120713233948, | |
| "step": 52224 | |
| }, | |
| { | |
| "epoch": 1.5215563480328202, | |
| "grad_norm": 1.1407897472381592, | |
| "learning_rate": 8.479023513380302e-06, | |
| "loss": 0.11975972354412079, | |
| "step": 52480 | |
| }, | |
| { | |
| "epoch": 1.528978574120785, | |
| "grad_norm": 1.1272715330123901, | |
| "learning_rate": 8.471630280362993e-06, | |
| "loss": 0.12049368768930435, | |
| "step": 52736 | |
| }, | |
| { | |
| "epoch": 1.5364008002087501, | |
| "grad_norm": 1.2129286527633667, | |
| "learning_rate": 8.46420805427503e-06, | |
| "loss": 0.1207490786910057, | |
| "step": 52992 | |
| }, | |
| { | |
| "epoch": 1.5438230262967152, | |
| "grad_norm": 0.9161160588264465, | |
| "learning_rate": 8.456785828187065e-06, | |
| "loss": 0.12305586785078049, | |
| "step": 53248 | |
| }, | |
| { | |
| "epoch": 1.55124525238468, | |
| "grad_norm": 1.0582292079925537, | |
| "learning_rate": 8.4493636020991e-06, | |
| "loss": 0.11973460018634796, | |
| "step": 53504 | |
| }, | |
| { | |
| "epoch": 1.558667478472645, | |
| "grad_norm": 4.305648326873779, | |
| "learning_rate": 8.441941376011134e-06, | |
| "loss": 0.11983916908502579, | |
| "step": 53760 | |
| }, | |
| { | |
| "epoch": 1.5660897045606101, | |
| "grad_norm": 2.1069681644439697, | |
| "learning_rate": 8.434519149923169e-06, | |
| "loss": 0.119334377348423, | |
| "step": 54016 | |
| }, | |
| { | |
| "epoch": 1.573511930648575, | |
| "grad_norm": 2.2981162071228027, | |
| "learning_rate": 8.427096923835203e-06, | |
| "loss": 0.12015485763549805, | |
| "step": 54272 | |
| }, | |
| { | |
| "epoch": 1.5809341567365398, | |
| "grad_norm": 2.0601630210876465, | |
| "learning_rate": 8.419674697747238e-06, | |
| "loss": 0.11952651292085648, | |
| "step": 54528 | |
| }, | |
| { | |
| "epoch": 1.588356382824505, | |
| "grad_norm": 1.8301986455917358, | |
| "learning_rate": 8.412252471659275e-06, | |
| "loss": 0.1208249032497406, | |
| "step": 54784 | |
| }, | |
| { | |
| "epoch": 1.59577860891247, | |
| "grad_norm": 1.0358697175979614, | |
| "learning_rate": 8.40483024557131e-06, | |
| "loss": 0.11711903661489487, | |
| "step": 55040 | |
| }, | |
| { | |
| "epoch": 1.6032008350004348, | |
| "grad_norm": 1.39198899269104, | |
| "learning_rate": 8.397408019483344e-06, | |
| "loss": 0.11665457487106323, | |
| "step": 55296 | |
| }, | |
| { | |
| "epoch": 1.6106230610884, | |
| "grad_norm": 1.3846735954284668, | |
| "learning_rate": 8.390014786466035e-06, | |
| "loss": 0.11981271952390671, | |
| "step": 55552 | |
| }, | |
| { | |
| "epoch": 1.6180452871763649, | |
| "grad_norm": 1.2096779346466064, | |
| "learning_rate": 8.38259256037807e-06, | |
| "loss": 0.11532194912433624, | |
| "step": 55808 | |
| }, | |
| { | |
| "epoch": 1.6254675132643297, | |
| "grad_norm": 2.1540744304656982, | |
| "learning_rate": 8.375170334290106e-06, | |
| "loss": 0.11524799466133118, | |
| "step": 56064 | |
| }, | |
| { | |
| "epoch": 1.6328897393522948, | |
| "grad_norm": 1.7781891822814941, | |
| "learning_rate": 8.367748108202141e-06, | |
| "loss": 0.11464376002550125, | |
| "step": 56320 | |
| }, | |
| { | |
| "epoch": 1.6403119654402598, | |
| "grad_norm": 1.4986239671707153, | |
| "learning_rate": 8.360325882114176e-06, | |
| "loss": 0.11543695628643036, | |
| "step": 56576 | |
| }, | |
| { | |
| "epoch": 1.6477341915282246, | |
| "grad_norm": 1.6460288763046265, | |
| "learning_rate": 8.35290365602621e-06, | |
| "loss": 0.11459029465913773, | |
| "step": 56832 | |
| }, | |
| { | |
| "epoch": 1.6551564176161897, | |
| "grad_norm": 1.1924717426300049, | |
| "learning_rate": 8.345481429938245e-06, | |
| "loss": 0.11774098873138428, | |
| "step": 57088 | |
| }, | |
| { | |
| "epoch": 1.6625786437041548, | |
| "grad_norm": 1.9193120002746582, | |
| "learning_rate": 8.33805920385028e-06, | |
| "loss": 0.11561363935470581, | |
| "step": 57344 | |
| }, | |
| { | |
| "epoch": 1.6700008697921196, | |
| "grad_norm": 1.1585633754730225, | |
| "learning_rate": 8.330636977762315e-06, | |
| "loss": 0.11557486653327942, | |
| "step": 57600 | |
| }, | |
| { | |
| "epoch": 1.6774230958800846, | |
| "grad_norm": 1.5214717388153076, | |
| "learning_rate": 8.323243744745006e-06, | |
| "loss": 0.11415628343820572, | |
| "step": 57856 | |
| }, | |
| { | |
| "epoch": 1.6848453219680497, | |
| "grad_norm": 2.019367218017578, | |
| "learning_rate": 8.31582151865704e-06, | |
| "loss": 0.1148451715707779, | |
| "step": 58112 | |
| }, | |
| { | |
| "epoch": 1.6922675480560145, | |
| "grad_norm": 1.1716550588607788, | |
| "learning_rate": 8.308399292569077e-06, | |
| "loss": 0.11534722149372101, | |
| "step": 58368 | |
| }, | |
| { | |
| "epoch": 1.6996897741439796, | |
| "grad_norm": 2.5370357036590576, | |
| "learning_rate": 8.300977066481112e-06, | |
| "loss": 0.11320699006319046, | |
| "step": 58624 | |
| }, | |
| { | |
| "epoch": 1.7071120002319446, | |
| "grad_norm": 1.1179746389389038, | |
| "learning_rate": 8.293554840393146e-06, | |
| "loss": 0.1134825050830841, | |
| "step": 58880 | |
| }, | |
| { | |
| "epoch": 1.7145342263199095, | |
| "grad_norm": 1.0148507356643677, | |
| "learning_rate": 8.286132614305181e-06, | |
| "loss": 0.11256734281778336, | |
| "step": 59136 | |
| }, | |
| { | |
| "epoch": 1.7219564524078745, | |
| "grad_norm": 1.0799450874328613, | |
| "learning_rate": 8.278710388217217e-06, | |
| "loss": 0.1122061163187027, | |
| "step": 59392 | |
| }, | |
| { | |
| "epoch": 1.7293786784958396, | |
| "grad_norm": 2.1830599308013916, | |
| "learning_rate": 8.271288162129252e-06, | |
| "loss": 0.11274608224630356, | |
| "step": 59648 | |
| }, | |
| { | |
| "epoch": 1.7368009045838044, | |
| "grad_norm": 1.5180538892745972, | |
| "learning_rate": 8.263865936041287e-06, | |
| "loss": 0.11179912835359573, | |
| "step": 59904 | |
| }, | |
| { | |
| "epoch": 1.7442231306717695, | |
| "grad_norm": 3.622218608856201, | |
| "learning_rate": 8.256443709953322e-06, | |
| "loss": 0.11142509430646896, | |
| "step": 60160 | |
| }, | |
| { | |
| "epoch": 1.7516453567597345, | |
| "grad_norm": 1.516839861869812, | |
| "learning_rate": 8.249021483865358e-06, | |
| "loss": 0.11382753401994705, | |
| "step": 60416 | |
| }, | |
| { | |
| "epoch": 1.7590675828476994, | |
| "grad_norm": 1.0474375486373901, | |
| "learning_rate": 8.241599257777393e-06, | |
| "loss": 0.11246595531702042, | |
| "step": 60672 | |
| }, | |
| { | |
| "epoch": 1.7664898089356644, | |
| "grad_norm": 1.2315608263015747, | |
| "learning_rate": 8.234177031689427e-06, | |
| "loss": 0.11168438196182251, | |
| "step": 60928 | |
| }, | |
| { | |
| "epoch": 1.7739120350236295, | |
| "grad_norm": 0.8689653873443604, | |
| "learning_rate": 8.226754805601462e-06, | |
| "loss": 0.11510151624679565, | |
| "step": 61184 | |
| }, | |
| { | |
| "epoch": 1.7813342611115943, | |
| "grad_norm": 1.1296117305755615, | |
| "learning_rate": 8.219332579513497e-06, | |
| "loss": 0.11331994086503983, | |
| "step": 61440 | |
| }, | |
| { | |
| "epoch": 1.7887564871995592, | |
| "grad_norm": 1.0127309560775757, | |
| "learning_rate": 8.211939346496188e-06, | |
| "loss": 0.11161452531814575, | |
| "step": 61696 | |
| }, | |
| { | |
| "epoch": 1.7961787132875244, | |
| "grad_norm": 1.368730902671814, | |
| "learning_rate": 8.204517120408223e-06, | |
| "loss": 0.11258195340633392, | |
| "step": 61952 | |
| }, | |
| { | |
| "epoch": 1.8036009393754893, | |
| "grad_norm": 1.853428840637207, | |
| "learning_rate": 8.197094894320257e-06, | |
| "loss": 0.11170430481433868, | |
| "step": 62208 | |
| }, | |
| { | |
| "epoch": 1.811023165463454, | |
| "grad_norm": 1.8468483686447144, | |
| "learning_rate": 8.189672668232294e-06, | |
| "loss": 0.11008735746145248, | |
| "step": 62464 | |
| }, | |
| { | |
| "epoch": 1.8184453915514194, | |
| "grad_norm": 2.3388783931732178, | |
| "learning_rate": 8.182250442144328e-06, | |
| "loss": 0.11217891424894333, | |
| "step": 62720 | |
| }, | |
| { | |
| "epoch": 1.8258676176393842, | |
| "grad_norm": 1.647824764251709, | |
| "learning_rate": 8.174828216056363e-06, | |
| "loss": 0.11034072190523148, | |
| "step": 62976 | |
| }, | |
| { | |
| "epoch": 1.833289843727349, | |
| "grad_norm": 1.8273885250091553, | |
| "learning_rate": 8.167405989968398e-06, | |
| "loss": 0.11056987196207047, | |
| "step": 63232 | |
| }, | |
| { | |
| "epoch": 1.840712069815314, | |
| "grad_norm": 1.7211445569992065, | |
| "learning_rate": 8.159983763880434e-06, | |
| "loss": 0.11001084744930267, | |
| "step": 63488 | |
| }, | |
| { | |
| "epoch": 1.8481342959032792, | |
| "grad_norm": 1.6466556787490845, | |
| "learning_rate": 8.152561537792469e-06, | |
| "loss": 0.111982062458992, | |
| "step": 63744 | |
| }, | |
| { | |
| "epoch": 1.855556521991244, | |
| "grad_norm": 1.6470916271209717, | |
| "learning_rate": 8.145139311704504e-06, | |
| "loss": 0.10999038070440292, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.862978748079209, | |
| "grad_norm": 1.1134272813796997, | |
| "learning_rate": 8.137746078687195e-06, | |
| "loss": 0.11305367946624756, | |
| "step": 64256 | |
| }, | |
| { | |
| "epoch": 1.870400974167174, | |
| "grad_norm": 1.1313408613204956, | |
| "learning_rate": 8.13032385259923e-06, | |
| "loss": 0.10864917188882828, | |
| "step": 64512 | |
| }, | |
| { | |
| "epoch": 1.877823200255139, | |
| "grad_norm": 1.1794612407684326, | |
| "learning_rate": 8.122901626511264e-06, | |
| "loss": 0.1088700145483017, | |
| "step": 64768 | |
| }, | |
| { | |
| "epoch": 1.885245426343104, | |
| "grad_norm": 1.4089781045913696, | |
| "learning_rate": 8.115479400423299e-06, | |
| "loss": 0.11004452407360077, | |
| "step": 65024 | |
| }, | |
| { | |
| "epoch": 1.892667652431069, | |
| "grad_norm": 1.1505563259124756, | |
| "learning_rate": 8.108057174335334e-06, | |
| "loss": 0.10707546025514603, | |
| "step": 65280 | |
| }, | |
| { | |
| "epoch": 1.9000898785190339, | |
| "grad_norm": 1.6187642812728882, | |
| "learning_rate": 8.10063494824737e-06, | |
| "loss": 0.10916116088628769, | |
| "step": 65536 | |
| }, | |
| { | |
| "epoch": 1.907512104606999, | |
| "grad_norm": 1.530271291732788, | |
| "learning_rate": 8.093212722159405e-06, | |
| "loss": 0.1079617589712143, | |
| "step": 65792 | |
| }, | |
| { | |
| "epoch": 1.914934330694964, | |
| "grad_norm": 0.9997887015342712, | |
| "learning_rate": 8.08579049607144e-06, | |
| "loss": 0.10774866491556168, | |
| "step": 66048 | |
| }, | |
| { | |
| "epoch": 1.9223565567829288, | |
| "grad_norm": 1.86561918258667, | |
| "learning_rate": 8.078368269983474e-06, | |
| "loss": 0.10869648307561874, | |
| "step": 66304 | |
| }, | |
| { | |
| "epoch": 1.9297787828708939, | |
| "grad_norm": 0.7563046216964722, | |
| "learning_rate": 8.07094604389551e-06, | |
| "loss": 0.10840298235416412, | |
| "step": 66560 | |
| }, | |
| { | |
| "epoch": 1.937201008958859, | |
| "grad_norm": 0.932775616645813, | |
| "learning_rate": 8.063552810878202e-06, | |
| "loss": 0.10603740066289902, | |
| "step": 66816 | |
| }, | |
| { | |
| "epoch": 1.9446232350468238, | |
| "grad_norm": 1.0463026762008667, | |
| "learning_rate": 8.056130584790237e-06, | |
| "loss": 0.10687943547964096, | |
| "step": 67072 | |
| }, | |
| { | |
| "epoch": 1.9520454611347888, | |
| "grad_norm": 1.870526671409607, | |
| "learning_rate": 8.048708358702271e-06, | |
| "loss": 0.10632742941379547, | |
| "step": 67328 | |
| }, | |
| { | |
| "epoch": 1.9594676872227539, | |
| "grad_norm": 1.9874995946884155, | |
| "learning_rate": 8.041286132614306e-06, | |
| "loss": 0.10460897535085678, | |
| "step": 67584 | |
| }, | |
| { | |
| "epoch": 1.9668899133107187, | |
| "grad_norm": 1.8464157581329346, | |
| "learning_rate": 8.03386390652634e-06, | |
| "loss": 0.10437272489070892, | |
| "step": 67840 | |
| }, | |
| { | |
| "epoch": 1.9743121393986836, | |
| "grad_norm": 2.218248128890991, | |
| "learning_rate": 8.026441680438375e-06, | |
| "loss": 0.1077708825469017, | |
| "step": 68096 | |
| }, | |
| { | |
| "epoch": 1.9817343654866488, | |
| "grad_norm": 1.2941707372665405, | |
| "learning_rate": 8.01901945435041e-06, | |
| "loss": 0.10709750652313232, | |
| "step": 68352 | |
| }, | |
| { | |
| "epoch": 1.9891565915746137, | |
| "grad_norm": 1.4901282787322998, | |
| "learning_rate": 8.011597228262445e-06, | |
| "loss": 0.10707870870828629, | |
| "step": 68608 | |
| }, | |
| { | |
| "epoch": 1.9965788176625785, | |
| "grad_norm": 1.0968399047851562, | |
| "learning_rate": 8.004175002174481e-06, | |
| "loss": 0.10603955388069153, | |
| "step": 68864 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_f1_B-DISEASE": 0.8233951346401412, | |
| "eval_f1_B-PROCEDURE": 0.8228046416527258, | |
| "eval_f1_B-SYMPTOM": 0.8032099730692381, | |
| "eval_f1_I-DISEASE": 0.8546127024840142, | |
| "eval_f1_I-PROCEDURE": 0.8602104264252335, | |
| "eval_f1_I-SYMPTOM": 0.8244893065033029, | |
| "eval_f1_O": 0.9276598726449387, | |
| "eval_f1_macro": 0.8451974367742278, | |
| "eval_f1_micro": 0.8891140662061447, | |
| "eval_loss": 0.07220370322465897, | |
| "eval_precision_B-DISEASE": 0.822598180902124, | |
| "eval_precision_B-PROCEDURE": 0.8242930488680801, | |
| "eval_precision_B-SYMPTOM": 0.8017820751853618, | |
| "eval_precision_I-DISEASE": 0.8372941389498603, | |
| "eval_precision_I-PROCEDURE": 0.8852814590217922, | |
| "eval_precision_I-SYMPTOM": 0.8002334635839129, | |
| "eval_precision_O": 0.941591476196215, | |
| "eval_precision_macro": 0.8447248346724779, | |
| "eval_precision_micro": 0.8927074638692154, | |
| "eval_rauc_macro": 0.9112110229653825, | |
| "eval_rauc_micro": 0.9335868422231772, | |
| "eval_recall_B-DISEASE": 0.8241936340932037, | |
| "eval_recall_B-PROCEDURE": 0.8213215999151801, | |
| "eval_recall_B-SYMPTOM": 0.8046429659283756, | |
| "eval_recall_I-DISEASE": 0.872662830840046, | |
| "eval_recall_I-PROCEDURE": 0.8365203021718602, | |
| "eval_recall_I-SYMPTOM": 0.8502615486075722, | |
| "eval_recall_O": 0.9141345168999658, | |
| "eval_recall_macro": 0.8462481997794578, | |
| "eval_recall_micro": 0.8855494814279414, | |
| "eval_roc_auc_B-DISEASE": 0.9101770523081748, | |
| "eval_roc_auc_B-PROCEDURE": 0.908765330030275, | |
| "eval_roc_auc_B-SYMPTOM": 0.8996852101804533, | |
| "eval_roc_auc_I-DISEASE": 0.9245444755324247, | |
| "eval_roc_auc_I-PROCEDURE": 0.9119212634345877, | |
| "eval_roc_auc_I-SYMPTOM": 0.9064134086726298, | |
| "eval_roc_auc_O": 0.9169704205991321, | |
| "eval_runtime": 50.5178, | |
| "eval_samples_per_second": 280.099, | |
| "eval_steps_per_second": 35.017, | |
| "step": 68982 | |
| }, | |
| { | |
| "epoch": 2.0040010437505438, | |
| "grad_norm": 1.5970464944839478, | |
| "learning_rate": 7.996752776086516e-06, | |
| "loss": 0.10319741070270538, | |
| "step": 69120 | |
| }, | |
| { | |
| "epoch": 2.0114232698385086, | |
| "grad_norm": 1.0218013525009155, | |
| "learning_rate": 7.98933054999855e-06, | |
| "loss": 0.10018162429332733, | |
| "step": 69376 | |
| }, | |
| { | |
| "epoch": 2.0188454959264734, | |
| "grad_norm": 0.8377583026885986, | |
| "learning_rate": 7.981908323910587e-06, | |
| "loss": 0.09980139136314392, | |
| "step": 69632 | |
| }, | |
| { | |
| "epoch": 2.0262677220144387, | |
| "grad_norm": 1.3614574670791626, | |
| "learning_rate": 7.974486097822622e-06, | |
| "loss": 0.10014472156763077, | |
| "step": 69888 | |
| }, | |
| { | |
| "epoch": 2.0336899481024036, | |
| "grad_norm": 1.567603588104248, | |
| "learning_rate": 7.967063871734656e-06, | |
| "loss": 0.1009359210729599, | |
| "step": 70144 | |
| }, | |
| { | |
| "epoch": 2.0411121741903684, | |
| "grad_norm": 1.2348442077636719, | |
| "learning_rate": 7.959641645646691e-06, | |
| "loss": 0.0986831933259964, | |
| "step": 70400 | |
| }, | |
| { | |
| "epoch": 2.0485344002783337, | |
| "grad_norm": 1.1694316864013672, | |
| "learning_rate": 7.952248412629382e-06, | |
| "loss": 0.09873685240745544, | |
| "step": 70656 | |
| }, | |
| { | |
| "epoch": 2.0559566263662985, | |
| "grad_norm": 1.4164974689483643, | |
| "learning_rate": 7.944826186541417e-06, | |
| "loss": 0.09888967871665955, | |
| "step": 70912 | |
| }, | |
| { | |
| "epoch": 2.0633788524542633, | |
| "grad_norm": 2.2125892639160156, | |
| "learning_rate": 7.937403960453452e-06, | |
| "loss": 0.0989769995212555, | |
| "step": 71168 | |
| }, | |
| { | |
| "epoch": 2.0708010785422286, | |
| "grad_norm": 1.680055856704712, | |
| "learning_rate": 7.929981734365487e-06, | |
| "loss": 0.0989062488079071, | |
| "step": 71424 | |
| }, | |
| { | |
| "epoch": 2.0782233046301934, | |
| "grad_norm": 1.0069916248321533, | |
| "learning_rate": 7.922559508277521e-06, | |
| "loss": 0.0979849249124527, | |
| "step": 71680 | |
| }, | |
| { | |
| "epoch": 2.0856455307181583, | |
| "grad_norm": 0.8232192993164062, | |
| "learning_rate": 7.915137282189558e-06, | |
| "loss": 0.09770826995372772, | |
| "step": 71936 | |
| }, | |
| { | |
| "epoch": 2.093067756806123, | |
| "grad_norm": 0.7202551364898682, | |
| "learning_rate": 7.907715056101592e-06, | |
| "loss": 0.0982343927025795, | |
| "step": 72192 | |
| }, | |
| { | |
| "epoch": 2.1004899828940884, | |
| "grad_norm": 2.105822801589966, | |
| "learning_rate": 7.900292830013627e-06, | |
| "loss": 0.09729639440774918, | |
| "step": 72448 | |
| }, | |
| { | |
| "epoch": 2.1079122089820532, | |
| "grad_norm": 1.5673470497131348, | |
| "learning_rate": 7.892899596996318e-06, | |
| "loss": 0.09895992279052734, | |
| "step": 72704 | |
| }, | |
| { | |
| "epoch": 2.115334435070018, | |
| "grad_norm": 1.1652296781539917, | |
| "learning_rate": 7.885477370908353e-06, | |
| "loss": 0.09693095833063126, | |
| "step": 72960 | |
| }, | |
| { | |
| "epoch": 2.1227566611579833, | |
| "grad_norm": 1.5735524892807007, | |
| "learning_rate": 7.87805514482039e-06, | |
| "loss": 0.09585839509963989, | |
| "step": 73216 | |
| }, | |
| { | |
| "epoch": 2.130178887245948, | |
| "grad_norm": 1.4391745328903198, | |
| "learning_rate": 7.87066191180308e-06, | |
| "loss": 0.09629447013139725, | |
| "step": 73472 | |
| }, | |
| { | |
| "epoch": 2.137601113333913, | |
| "grad_norm": 2.274918794631958, | |
| "learning_rate": 7.863239685715115e-06, | |
| "loss": 0.09710395336151123, | |
| "step": 73728 | |
| }, | |
| { | |
| "epoch": 2.1450233394218783, | |
| "grad_norm": 1.181287407875061, | |
| "learning_rate": 7.85581745962715e-06, | |
| "loss": 0.09741077572107315, | |
| "step": 73984 | |
| }, | |
| { | |
| "epoch": 2.152445565509843, | |
| "grad_norm": 1.2975316047668457, | |
| "learning_rate": 7.848395233539185e-06, | |
| "loss": 0.09615115821361542, | |
| "step": 74240 | |
| }, | |
| { | |
| "epoch": 2.159867791597808, | |
| "grad_norm": 1.1137505769729614, | |
| "learning_rate": 7.84097300745122e-06, | |
| "loss": 0.09520907700061798, | |
| "step": 74496 | |
| }, | |
| { | |
| "epoch": 2.1672900176857732, | |
| "grad_norm": 1.833809494972229, | |
| "learning_rate": 7.833550781363254e-06, | |
| "loss": 0.09584689140319824, | |
| "step": 74752 | |
| }, | |
| { | |
| "epoch": 2.174712243773738, | |
| "grad_norm": 1.0752769708633423, | |
| "learning_rate": 7.826128555275289e-06, | |
| "loss": 0.09547276794910431, | |
| "step": 75008 | |
| }, | |
| { | |
| "epoch": 2.182134469861703, | |
| "grad_norm": 1.0917208194732666, | |
| "learning_rate": 7.818706329187324e-06, | |
| "loss": 0.09638134390115738, | |
| "step": 75264 | |
| }, | |
| { | |
| "epoch": 2.189556695949668, | |
| "grad_norm": 0.8327083587646484, | |
| "learning_rate": 7.81128410309936e-06, | |
| "loss": 0.09734158962965012, | |
| "step": 75520 | |
| }, | |
| { | |
| "epoch": 2.196978922037633, | |
| "grad_norm": 1.6501367092132568, | |
| "learning_rate": 7.803861877011395e-06, | |
| "loss": 0.09588190913200378, | |
| "step": 75776 | |
| }, | |
| { | |
| "epoch": 2.204401148125598, | |
| "grad_norm": 1.445810317993164, | |
| "learning_rate": 7.796468643994086e-06, | |
| "loss": 0.09707304835319519, | |
| "step": 76032 | |
| }, | |
| { | |
| "epoch": 2.211823374213563, | |
| "grad_norm": 3.2810208797454834, | |
| "learning_rate": 7.78904641790612e-06, | |
| "loss": 0.09537787735462189, | |
| "step": 76288 | |
| }, | |
| { | |
| "epoch": 2.219245600301528, | |
| "grad_norm": 1.1140615940093994, | |
| "learning_rate": 7.781624191818155e-06, | |
| "loss": 0.09657083451747894, | |
| "step": 76544 | |
| }, | |
| { | |
| "epoch": 2.226667826389493, | |
| "grad_norm": 2.93180251121521, | |
| "learning_rate": 7.774201965730192e-06, | |
| "loss": 0.09558647125959396, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 2.234090052477458, | |
| "grad_norm": 1.2372024059295654, | |
| "learning_rate": 7.766779739642227e-06, | |
| "loss": 0.09815304726362228, | |
| "step": 77056 | |
| }, | |
| { | |
| "epoch": 2.241512278565423, | |
| "grad_norm": 1.6396515369415283, | |
| "learning_rate": 7.759357513554261e-06, | |
| "loss": 0.09785998612642288, | |
| "step": 77312 | |
| }, | |
| { | |
| "epoch": 2.2489345046533877, | |
| "grad_norm": 2.0098323822021484, | |
| "learning_rate": 7.751935287466296e-06, | |
| "loss": 0.09462552517652512, | |
| "step": 77568 | |
| }, | |
| { | |
| "epoch": 2.256356730741353, | |
| "grad_norm": 2.05253529548645, | |
| "learning_rate": 7.744513061378332e-06, | |
| "loss": 0.0934867262840271, | |
| "step": 77824 | |
| }, | |
| { | |
| "epoch": 2.263778956829318, | |
| "grad_norm": 0.9149698615074158, | |
| "learning_rate": 7.737090835290367e-06, | |
| "loss": 0.09518175572156906, | |
| "step": 78080 | |
| }, | |
| { | |
| "epoch": 2.2712011829172827, | |
| "grad_norm": 1.2371445894241333, | |
| "learning_rate": 7.729668609202402e-06, | |
| "loss": 0.09454512596130371, | |
| "step": 78336 | |
| }, | |
| { | |
| "epoch": 2.278623409005248, | |
| "grad_norm": 1.2862200736999512, | |
| "learning_rate": 7.722246383114436e-06, | |
| "loss": 0.09522521495819092, | |
| "step": 78592 | |
| }, | |
| { | |
| "epoch": 2.286045635093213, | |
| "grad_norm": 0.6058925986289978, | |
| "learning_rate": 7.714824157026471e-06, | |
| "loss": 0.09450002759695053, | |
| "step": 78848 | |
| }, | |
| { | |
| "epoch": 2.2934678611811776, | |
| "grad_norm": 1.2032670974731445, | |
| "learning_rate": 7.707401930938506e-06, | |
| "loss": 0.09529651701450348, | |
| "step": 79104 | |
| }, | |
| { | |
| "epoch": 2.300890087269143, | |
| "grad_norm": 2.0551536083221436, | |
| "learning_rate": 7.69997970485054e-06, | |
| "loss": 0.0937376618385315, | |
| "step": 79360 | |
| }, | |
| { | |
| "epoch": 2.3083123133571077, | |
| "grad_norm": 1.9434480667114258, | |
| "learning_rate": 7.692557478762577e-06, | |
| "loss": 0.09268642216920853, | |
| "step": 79616 | |
| }, | |
| { | |
| "epoch": 2.3157345394450726, | |
| "grad_norm": 1.234994649887085, | |
| "learning_rate": 7.685135252674612e-06, | |
| "loss": 0.0928453728556633, | |
| "step": 79872 | |
| }, | |
| { | |
| "epoch": 2.323156765533038, | |
| "grad_norm": 1.302419900894165, | |
| "learning_rate": 7.677742019657303e-06, | |
| "loss": 0.0948258712887764, | |
| "step": 80128 | |
| }, | |
| { | |
| "epoch": 2.3305789916210027, | |
| "grad_norm": 1.2769848108291626, | |
| "learning_rate": 7.670319793569338e-06, | |
| "loss": 0.09256388247013092, | |
| "step": 80384 | |
| }, | |
| { | |
| "epoch": 2.3380012177089675, | |
| "grad_norm": 1.245105266571045, | |
| "learning_rate": 7.662897567481372e-06, | |
| "loss": 0.09323182702064514, | |
| "step": 80640 | |
| }, | |
| { | |
| "epoch": 2.3454234437969324, | |
| "grad_norm": 1.4147142171859741, | |
| "learning_rate": 7.655475341393409e-06, | |
| "loss": 0.0920487642288208, | |
| "step": 80896 | |
| }, | |
| { | |
| "epoch": 2.3528456698848976, | |
| "grad_norm": 0.9135209321975708, | |
| "learning_rate": 7.648053115305443e-06, | |
| "loss": 0.09392034262418747, | |
| "step": 81152 | |
| }, | |
| { | |
| "epoch": 2.3602678959728625, | |
| "grad_norm": 3.3641445636749268, | |
| "learning_rate": 7.640630889217478e-06, | |
| "loss": 0.09316613525152206, | |
| "step": 81408 | |
| }, | |
| { | |
| "epoch": 2.3676901220608273, | |
| "grad_norm": 1.657092809677124, | |
| "learning_rate": 7.633208663129513e-06, | |
| "loss": 0.09134019911289215, | |
| "step": 81664 | |
| }, | |
| { | |
| "epoch": 2.3751123481487926, | |
| "grad_norm": 1.7811989784240723, | |
| "learning_rate": 7.625786437041548e-06, | |
| "loss": 0.09261273592710495, | |
| "step": 81920 | |
| }, | |
| { | |
| "epoch": 2.3825345742367574, | |
| "grad_norm": 1.0955129861831665, | |
| "learning_rate": 7.6184221970948945e-06, | |
| "loss": 0.09145151078701019, | |
| "step": 82176 | |
| }, | |
| { | |
| "epoch": 2.3899568003247222, | |
| "grad_norm": 1.3998067378997803, | |
| "learning_rate": 7.61099997100693e-06, | |
| "loss": 0.09198223054409027, | |
| "step": 82432 | |
| }, | |
| { | |
| "epoch": 2.3973790264126875, | |
| "grad_norm": 1.798430323600769, | |
| "learning_rate": 7.603577744918965e-06, | |
| "loss": 0.09014385938644409, | |
| "step": 82688 | |
| }, | |
| { | |
| "epoch": 2.4048012525006524, | |
| "grad_norm": 0.867472767829895, | |
| "learning_rate": 7.5961555188309995e-06, | |
| "loss": 0.09200616925954819, | |
| "step": 82944 | |
| }, | |
| { | |
| "epoch": 2.412223478588617, | |
| "grad_norm": 1.5483663082122803, | |
| "learning_rate": 7.588733292743035e-06, | |
| "loss": 0.09048184007406235, | |
| "step": 83200 | |
| }, | |
| { | |
| "epoch": 2.4196457046765825, | |
| "grad_norm": 1.0577336549758911, | |
| "learning_rate": 7.5813110666550706e-06, | |
| "loss": 0.0909578874707222, | |
| "step": 83456 | |
| }, | |
| { | |
| "epoch": 2.4270679307645473, | |
| "grad_norm": 1.7594668865203857, | |
| "learning_rate": 7.573888840567105e-06, | |
| "loss": 0.09119825810194016, | |
| "step": 83712 | |
| }, | |
| { | |
| "epoch": 2.434490156852512, | |
| "grad_norm": 1.3315958976745605, | |
| "learning_rate": 7.56646661447914e-06, | |
| "loss": 0.09173890948295593, | |
| "step": 83968 | |
| }, | |
| { | |
| "epoch": 2.4419123829404774, | |
| "grad_norm": 1.1693644523620605, | |
| "learning_rate": 7.559044388391175e-06, | |
| "loss": 0.09135568141937256, | |
| "step": 84224 | |
| }, | |
| { | |
| "epoch": 2.4493346090284422, | |
| "grad_norm": 1.1136761903762817, | |
| "learning_rate": 7.55162216230321e-06, | |
| "loss": 0.09019352495670319, | |
| "step": 84480 | |
| }, | |
| { | |
| "epoch": 2.456756835116407, | |
| "grad_norm": 1.3075796365737915, | |
| "learning_rate": 7.544199936215245e-06, | |
| "loss": 0.08919453620910645, | |
| "step": 84736 | |
| }, | |
| { | |
| "epoch": 2.464179061204372, | |
| "grad_norm": 1.0504467487335205, | |
| "learning_rate": 7.53677771012728e-06, | |
| "loss": 0.09064843505620956, | |
| "step": 84992 | |
| }, | |
| { | |
| "epoch": 2.471601287292337, | |
| "grad_norm": 1.0015623569488525, | |
| "learning_rate": 7.529355484039316e-06, | |
| "loss": 0.08907545357942581, | |
| "step": 85248 | |
| }, | |
| { | |
| "epoch": 2.479023513380302, | |
| "grad_norm": 2.6646406650543213, | |
| "learning_rate": 7.521933257951351e-06, | |
| "loss": 0.09090094268321991, | |
| "step": 85504 | |
| }, | |
| { | |
| "epoch": 2.486445739468267, | |
| "grad_norm": 0.9044279456138611, | |
| "learning_rate": 7.5145110318633854e-06, | |
| "loss": 0.08867330104112625, | |
| "step": 85760 | |
| }, | |
| { | |
| "epoch": 2.493867965556232, | |
| "grad_norm": 1.5245310068130493, | |
| "learning_rate": 7.507117798846077e-06, | |
| "loss": 0.08801846206188202, | |
| "step": 86016 | |
| }, | |
| { | |
| "epoch": 2.501290191644197, | |
| "grad_norm": 1.9575257301330566, | |
| "learning_rate": 7.499695572758111e-06, | |
| "loss": 0.0878206118941307, | |
| "step": 86272 | |
| }, | |
| { | |
| "epoch": 2.508712417732162, | |
| "grad_norm": 1.9037858247756958, | |
| "learning_rate": 7.492302339740803e-06, | |
| "loss": 0.08872802555561066, | |
| "step": 86528 | |
| }, | |
| { | |
| "epoch": 2.516134643820127, | |
| "grad_norm": 1.316570520401001, | |
| "learning_rate": 7.484880113652837e-06, | |
| "loss": 0.08829988539218903, | |
| "step": 86784 | |
| }, | |
| { | |
| "epoch": 2.523556869908092, | |
| "grad_norm": 1.5093870162963867, | |
| "learning_rate": 7.477457887564873e-06, | |
| "loss": 0.08735402673482895, | |
| "step": 87040 | |
| }, | |
| { | |
| "epoch": 2.5309790959960567, | |
| "grad_norm": 1.1591265201568604, | |
| "learning_rate": 7.470035661476908e-06, | |
| "loss": 0.08951527625322342, | |
| "step": 87296 | |
| }, | |
| { | |
| "epoch": 2.538401322084022, | |
| "grad_norm": 2.379953384399414, | |
| "learning_rate": 7.462613435388942e-06, | |
| "loss": 0.08941785246133804, | |
| "step": 87552 | |
| }, | |
| { | |
| "epoch": 2.545823548171987, | |
| "grad_norm": 1.987851858139038, | |
| "learning_rate": 7.455191209300978e-06, | |
| "loss": 0.08783798664808273, | |
| "step": 87808 | |
| }, | |
| { | |
| "epoch": 2.5532457742599517, | |
| "grad_norm": 1.3022395372390747, | |
| "learning_rate": 7.447768983213013e-06, | |
| "loss": 0.08796506375074387, | |
| "step": 88064 | |
| }, | |
| { | |
| "epoch": 2.560668000347917, | |
| "grad_norm": 1.0609015226364136, | |
| "learning_rate": 7.440346757125047e-06, | |
| "loss": 0.0878596156835556, | |
| "step": 88320 | |
| }, | |
| { | |
| "epoch": 2.568090226435882, | |
| "grad_norm": 1.2358717918395996, | |
| "learning_rate": 7.432924531037082e-06, | |
| "loss": 0.08771014213562012, | |
| "step": 88576 | |
| }, | |
| { | |
| "epoch": 2.5755124525238466, | |
| "grad_norm": 0.5960157513618469, | |
| "learning_rate": 7.425502304949118e-06, | |
| "loss": 0.08743297308683395, | |
| "step": 88832 | |
| }, | |
| { | |
| "epoch": 2.582934678611812, | |
| "grad_norm": 1.6683177947998047, | |
| "learning_rate": 7.418080078861153e-06, | |
| "loss": 0.08782097697257996, | |
| "step": 89088 | |
| }, | |
| { | |
| "epoch": 2.5903569046997768, | |
| "grad_norm": 1.572849154472351, | |
| "learning_rate": 7.410657852773188e-06, | |
| "loss": 0.08598018437623978, | |
| "step": 89344 | |
| }, | |
| { | |
| "epoch": 2.5977791307877416, | |
| "grad_norm": 1.9523414373397827, | |
| "learning_rate": 7.4032356266852225e-06, | |
| "loss": 0.087459497153759, | |
| "step": 89600 | |
| }, | |
| { | |
| "epoch": 2.605201356875707, | |
| "grad_norm": 0.7449246048927307, | |
| "learning_rate": 7.395813400597258e-06, | |
| "loss": 0.08550211042165756, | |
| "step": 89856 | |
| }, | |
| { | |
| "epoch": 2.6126235829636717, | |
| "grad_norm": 1.184494972229004, | |
| "learning_rate": 7.388391174509293e-06, | |
| "loss": 0.08633177727460861, | |
| "step": 90112 | |
| }, | |
| { | |
| "epoch": 2.6200458090516365, | |
| "grad_norm": 1.3583768606185913, | |
| "learning_rate": 7.3809689484213275e-06, | |
| "loss": 0.08681616932153702, | |
| "step": 90368 | |
| }, | |
| { | |
| "epoch": 2.627468035139602, | |
| "grad_norm": 1.078107237815857, | |
| "learning_rate": 7.373575715404019e-06, | |
| "loss": 0.0859987735748291, | |
| "step": 90624 | |
| }, | |
| { | |
| "epoch": 2.6348902612275666, | |
| "grad_norm": 1.467429757118225, | |
| "learning_rate": 7.36618248238671e-06, | |
| "loss": 0.08708871901035309, | |
| "step": 90880 | |
| }, | |
| { | |
| "epoch": 2.6423124873155315, | |
| "grad_norm": 1.6195470094680786, | |
| "learning_rate": 7.358760256298745e-06, | |
| "loss": 0.083903968334198, | |
| "step": 91136 | |
| }, | |
| { | |
| "epoch": 2.6497347134034968, | |
| "grad_norm": 1.304286241531372, | |
| "learning_rate": 7.35133803021078e-06, | |
| "loss": 0.08599025011062622, | |
| "step": 91392 | |
| }, | |
| { | |
| "epoch": 2.6571569394914616, | |
| "grad_norm": 1.6564602851867676, | |
| "learning_rate": 7.343915804122815e-06, | |
| "loss": 0.08739902824163437, | |
| "step": 91648 | |
| }, | |
| { | |
| "epoch": 2.6645791655794264, | |
| "grad_norm": 1.9799528121948242, | |
| "learning_rate": 7.33649357803485e-06, | |
| "loss": 0.08461599797010422, | |
| "step": 91904 | |
| }, | |
| { | |
| "epoch": 2.6720013916673917, | |
| "grad_norm": 1.25792396068573, | |
| "learning_rate": 7.329071351946885e-06, | |
| "loss": 0.08518636226654053, | |
| "step": 92160 | |
| }, | |
| { | |
| "epoch": 2.6794236177553565, | |
| "grad_norm": 1.426477313041687, | |
| "learning_rate": 7.321649125858921e-06, | |
| "loss": 0.08482255041599274, | |
| "step": 92416 | |
| }, | |
| { | |
| "epoch": 2.6868458438433214, | |
| "grad_norm": 0.8972841501235962, | |
| "learning_rate": 7.3142268997709555e-06, | |
| "loss": 0.08393040299415588, | |
| "step": 92672 | |
| }, | |
| { | |
| "epoch": 2.6942680699312866, | |
| "grad_norm": 1.1803803443908691, | |
| "learning_rate": 7.30680467368299e-06, | |
| "loss": 0.08566819876432419, | |
| "step": 92928 | |
| }, | |
| { | |
| "epoch": 2.7016902960192515, | |
| "grad_norm": 0.8352828025817871, | |
| "learning_rate": 7.299382447595026e-06, | |
| "loss": 0.0829949676990509, | |
| "step": 93184 | |
| }, | |
| { | |
| "epoch": 2.7091125221072163, | |
| "grad_norm": 1.5144259929656982, | |
| "learning_rate": 7.2919602215070604e-06, | |
| "loss": 0.08488274365663528, | |
| "step": 93440 | |
| }, | |
| { | |
| "epoch": 2.7165347481951816, | |
| "grad_norm": 1.2217360734939575, | |
| "learning_rate": 7.284537995419095e-06, | |
| "loss": 0.0851670354604721, | |
| "step": 93696 | |
| }, | |
| { | |
| "epoch": 2.7239569742831464, | |
| "grad_norm": 1.0168439149856567, | |
| "learning_rate": 7.27711576933113e-06, | |
| "loss": 0.08516183495521545, | |
| "step": 93952 | |
| }, | |
| { | |
| "epoch": 2.7313792003711113, | |
| "grad_norm": 1.9827710390090942, | |
| "learning_rate": 7.269693543243166e-06, | |
| "loss": 0.08326787501573563, | |
| "step": 94208 | |
| }, | |
| { | |
| "epoch": 2.7388014264590765, | |
| "grad_norm": 1.0075639486312866, | |
| "learning_rate": 7.262271317155201e-06, | |
| "loss": 0.08202008903026581, | |
| "step": 94464 | |
| }, | |
| { | |
| "epoch": 2.7462236525470414, | |
| "grad_norm": 2.176619529724121, | |
| "learning_rate": 7.254849091067236e-06, | |
| "loss": 0.08481749892234802, | |
| "step": 94720 | |
| }, | |
| { | |
| "epoch": 2.753645878635006, | |
| "grad_norm": 1.210526466369629, | |
| "learning_rate": 7.247455858049927e-06, | |
| "loss": 0.08310143649578094, | |
| "step": 94976 | |
| }, | |
| { | |
| "epoch": 2.7610681047229715, | |
| "grad_norm": 2.6383216381073, | |
| "learning_rate": 7.240033631961962e-06, | |
| "loss": 0.08345344662666321, | |
| "step": 95232 | |
| }, | |
| { | |
| "epoch": 2.7684903308109363, | |
| "grad_norm": 0.7562350034713745, | |
| "learning_rate": 7.232611405873997e-06, | |
| "loss": 0.08353827148675919, | |
| "step": 95488 | |
| }, | |
| { | |
| "epoch": 2.775912556898901, | |
| "grad_norm": 1.743805170059204, | |
| "learning_rate": 7.225218172856688e-06, | |
| "loss": 0.08417970687150955, | |
| "step": 95744 | |
| }, | |
| { | |
| "epoch": 2.783334782986866, | |
| "grad_norm": 1.4898192882537842, | |
| "learning_rate": 7.217795946768723e-06, | |
| "loss": 0.08327046781778336, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 2.7907570090748313, | |
| "grad_norm": 0.9221867918968201, | |
| "learning_rate": 7.210373720680758e-06, | |
| "loss": 0.08236086368560791, | |
| "step": 96256 | |
| }, | |
| { | |
| "epoch": 2.798179235162796, | |
| "grad_norm": 1.2286884784698486, | |
| "learning_rate": 7.2029514945927925e-06, | |
| "loss": 0.08261434733867645, | |
| "step": 96512 | |
| }, | |
| { | |
| "epoch": 2.805601461250761, | |
| "grad_norm": 0.7629371881484985, | |
| "learning_rate": 7.195529268504828e-06, | |
| "loss": 0.08190848678350449, | |
| "step": 96768 | |
| }, | |
| { | |
| "epoch": 2.813023687338726, | |
| "grad_norm": 0.9889212846755981, | |
| "learning_rate": 7.188107042416863e-06, | |
| "loss": 0.08423368632793427, | |
| "step": 97024 | |
| }, | |
| { | |
| "epoch": 2.820445913426691, | |
| "grad_norm": 1.484141230583191, | |
| "learning_rate": 7.1806848163288975e-06, | |
| "loss": 0.08237377554178238, | |
| "step": 97280 | |
| }, | |
| { | |
| "epoch": 2.827868139514656, | |
| "grad_norm": 1.1202856302261353, | |
| "learning_rate": 7.173262590240932e-06, | |
| "loss": 0.08046771585941315, | |
| "step": 97536 | |
| }, | |
| { | |
| "epoch": 2.8352903656026207, | |
| "grad_norm": 1.5413111448287964, | |
| "learning_rate": 7.165840364152969e-06, | |
| "loss": 0.08320514112710953, | |
| "step": 97792 | |
| }, | |
| { | |
| "epoch": 2.842712591690586, | |
| "grad_norm": 1.2286149263381958, | |
| "learning_rate": 7.158418138065003e-06, | |
| "loss": 0.08079677075147629, | |
| "step": 98048 | |
| }, | |
| { | |
| "epoch": 2.850134817778551, | |
| "grad_norm": 1.4287984371185303, | |
| "learning_rate": 7.150995911977038e-06, | |
| "loss": 0.08117518573999405, | |
| "step": 98304 | |
| }, | |
| { | |
| "epoch": 2.8575570438665157, | |
| "grad_norm": 0.7255827188491821, | |
| "learning_rate": 7.1435736858890736e-06, | |
| "loss": 0.08171623200178146, | |
| "step": 98560 | |
| }, | |
| { | |
| "epoch": 2.864979269954481, | |
| "grad_norm": 1.8379969596862793, | |
| "learning_rate": 7.136151459801108e-06, | |
| "loss": 0.0811542272567749, | |
| "step": 98816 | |
| }, | |
| { | |
| "epoch": 2.8724014960424458, | |
| "grad_norm": 0.9097608923912048, | |
| "learning_rate": 7.128729233713143e-06, | |
| "loss": 0.08120869100093842, | |
| "step": 99072 | |
| }, | |
| { | |
| "epoch": 2.8798237221304106, | |
| "grad_norm": 1.5451022386550903, | |
| "learning_rate": 7.121307007625178e-06, | |
| "loss": 0.0811024010181427, | |
| "step": 99328 | |
| }, | |
| { | |
| "epoch": 2.887245948218376, | |
| "grad_norm": 1.554936170578003, | |
| "learning_rate": 7.113884781537213e-06, | |
| "loss": 0.07978853583335876, | |
| "step": 99584 | |
| }, | |
| { | |
| "epoch": 2.8946681743063407, | |
| "grad_norm": 0.9636830687522888, | |
| "learning_rate": 7.1064915485199045e-06, | |
| "loss": 0.07950424402952194, | |
| "step": 99840 | |
| }, | |
| { | |
| "epoch": 2.9020904003943055, | |
| "grad_norm": 1.3318116664886475, | |
| "learning_rate": 7.099069322431939e-06, | |
| "loss": 0.08004064112901688, | |
| "step": 100096 | |
| }, | |
| { | |
| "epoch": 2.909512626482271, | |
| "grad_norm": 1.3676046133041382, | |
| "learning_rate": 7.091647096343974e-06, | |
| "loss": 0.0820651426911354, | |
| "step": 100352 | |
| }, | |
| { | |
| "epoch": 2.9169348525702357, | |
| "grad_norm": 1.163332223892212, | |
| "learning_rate": 7.0842248702560094e-06, | |
| "loss": 0.08003416657447815, | |
| "step": 100608 | |
| }, | |
| { | |
| "epoch": 2.9243570786582005, | |
| "grad_norm": 0.9317662119865417, | |
| "learning_rate": 7.076802644168045e-06, | |
| "loss": 0.08074063807725906, | |
| "step": 100864 | |
| }, | |
| { | |
| "epoch": 2.9317793047461658, | |
| "grad_norm": 0.913896381855011, | |
| "learning_rate": 7.06938041808008e-06, | |
| "loss": 0.07903257012367249, | |
| "step": 101120 | |
| }, | |
| { | |
| "epoch": 2.9392015308341306, | |
| "grad_norm": 1.0926191806793213, | |
| "learning_rate": 7.061958191992114e-06, | |
| "loss": 0.08051859587430954, | |
| "step": 101376 | |
| }, | |
| { | |
| "epoch": 2.9466237569220954, | |
| "grad_norm": 1.3655383586883545, | |
| "learning_rate": 7.054535965904149e-06, | |
| "loss": 0.07894966751337051, | |
| "step": 101632 | |
| }, | |
| { | |
| "epoch": 2.9540459830100607, | |
| "grad_norm": 2.0877397060394287, | |
| "learning_rate": 7.04714273288684e-06, | |
| "loss": 0.07975777238607407, | |
| "step": 101888 | |
| }, | |
| { | |
| "epoch": 2.9614682090980256, | |
| "grad_norm": 1.5793073177337646, | |
| "learning_rate": 7.039720506798876e-06, | |
| "loss": 0.08113066852092743, | |
| "step": 102144 | |
| }, | |
| { | |
| "epoch": 2.9688904351859904, | |
| "grad_norm": 1.9515403509140015, | |
| "learning_rate": 7.032327273781567e-06, | |
| "loss": 0.07889179140329361, | |
| "step": 102400 | |
| }, | |
| { | |
| "epoch": 2.9763126612739557, | |
| "grad_norm": 0.9475187659263611, | |
| "learning_rate": 7.024905047693602e-06, | |
| "loss": 0.07782009243965149, | |
| "step": 102656 | |
| }, | |
| { | |
| "epoch": 2.9837348873619205, | |
| "grad_norm": 1.298964262008667, | |
| "learning_rate": 7.017482821605637e-06, | |
| "loss": 0.07958176732063293, | |
| "step": 102912 | |
| }, | |
| { | |
| "epoch": 2.9911571134498853, | |
| "grad_norm": 1.1061244010925293, | |
| "learning_rate": 7.010060595517671e-06, | |
| "loss": 0.07731985300779343, | |
| "step": 103168 | |
| }, | |
| { | |
| "epoch": 2.9985793395378506, | |
| "grad_norm": 1.25368332862854, | |
| "learning_rate": 7.002638369429708e-06, | |
| "loss": 0.07639598101377487, | |
| "step": 103424 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_f1_B-DISEASE": 0.8442475981480986, | |
| "eval_f1_B-PROCEDURE": 0.8438875990713314, | |
| "eval_f1_B-SYMPTOM": 0.8249060678515155, | |
| "eval_f1_I-DISEASE": 0.8754347692566669, | |
| "eval_f1_I-PROCEDURE": 0.8818953984838734, | |
| "eval_f1_I-SYMPTOM": 0.848827050936612, | |
| "eval_f1_O": 0.9377095345355524, | |
| "eval_f1_macro": 0.8652725740405215, | |
| "eval_f1_micro": 0.9047197910661565, | |
| "eval_loss": 0.06751061230897903, | |
| "eval_precision_B-DISEASE": 0.8504151390985552, | |
| "eval_precision_B-PROCEDURE": 0.8496466858325049, | |
| "eval_precision_B-SYMPTOM": 0.8356513003683447, | |
| "eval_precision_I-DISEASE": 0.8653930047537453, | |
| "eval_precision_I-PROCEDURE": 0.8778192990756047, | |
| "eval_precision_I-SYMPTOM": 0.8335410322818348, | |
| "eval_precision_O": 0.9469512052010233, | |
| "eval_precision_macro": 0.8656310952302304, | |
| "eval_precision_micro": 0.9060954239896774, | |
| "eval_rauc_macro": 0.9219726884290445, | |
| "eval_rauc_micro": 0.9435923090115219, | |
| "eval_recall_B-DISEASE": 0.8381688718847973, | |
| "eval_recall_B-PROCEDURE": 0.8382060593209107, | |
| "eval_recall_B-SYMPTOM": 0.8144336625908359, | |
| "eval_recall_I-DISEASE": 0.8857123130034522, | |
| "eval_recall_I-PROCEDURE": 0.8860095287149111, | |
| "eval_recall_I-SYMPTOM": 0.8646841931712581, | |
| "eval_recall_O": 0.9286465066498901, | |
| "eval_recall_macro": 0.8651230193337222, | |
| "eval_recall_micro": 0.9033483287791003, | |
| "eval_roc_auc_B-DISEASE": 0.9174920929627975, | |
| "eval_roc_auc_B-PROCEDURE": 0.9174971175161323, | |
| "eval_roc_auc_B-SYMPTOM": 0.9050940881877013, | |
| "eval_roc_auc_I-DISEASE": 0.9332803276470311, | |
| "eval_roc_auc_I-PROCEDURE": 0.9357933541086877, | |
| "eval_roc_auc_I-SYMPTOM": 0.9171147621275024, | |
| "eval_roc_auc_O": 0.9275370764534587, | |
| "eval_runtime": 54.0164, | |
| "eval_samples_per_second": 261.958, | |
| "eval_steps_per_second": 32.749, | |
| "step": 103473 | |
| }, | |
| { | |
| "epoch": 3.0060015656258154, | |
| "grad_norm": 1.659386396408081, | |
| "learning_rate": 6.995216143341742e-06, | |
| "loss": 0.0724564790725708, | |
| "step": 103680 | |
| }, | |
| { | |
| "epoch": 3.0134237917137803, | |
| "grad_norm": 2.2287073135375977, | |
| "learning_rate": 6.987793917253777e-06, | |
| "loss": 0.0720382034778595, | |
| "step": 103936 | |
| }, | |
| { | |
| "epoch": 3.0208460178017456, | |
| "grad_norm": 1.24174165725708, | |
| "learning_rate": 6.980371691165812e-06, | |
| "loss": 0.07275696098804474, | |
| "step": 104192 | |
| }, | |
| { | |
| "epoch": 3.0282682438897104, | |
| "grad_norm": 1.0244415998458862, | |
| "learning_rate": 6.972949465077847e-06, | |
| "loss": 0.07141397893428802, | |
| "step": 104448 | |
| }, | |
| { | |
| "epoch": 3.035690469977675, | |
| "grad_norm": 1.217551827430725, | |
| "learning_rate": 6.965527238989882e-06, | |
| "loss": 0.07081952691078186, | |
| "step": 104704 | |
| }, | |
| { | |
| "epoch": 3.0431126960656405, | |
| "grad_norm": 0.9922760128974915, | |
| "learning_rate": 6.958105012901917e-06, | |
| "loss": 0.07204527407884598, | |
| "step": 104960 | |
| }, | |
| { | |
| "epoch": 3.0505349221536053, | |
| "grad_norm": 1.2190998792648315, | |
| "learning_rate": 6.950682786813952e-06, | |
| "loss": 0.07317464053630829, | |
| "step": 105216 | |
| }, | |
| { | |
| "epoch": 3.05795714824157, | |
| "grad_norm": 1.0417033433914185, | |
| "learning_rate": 6.943260560725987e-06, | |
| "loss": 0.07095565646886826, | |
| "step": 105472 | |
| }, | |
| { | |
| "epoch": 3.0653793743295354, | |
| "grad_norm": 1.5745536088943481, | |
| "learning_rate": 6.935838334638022e-06, | |
| "loss": 0.07097219675779343, | |
| "step": 105728 | |
| }, | |
| { | |
| "epoch": 3.0728016004175003, | |
| "grad_norm": 0.9291413426399231, | |
| "learning_rate": 6.9284161085500564e-06, | |
| "loss": 0.07165142148733139, | |
| "step": 105984 | |
| }, | |
| { | |
| "epoch": 3.080223826505465, | |
| "grad_norm": 3.8608274459838867, | |
| "learning_rate": 6.921022875532748e-06, | |
| "loss": 0.0695074051618576, | |
| "step": 106240 | |
| }, | |
| { | |
| "epoch": 3.0876460525934304, | |
| "grad_norm": 2.1987979412078857, | |
| "learning_rate": 6.913600649444782e-06, | |
| "loss": 0.07008818536996841, | |
| "step": 106496 | |
| }, | |
| { | |
| "epoch": 3.0950682786813952, | |
| "grad_norm": 1.1189053058624268, | |
| "learning_rate": 6.906178423356819e-06, | |
| "loss": 0.07337635010480881, | |
| "step": 106752 | |
| }, | |
| { | |
| "epoch": 3.10249050476936, | |
| "grad_norm": 2.2442538738250732, | |
| "learning_rate": 6.8987561972688535e-06, | |
| "loss": 0.07152988761663437, | |
| "step": 107008 | |
| }, | |
| { | |
| "epoch": 3.109912730857325, | |
| "grad_norm": 1.84907865524292, | |
| "learning_rate": 6.891333971180888e-06, | |
| "loss": 0.07006312906742096, | |
| "step": 107264 | |
| }, | |
| { | |
| "epoch": 3.11733495694529, | |
| "grad_norm": 0.8491196036338806, | |
| "learning_rate": 6.883911745092924e-06, | |
| "loss": 0.07236039638519287, | |
| "step": 107520 | |
| }, | |
| { | |
| "epoch": 3.124757183033255, | |
| "grad_norm": 2.2687666416168213, | |
| "learning_rate": 6.8764895190049585e-06, | |
| "loss": 0.06880185008049011, | |
| "step": 107776 | |
| }, | |
| { | |
| "epoch": 3.13217940912122, | |
| "grad_norm": 1.2143901586532593, | |
| "learning_rate": 6.869067292916993e-06, | |
| "loss": 0.07060759514570236, | |
| "step": 108032 | |
| }, | |
| { | |
| "epoch": 3.139601635209185, | |
| "grad_norm": 1.140826940536499, | |
| "learning_rate": 6.861645066829028e-06, | |
| "loss": 0.07119274884462357, | |
| "step": 108288 | |
| }, | |
| { | |
| "epoch": 3.14702386129715, | |
| "grad_norm": 0.9580644965171814, | |
| "learning_rate": 6.854251833811719e-06, | |
| "loss": 0.0698171928524971, | |
| "step": 108544 | |
| }, | |
| { | |
| "epoch": 3.154446087385115, | |
| "grad_norm": 0.8632366061210632, | |
| "learning_rate": 6.846829607723755e-06, | |
| "loss": 0.0699184387922287, | |
| "step": 108800 | |
| }, | |
| { | |
| "epoch": 3.16186831347308, | |
| "grad_norm": 0.9587484002113342, | |
| "learning_rate": 6.839407381635789e-06, | |
| "loss": 0.0697576180100441, | |
| "step": 109056 | |
| }, | |
| { | |
| "epoch": 3.169290539561045, | |
| "grad_norm": 1.705635666847229, | |
| "learning_rate": 6.831985155547824e-06, | |
| "loss": 0.07022682577371597, | |
| "step": 109312 | |
| }, | |
| { | |
| "epoch": 3.1767127656490097, | |
| "grad_norm": 1.2647652626037598, | |
| "learning_rate": 6.824562929459859e-06, | |
| "loss": 0.07101025432348251, | |
| "step": 109568 | |
| }, | |
| { | |
| "epoch": 3.184134991736975, | |
| "grad_norm": 1.1325069665908813, | |
| "learning_rate": 6.817140703371895e-06, | |
| "loss": 0.07074710726737976, | |
| "step": 109824 | |
| }, | |
| { | |
| "epoch": 3.19155721782494, | |
| "grad_norm": 0.8290653228759766, | |
| "learning_rate": 6.80971847728393e-06, | |
| "loss": 0.07273830473423004, | |
| "step": 110080 | |
| }, | |
| { | |
| "epoch": 3.1989794439129047, | |
| "grad_norm": 1.3392220735549927, | |
| "learning_rate": 6.802296251195965e-06, | |
| "loss": 0.07128596305847168, | |
| "step": 110336 | |
| }, | |
| { | |
| "epoch": 3.20640167000087, | |
| "grad_norm": 1.2070839405059814, | |
| "learning_rate": 6.794903018178656e-06, | |
| "loss": 0.07207676023244858, | |
| "step": 110592 | |
| }, | |
| { | |
| "epoch": 3.213823896088835, | |
| "grad_norm": 2.795504570007324, | |
| "learning_rate": 6.7874807920906906e-06, | |
| "loss": 0.06816298514604568, | |
| "step": 110848 | |
| }, | |
| { | |
| "epoch": 3.2212461221767996, | |
| "grad_norm": 0.9610592722892761, | |
| "learning_rate": 6.780058566002726e-06, | |
| "loss": 0.06956319510936737, | |
| "step": 111104 | |
| }, | |
| { | |
| "epoch": 3.228668348264765, | |
| "grad_norm": 1.0557652711868286, | |
| "learning_rate": 6.772636339914761e-06, | |
| "loss": 0.06906887143850327, | |
| "step": 111360 | |
| }, | |
| { | |
| "epoch": 3.2360905743527297, | |
| "grad_norm": 1.55743408203125, | |
| "learning_rate": 6.7652141138267955e-06, | |
| "loss": 0.07004261761903763, | |
| "step": 111616 | |
| }, | |
| { | |
| "epoch": 3.2435128004406946, | |
| "grad_norm": 1.1939220428466797, | |
| "learning_rate": 6.75779188773883e-06, | |
| "loss": 0.07050329446792603, | |
| "step": 111872 | |
| }, | |
| { | |
| "epoch": 3.25093502652866, | |
| "grad_norm": 5.080704689025879, | |
| "learning_rate": 6.750369661650867e-06, | |
| "loss": 0.0688510313630104, | |
| "step": 112128 | |
| }, | |
| { | |
| "epoch": 3.2583572526166247, | |
| "grad_norm": 1.5225037336349487, | |
| "learning_rate": 6.742947435562901e-06, | |
| "loss": 0.07025352865457535, | |
| "step": 112384 | |
| }, | |
| { | |
| "epoch": 3.2657794787045895, | |
| "grad_norm": 1.0847176313400269, | |
| "learning_rate": 6.735583195616249e-06, | |
| "loss": 0.07004053890705109, | |
| "step": 112640 | |
| }, | |
| { | |
| "epoch": 3.2732017047925543, | |
| "grad_norm": 1.6107685565948486, | |
| "learning_rate": 6.728160969528284e-06, | |
| "loss": 0.06852912157773972, | |
| "step": 112896 | |
| }, | |
| { | |
| "epoch": 3.2806239308805196, | |
| "grad_norm": 1.239355206489563, | |
| "learning_rate": 6.7207387434403186e-06, | |
| "loss": 0.06782490760087967, | |
| "step": 113152 | |
| }, | |
| { | |
| "epoch": 3.2880461569684845, | |
| "grad_norm": 2.2779667377471924, | |
| "learning_rate": 6.713316517352353e-06, | |
| "loss": 0.06995242834091187, | |
| "step": 113408 | |
| }, | |
| { | |
| "epoch": 3.2954683830564493, | |
| "grad_norm": 2.5409092903137207, | |
| "learning_rate": 6.705894291264389e-06, | |
| "loss": 0.06850997358560562, | |
| "step": 113664 | |
| }, | |
| { | |
| "epoch": 3.3028906091444146, | |
| "grad_norm": 0.9244787693023682, | |
| "learning_rate": 6.6984720651764235e-06, | |
| "loss": 0.06772230565547943, | |
| "step": 113920 | |
| }, | |
| { | |
| "epoch": 3.3103128352323794, | |
| "grad_norm": 1.1681801080703735, | |
| "learning_rate": 6.691049839088458e-06, | |
| "loss": 0.06610868871212006, | |
| "step": 114176 | |
| }, | |
| { | |
| "epoch": 3.3177350613203442, | |
| "grad_norm": 0.9614848494529724, | |
| "learning_rate": 6.683627613000493e-06, | |
| "loss": 0.06784440577030182, | |
| "step": 114432 | |
| }, | |
| { | |
| "epoch": 3.3251572874083095, | |
| "grad_norm": 0.9045601487159729, | |
| "learning_rate": 6.6762053869125285e-06, | |
| "loss": 0.06725604832172394, | |
| "step": 114688 | |
| }, | |
| { | |
| "epoch": 3.3325795134962743, | |
| "grad_norm": 1.1413661241531372, | |
| "learning_rate": 6.668783160824563e-06, | |
| "loss": 0.06764224916696548, | |
| "step": 114944 | |
| }, | |
| { | |
| "epoch": 3.340001739584239, | |
| "grad_norm": 2.1499950885772705, | |
| "learning_rate": 6.661360934736598e-06, | |
| "loss": 0.06848239153623581, | |
| "step": 115200 | |
| }, | |
| { | |
| "epoch": 3.3474239656722045, | |
| "grad_norm": 1.384770393371582, | |
| "learning_rate": 6.653967701719289e-06, | |
| "loss": 0.06827209889888763, | |
| "step": 115456 | |
| }, | |
| { | |
| "epoch": 3.3548461917601693, | |
| "grad_norm": 1.5378497838974, | |
| "learning_rate": 6.646545475631324e-06, | |
| "loss": 0.06868505477905273, | |
| "step": 115712 | |
| }, | |
| { | |
| "epoch": 3.362268417848134, | |
| "grad_norm": 1.9188313484191895, | |
| "learning_rate": 6.63912324954336e-06, | |
| "loss": 0.0663612112402916, | |
| "step": 115968 | |
| }, | |
| { | |
| "epoch": 3.3696906439360994, | |
| "grad_norm": 0.8880401253700256, | |
| "learning_rate": 6.631701023455395e-06, | |
| "loss": 0.06696955859661102, | |
| "step": 116224 | |
| }, | |
| { | |
| "epoch": 3.3771128700240642, | |
| "grad_norm": 1.6643853187561035, | |
| "learning_rate": 6.62427879736743e-06, | |
| "loss": 0.06680089980363846, | |
| "step": 116480 | |
| }, | |
| { | |
| "epoch": 3.384535096112029, | |
| "grad_norm": 3.724933385848999, | |
| "learning_rate": 6.616856571279465e-06, | |
| "loss": 0.06643035262823105, | |
| "step": 116736 | |
| }, | |
| { | |
| "epoch": 3.3919573221999944, | |
| "grad_norm": 1.1622769832611084, | |
| "learning_rate": 6.6094343451915e-06, | |
| "loss": 0.06458131223917007, | |
| "step": 116992 | |
| }, | |
| { | |
| "epoch": 3.399379548287959, | |
| "grad_norm": 1.4477862119674683, | |
| "learning_rate": 6.602012119103535e-06, | |
| "loss": 0.06673602014780045, | |
| "step": 117248 | |
| }, | |
| { | |
| "epoch": 3.406801774375924, | |
| "grad_norm": 1.3256279230117798, | |
| "learning_rate": 6.594589893015569e-06, | |
| "loss": 0.06612307578325272, | |
| "step": 117504 | |
| }, | |
| { | |
| "epoch": 3.4142240004638893, | |
| "grad_norm": 1.2756476402282715, | |
| "learning_rate": 6.587167666927605e-06, | |
| "loss": 0.06554066389799118, | |
| "step": 117760 | |
| }, | |
| { | |
| "epoch": 3.421646226551854, | |
| "grad_norm": 1.5321786403656006, | |
| "learning_rate": 6.5797454408396396e-06, | |
| "loss": 0.06720390915870667, | |
| "step": 118016 | |
| }, | |
| { | |
| "epoch": 3.429068452639819, | |
| "grad_norm": 13.644562721252441, | |
| "learning_rate": 6.572323214751674e-06, | |
| "loss": 0.06655475497245789, | |
| "step": 118272 | |
| }, | |
| { | |
| "epoch": 3.4364906787277842, | |
| "grad_norm": 1.0022516250610352, | |
| "learning_rate": 6.564900988663709e-06, | |
| "loss": 0.06593410670757294, | |
| "step": 118528 | |
| }, | |
| { | |
| "epoch": 3.443912904815749, | |
| "grad_norm": 1.2650278806686401, | |
| "learning_rate": 6.557478762575745e-06, | |
| "loss": 0.06658318638801575, | |
| "step": 118784 | |
| }, | |
| { | |
| "epoch": 3.451335130903714, | |
| "grad_norm": 1.8073201179504395, | |
| "learning_rate": 6.55005653648778e-06, | |
| "loss": 0.06579141318798065, | |
| "step": 119040 | |
| }, | |
| { | |
| "epoch": 3.458757356991679, | |
| "grad_norm": 1.5507023334503174, | |
| "learning_rate": 6.542663303470471e-06, | |
| "loss": 0.06475941836833954, | |
| "step": 119296 | |
| }, | |
| { | |
| "epoch": 3.466179583079644, | |
| "grad_norm": 0.9973443150520325, | |
| "learning_rate": 6.535241077382506e-06, | |
| "loss": 0.06540407240390778, | |
| "step": 119552 | |
| }, | |
| { | |
| "epoch": 3.473601809167609, | |
| "grad_norm": 1.1612350940704346, | |
| "learning_rate": 6.527818851294541e-06, | |
| "loss": 0.06724215298891068, | |
| "step": 119808 | |
| }, | |
| { | |
| "epoch": 3.481024035255574, | |
| "grad_norm": 1.2308496236801147, | |
| "learning_rate": 6.520396625206576e-06, | |
| "loss": 0.06539810448884964, | |
| "step": 120064 | |
| }, | |
| { | |
| "epoch": 3.488446261343539, | |
| "grad_norm": 0.9591255187988281, | |
| "learning_rate": 6.512974399118611e-06, | |
| "loss": 0.06621332466602325, | |
| "step": 120320 | |
| }, | |
| { | |
| "epoch": 3.495868487431504, | |
| "grad_norm": 1.1396393775939941, | |
| "learning_rate": 6.505552173030646e-06, | |
| "loss": 0.0646854117512703, | |
| "step": 120576 | |
| }, | |
| { | |
| "epoch": 3.503290713519469, | |
| "grad_norm": 1.7772917747497559, | |
| "learning_rate": 6.498129946942682e-06, | |
| "loss": 0.06502901762723923, | |
| "step": 120832 | |
| }, | |
| { | |
| "epoch": 3.510712939607434, | |
| "grad_norm": 1.9355366230010986, | |
| "learning_rate": 6.490707720854717e-06, | |
| "loss": 0.06590066105127335, | |
| "step": 121088 | |
| }, | |
| { | |
| "epoch": 3.5181351656953987, | |
| "grad_norm": 1.1407538652420044, | |
| "learning_rate": 6.4832854947667515e-06, | |
| "loss": 0.06572704762220383, | |
| "step": 121344 | |
| }, | |
| { | |
| "epoch": 3.525557391783364, | |
| "grad_norm": 5.394893169403076, | |
| "learning_rate": 6.475863268678786e-06, | |
| "loss": 0.06465983390808105, | |
| "step": 121600 | |
| }, | |
| { | |
| "epoch": 3.532979617871329, | |
| "grad_norm": 1.164096474647522, | |
| "learning_rate": 6.4684700356614775e-06, | |
| "loss": 0.06483496725559235, | |
| "step": 121856 | |
| }, | |
| { | |
| "epoch": 3.5404018439592937, | |
| "grad_norm": 1.5464900732040405, | |
| "learning_rate": 6.461047809573513e-06, | |
| "loss": 0.06383845210075378, | |
| "step": 122112 | |
| }, | |
| { | |
| "epoch": 3.547824070047259, | |
| "grad_norm": 0.8942134976387024, | |
| "learning_rate": 6.453625583485548e-06, | |
| "loss": 0.0651378259062767, | |
| "step": 122368 | |
| }, | |
| { | |
| "epoch": 3.555246296135224, | |
| "grad_norm": 1.5484808683395386, | |
| "learning_rate": 6.4462033573975824e-06, | |
| "loss": 0.06251890957355499, | |
| "step": 122624 | |
| }, | |
| { | |
| "epoch": 3.5626685222231886, | |
| "grad_norm": 2.1657159328460693, | |
| "learning_rate": 6.438781131309617e-06, | |
| "loss": 0.0626024454832077, | |
| "step": 122880 | |
| }, | |
| { | |
| "epoch": 3.5700907483111535, | |
| "grad_norm": 1.1823967695236206, | |
| "learning_rate": 6.431358905221653e-06, | |
| "loss": 0.06294266134500504, | |
| "step": 123136 | |
| }, | |
| { | |
| "epoch": 3.5775129743991188, | |
| "grad_norm": 1.2693891525268555, | |
| "learning_rate": 6.423936679133687e-06, | |
| "loss": 0.06377562880516052, | |
| "step": 123392 | |
| }, | |
| { | |
| "epoch": 3.5849352004870836, | |
| "grad_norm": 1.1565278768539429, | |
| "learning_rate": 6.416514453045722e-06, | |
| "loss": 0.06456663459539413, | |
| "step": 123648 | |
| }, | |
| { | |
| "epoch": 3.5923574265750484, | |
| "grad_norm": 1.4107431173324585, | |
| "learning_rate": 6.409121220028413e-06, | |
| "loss": 0.06456474959850311, | |
| "step": 123904 | |
| }, | |
| { | |
| "epoch": 3.5997796526630137, | |
| "grad_norm": 1.2529884576797485, | |
| "learning_rate": 6.401698993940448e-06, | |
| "loss": 0.06332822889089584, | |
| "step": 124160 | |
| }, | |
| { | |
| "epoch": 3.6072018787509785, | |
| "grad_norm": 1.263092041015625, | |
| "learning_rate": 6.3942767678524845e-06, | |
| "loss": 0.06439608335494995, | |
| "step": 124416 | |
| }, | |
| { | |
| "epoch": 3.6146241048389434, | |
| "grad_norm": 1.6909823417663574, | |
| "learning_rate": 6.386854541764519e-06, | |
| "loss": 0.06379682570695877, | |
| "step": 124672 | |
| }, | |
| { | |
| "epoch": 3.622046330926908, | |
| "grad_norm": 1.1804763078689575, | |
| "learning_rate": 6.379432315676554e-06, | |
| "loss": 0.06351881474256516, | |
| "step": 124928 | |
| }, | |
| { | |
| "epoch": 3.6294685570148735, | |
| "grad_norm": 0.8701578974723816, | |
| "learning_rate": 6.372010089588589e-06, | |
| "loss": 0.062777079641819, | |
| "step": 125184 | |
| }, | |
| { | |
| "epoch": 3.6368907831028383, | |
| "grad_norm": 1.331722378730774, | |
| "learning_rate": 6.364587863500624e-06, | |
| "loss": 0.06259626895189285, | |
| "step": 125440 | |
| }, | |
| { | |
| "epoch": 3.644313009190803, | |
| "grad_norm": 1.3232914209365845, | |
| "learning_rate": 6.357165637412659e-06, | |
| "loss": 0.06263776123523712, | |
| "step": 125696 | |
| }, | |
| { | |
| "epoch": 3.6517352352787684, | |
| "grad_norm": 2.3550124168395996, | |
| "learning_rate": 6.3497434113246935e-06, | |
| "loss": 0.06342460960149765, | |
| "step": 125952 | |
| }, | |
| { | |
| "epoch": 3.6591574613667333, | |
| "grad_norm": 1.2501484155654907, | |
| "learning_rate": 6.342350178307385e-06, | |
| "loss": 0.06399497389793396, | |
| "step": 126208 | |
| }, | |
| { | |
| "epoch": 3.666579687454698, | |
| "grad_norm": 1.6568396091461182, | |
| "learning_rate": 6.3349279522194195e-06, | |
| "loss": 0.06301023066043854, | |
| "step": 126464 | |
| }, | |
| { | |
| "epoch": 3.6740019135426634, | |
| "grad_norm": 3.808454990386963, | |
| "learning_rate": 6.327505726131455e-06, | |
| "loss": 0.06246251240372658, | |
| "step": 126720 | |
| }, | |
| { | |
| "epoch": 3.681424139630628, | |
| "grad_norm": 1.1286664009094238, | |
| "learning_rate": 6.32008350004349e-06, | |
| "loss": 0.061383429914712906, | |
| "step": 126976 | |
| }, | |
| { | |
| "epoch": 3.688846365718593, | |
| "grad_norm": 2.1331582069396973, | |
| "learning_rate": 6.3126612739555245e-06, | |
| "loss": 0.061110664159059525, | |
| "step": 127232 | |
| }, | |
| { | |
| "epoch": 3.6962685918065583, | |
| "grad_norm": 1.4336588382720947, | |
| "learning_rate": 6.305239047867561e-06, | |
| "loss": 0.060820192098617554, | |
| "step": 127488 | |
| }, | |
| { | |
| "epoch": 3.703690817894523, | |
| "grad_norm": 1.408690333366394, | |
| "learning_rate": 6.297816821779596e-06, | |
| "loss": 0.06147029623389244, | |
| "step": 127744 | |
| }, | |
| { | |
| "epoch": 3.711113043982488, | |
| "grad_norm": 1.0031101703643799, | |
| "learning_rate": 6.29039459569163e-06, | |
| "loss": 0.06205154210329056, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 3.7185352700704533, | |
| "grad_norm": 1.0147720575332642, | |
| "learning_rate": 6.2830013626743215e-06, | |
| "loss": 0.06094561889767647, | |
| "step": 128256 | |
| }, | |
| { | |
| "epoch": 3.725957496158418, | |
| "grad_norm": 1.0210496187210083, | |
| "learning_rate": 6.275579136586356e-06, | |
| "loss": 0.060289010405540466, | |
| "step": 128512 | |
| }, | |
| { | |
| "epoch": 3.733379722246383, | |
| "grad_norm": 1.4809277057647705, | |
| "learning_rate": 6.268156910498392e-06, | |
| "loss": 0.06048288941383362, | |
| "step": 128768 | |
| }, | |
| { | |
| "epoch": 3.740801948334348, | |
| "grad_norm": 1.2163718938827515, | |
| "learning_rate": 6.2607346844104265e-06, | |
| "loss": 0.06114616617560387, | |
| "step": 129024 | |
| }, | |
| { | |
| "epoch": 3.748224174422313, | |
| "grad_norm": 1.1666347980499268, | |
| "learning_rate": 6.253312458322461e-06, | |
| "loss": 0.06302626430988312, | |
| "step": 129280 | |
| }, | |
| { | |
| "epoch": 3.755646400510278, | |
| "grad_norm": 0.9073694348335266, | |
| "learning_rate": 6.245890232234496e-06, | |
| "loss": 0.06126226857304573, | |
| "step": 129536 | |
| }, | |
| { | |
| "epoch": 3.763068626598243, | |
| "grad_norm": 0.8603505492210388, | |
| "learning_rate": 6.238468006146532e-06, | |
| "loss": 0.06161976233124733, | |
| "step": 129792 | |
| }, | |
| { | |
| "epoch": 3.770490852686208, | |
| "grad_norm": 1.128769040107727, | |
| "learning_rate": 6.231045780058567e-06, | |
| "loss": 0.06006836146116257, | |
| "step": 130048 | |
| }, | |
| { | |
| "epoch": 3.777913078774173, | |
| "grad_norm": 1.206339955329895, | |
| "learning_rate": 6.223623553970602e-06, | |
| "loss": 0.060498420149087906, | |
| "step": 130304 | |
| }, | |
| { | |
| "epoch": 3.785335304862138, | |
| "grad_norm": 1.249027132987976, | |
| "learning_rate": 6.216230320953293e-06, | |
| "loss": 0.05888880789279938, | |
| "step": 130560 | |
| }, | |
| { | |
| "epoch": 3.792757530950103, | |
| "grad_norm": 0.8686812520027161, | |
| "learning_rate": 6.208808094865328e-06, | |
| "loss": 0.05897178500890732, | |
| "step": 130816 | |
| }, | |
| { | |
| "epoch": 3.8001797570380678, | |
| "grad_norm": 0.6762338280677795, | |
| "learning_rate": 6.201385868777363e-06, | |
| "loss": 0.05932297185063362, | |
| "step": 131072 | |
| }, | |
| { | |
| "epoch": 3.807601983126033, | |
| "grad_norm": 1.6166266202926636, | |
| "learning_rate": 6.193963642689398e-06, | |
| "loss": 0.06046015024185181, | |
| "step": 131328 | |
| }, | |
| { | |
| "epoch": 3.815024209213998, | |
| "grad_norm": 1.1673097610473633, | |
| "learning_rate": 6.186541416601433e-06, | |
| "loss": 0.06021393463015556, | |
| "step": 131584 | |
| }, | |
| { | |
| "epoch": 3.8224464353019627, | |
| "grad_norm": 1.8146482706069946, | |
| "learning_rate": 6.179119190513467e-06, | |
| "loss": 0.05905488133430481, | |
| "step": 131840 | |
| }, | |
| { | |
| "epoch": 3.829868661389928, | |
| "grad_norm": 1.2830371856689453, | |
| "learning_rate": 6.171696964425503e-06, | |
| "loss": 0.059033047407865524, | |
| "step": 132096 | |
| }, | |
| { | |
| "epoch": 3.837290887477893, | |
| "grad_norm": 1.280580759048462, | |
| "learning_rate": 6.164274738337538e-06, | |
| "loss": 0.059182584285736084, | |
| "step": 132352 | |
| }, | |
| { | |
| "epoch": 3.8447131135658577, | |
| "grad_norm": 1.4833192825317383, | |
| "learning_rate": 6.156852512249572e-06, | |
| "loss": 0.060947176069021225, | |
| "step": 132608 | |
| }, | |
| { | |
| "epoch": 3.852135339653823, | |
| "grad_norm": 1.5997072458267212, | |
| "learning_rate": 6.1494592792322636e-06, | |
| "loss": 0.058467965573072433, | |
| "step": 132864 | |
| }, | |
| { | |
| "epoch": 3.8595575657417878, | |
| "grad_norm": 1.6831973791122437, | |
| "learning_rate": 6.142037053144298e-06, | |
| "loss": 0.05896062031388283, | |
| "step": 133120 | |
| }, | |
| { | |
| "epoch": 3.8669797918297526, | |
| "grad_norm": 2.135114908218384, | |
| "learning_rate": 6.134614827056335e-06, | |
| "loss": 0.0595758892595768, | |
| "step": 133376 | |
| }, | |
| { | |
| "epoch": 3.874402017917718, | |
| "grad_norm": 2.515566110610962, | |
| "learning_rate": 6.127192600968369e-06, | |
| "loss": 0.05884886533021927, | |
| "step": 133632 | |
| }, | |
| { | |
| "epoch": 3.8818242440056827, | |
| "grad_norm": 1.1714390516281128, | |
| "learning_rate": 6.119770374880404e-06, | |
| "loss": 0.058262407779693604, | |
| "step": 133888 | |
| }, | |
| { | |
| "epoch": 3.8892464700936475, | |
| "grad_norm": 0.7722835540771484, | |
| "learning_rate": 6.11234814879244e-06, | |
| "loss": 0.05952470749616623, | |
| "step": 134144 | |
| }, | |
| { | |
| "epoch": 3.896668696181613, | |
| "grad_norm": 1.1208815574645996, | |
| "learning_rate": 6.104925922704474e-06, | |
| "loss": 0.06004638597369194, | |
| "step": 134400 | |
| }, | |
| { | |
| "epoch": 3.9040909222695777, | |
| "grad_norm": 1.2898955345153809, | |
| "learning_rate": 6.097503696616509e-06, | |
| "loss": 0.059184350073337555, | |
| "step": 134656 | |
| }, | |
| { | |
| "epoch": 3.9115131483575425, | |
| "grad_norm": 0.9606613516807556, | |
| "learning_rate": 6.090081470528544e-06, | |
| "loss": 0.0595993809401989, | |
| "step": 134912 | |
| }, | |
| { | |
| "epoch": 3.9189353744455078, | |
| "grad_norm": 1.0130254030227661, | |
| "learning_rate": 6.082688237511235e-06, | |
| "loss": 0.05890991911292076, | |
| "step": 135168 | |
| }, | |
| { | |
| "epoch": 3.9263576005334726, | |
| "grad_norm": 1.2042089700698853, | |
| "learning_rate": 6.0752660114232706e-06, | |
| "loss": 0.057879913598299026, | |
| "step": 135424 | |
| }, | |
| { | |
| "epoch": 3.9337798266214374, | |
| "grad_norm": 1.8819071054458618, | |
| "learning_rate": 6.067843785335305e-06, | |
| "loss": 0.05783630907535553, | |
| "step": 135680 | |
| }, | |
| { | |
| "epoch": 3.9412020527094027, | |
| "grad_norm": 1.1659451723098755, | |
| "learning_rate": 6.06042155924734e-06, | |
| "loss": 0.05880802497267723, | |
| "step": 135936 | |
| }, | |
| { | |
| "epoch": 3.9486242787973675, | |
| "grad_norm": 0.6728376150131226, | |
| "learning_rate": 6.052999333159375e-06, | |
| "loss": 0.05816688388586044, | |
| "step": 136192 | |
| }, | |
| { | |
| "epoch": 3.9560465048853324, | |
| "grad_norm": 4.910149574279785, | |
| "learning_rate": 6.045577107071411e-06, | |
| "loss": 0.05902698263525963, | |
| "step": 136448 | |
| }, | |
| { | |
| "epoch": 3.963468730973297, | |
| "grad_norm": 1.8614972829818726, | |
| "learning_rate": 6.038154880983446e-06, | |
| "loss": 0.05790010467171669, | |
| "step": 136704 | |
| }, | |
| { | |
| "epoch": 3.9708909570612625, | |
| "grad_norm": 2.0103042125701904, | |
| "learning_rate": 6.0307326548954805e-06, | |
| "loss": 0.05860760807991028, | |
| "step": 136960 | |
| }, | |
| { | |
| "epoch": 3.9783131831492273, | |
| "grad_norm": 1.8316086530685425, | |
| "learning_rate": 6.023310428807515e-06, | |
| "loss": 0.05606749281287193, | |
| "step": 137216 | |
| }, | |
| { | |
| "epoch": 3.985735409237192, | |
| "grad_norm": 0.9380112290382385, | |
| "learning_rate": 6.015888202719551e-06, | |
| "loss": 0.05585785582661629, | |
| "step": 137472 | |
| }, | |
| { | |
| "epoch": 3.9931576353251574, | |
| "grad_norm": 0.7350268363952637, | |
| "learning_rate": 6.008494969702242e-06, | |
| "loss": 0.05835752934217453, | |
| "step": 137728 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_f1_B-DISEASE": 0.858496024110975, | |
| "eval_f1_B-PROCEDURE": 0.8520623036299217, | |
| "eval_f1_B-SYMPTOM": 0.8441152081375422, | |
| "eval_f1_I-DISEASE": 0.8905982515806533, | |
| "eval_f1_I-PROCEDURE": 0.8908012509435997, | |
| "eval_f1_I-SYMPTOM": 0.8672686170664062, | |
| "eval_f1_O": 0.943749510567655, | |
| "eval_f1_macro": 0.8781558808623934, | |
| "eval_f1_micro": 0.9148746623728682, | |
| "eval_loss": 0.06755498051643372, | |
| "eval_precision_B-DISEASE": 0.8696469946844759, | |
| "eval_precision_B-PROCEDURE": 0.8596363440164023, | |
| "eval_precision_B-SYMPTOM": 0.8581480565609334, | |
| "eval_precision_I-DISEASE": 0.8833197500679163, | |
| "eval_precision_I-PROCEDURE": 0.8952616291672085, | |
| "eval_precision_I-SYMPTOM": 0.8727928201430658, | |
| "eval_precision_O": 0.9455461223813957, | |
| "eval_precision_macro": 0.8834788167173426, | |
| "eval_precision_micro": 0.9170394475978084, | |
| "eval_rauc_macro": 0.9267200320416903, | |
| "eval_rauc_micro": 0.949232093592823, | |
| "eval_recall_B-DISEASE": 0.8476273978426059, | |
| "eval_recall_B-PROCEDURE": 0.844620563522146, | |
| "eval_recall_B-SYMPTOM": 0.8305339193246595, | |
| "eval_recall_I-DISEASE": 0.8979976985040277, | |
| "eval_recall_I-PROCEDURE": 0.8863850974332561, | |
| "eval_recall_I-SYMPTOM": 0.8618139032944633, | |
| "eval_recall_O": 0.9419597132127006, | |
| "eval_recall_macro": 0.8729911847334083, | |
| "eval_recall_micro": 0.9127200735654438, | |
| "eval_roc_auc_B-DISEASE": 0.9224414545297994, | |
| "eval_roc_auc_B-PROCEDURE": 0.9208171507948626, | |
| "eval_roc_auc_B-SYMPTOM": 0.9134475479522384, | |
| "eval_roc_auc_I-DISEASE": 0.940753974224099, | |
| "eval_roc_auc_I-PROCEDURE": 0.9371285009773118, | |
| "eval_roc_auc_I-SYMPTOM": 0.9198305139506655, | |
| "eval_roc_auc_O": 0.9326210818628553, | |
| "eval_runtime": 50.3166, | |
| "eval_samples_per_second": 281.219, | |
| "eval_steps_per_second": 35.157, | |
| "step": 137964 | |
| }, | |
| { | |
| "epoch": 4.000579861413122, | |
| "grad_norm": 0.9633011817932129, | |
| "learning_rate": 6.001072743614277e-06, | |
| "loss": 0.05613144487142563, | |
| "step": 137984 | |
| }, | |
| { | |
| "epoch": 4.0080020875010876, | |
| "grad_norm": 1.7201664447784424, | |
| "learning_rate": 5.993650517526311e-06, | |
| "loss": 0.051452554762363434, | |
| "step": 138240 | |
| }, | |
| { | |
| "epoch": 4.015424313589052, | |
| "grad_norm": 1.1845417022705078, | |
| "learning_rate": 5.986228291438346e-06, | |
| "loss": 0.051357295364141464, | |
| "step": 138496 | |
| }, | |
| { | |
| "epoch": 4.022846539677017, | |
| "grad_norm": 1.3591521978378296, | |
| "learning_rate": 5.9788060653503825e-06, | |
| "loss": 0.052172135561704636, | |
| "step": 138752 | |
| }, | |
| { | |
| "epoch": 4.0302687657649825, | |
| "grad_norm": 1.817952036857605, | |
| "learning_rate": 5.971412832333073e-06, | |
| "loss": 0.05360673740506172, | |
| "step": 139008 | |
| }, | |
| { | |
| "epoch": 4.037690991852947, | |
| "grad_norm": 1.7175657749176025, | |
| "learning_rate": 5.9639906062451085e-06, | |
| "loss": 0.05153278261423111, | |
| "step": 139264 | |
| }, | |
| { | |
| "epoch": 4.045113217940912, | |
| "grad_norm": 1.7851570844650269, | |
| "learning_rate": 5.956568380157143e-06, | |
| "loss": 0.05139975994825363, | |
| "step": 139520 | |
| }, | |
| { | |
| "epoch": 4.052535444028877, | |
| "grad_norm": 1.0840373039245605, | |
| "learning_rate": 5.949146154069178e-06, | |
| "loss": 0.0530707910656929, | |
| "step": 139776 | |
| }, | |
| { | |
| "epoch": 4.059957670116842, | |
| "grad_norm": 1.3619728088378906, | |
| "learning_rate": 5.9417239279812134e-06, | |
| "loss": 0.05101487785577774, | |
| "step": 140032 | |
| }, | |
| { | |
| "epoch": 4.067379896204807, | |
| "grad_norm": 1.3447620868682861, | |
| "learning_rate": 5.934301701893248e-06, | |
| "loss": 0.05232955515384674, | |
| "step": 140288 | |
| }, | |
| { | |
| "epoch": 4.074802122292772, | |
| "grad_norm": 1.3875212669372559, | |
| "learning_rate": 5.926879475805283e-06, | |
| "loss": 0.05181929096579552, | |
| "step": 140544 | |
| }, | |
| { | |
| "epoch": 4.082224348380737, | |
| "grad_norm": 4.292051792144775, | |
| "learning_rate": 5.9194572497173175e-06, | |
| "loss": 0.05065387114882469, | |
| "step": 140800 | |
| }, | |
| { | |
| "epoch": 4.089646574468702, | |
| "grad_norm": 1.77179753780365, | |
| "learning_rate": 5.912035023629353e-06, | |
| "loss": 0.05314262583851814, | |
| "step": 141056 | |
| }, | |
| { | |
| "epoch": 4.097068800556667, | |
| "grad_norm": 1.2534422874450684, | |
| "learning_rate": 5.904612797541388e-06, | |
| "loss": 0.05201578140258789, | |
| "step": 141312 | |
| }, | |
| { | |
| "epoch": 4.104491026644632, | |
| "grad_norm": 1.2724168300628662, | |
| "learning_rate": 5.8971905714534225e-06, | |
| "loss": 0.05249081552028656, | |
| "step": 141568 | |
| }, | |
| { | |
| "epoch": 4.111913252732597, | |
| "grad_norm": 1.055077075958252, | |
| "learning_rate": 5.889768345365459e-06, | |
| "loss": 0.05045388266444206, | |
| "step": 141824 | |
| }, | |
| { | |
| "epoch": 4.119335478820562, | |
| "grad_norm": 1.0675854682922363, | |
| "learning_rate": 5.882346119277494e-06, | |
| "loss": 0.051203496754169464, | |
| "step": 142080 | |
| }, | |
| { | |
| "epoch": 4.126757704908527, | |
| "grad_norm": 1.8462070226669312, | |
| "learning_rate": 5.874923893189528e-06, | |
| "loss": 0.05122574791312218, | |
| "step": 142336 | |
| }, | |
| { | |
| "epoch": 4.134179930996492, | |
| "grad_norm": 1.3658422231674194, | |
| "learning_rate": 5.867501667101563e-06, | |
| "loss": 0.04984445869922638, | |
| "step": 142592 | |
| }, | |
| { | |
| "epoch": 4.141602157084457, | |
| "grad_norm": 1.2870848178863525, | |
| "learning_rate": 5.8600794410135986e-06, | |
| "loss": 0.049068402498960495, | |
| "step": 142848 | |
| }, | |
| { | |
| "epoch": 4.149024383172422, | |
| "grad_norm": 1.0964611768722534, | |
| "learning_rate": 5.852657214925633e-06, | |
| "loss": 0.05185756832361221, | |
| "step": 143104 | |
| }, | |
| { | |
| "epoch": 4.156446609260387, | |
| "grad_norm": 1.0101662874221802, | |
| "learning_rate": 5.8452639819083245e-06, | |
| "loss": 0.051466602832078934, | |
| "step": 143360 | |
| }, | |
| { | |
| "epoch": 4.163868835348352, | |
| "grad_norm": 1.16269052028656, | |
| "learning_rate": 5.837841755820359e-06, | |
| "loss": 0.05081786960363388, | |
| "step": 143616 | |
| }, | |
| { | |
| "epoch": 4.171291061436317, | |
| "grad_norm": 0.876474142074585, | |
| "learning_rate": 5.830419529732394e-06, | |
| "loss": 0.04983343929052353, | |
| "step": 143872 | |
| }, | |
| { | |
| "epoch": 4.178713287524282, | |
| "grad_norm": 1.0422379970550537, | |
| "learning_rate": 5.8229973036444295e-06, | |
| "loss": 0.05065804719924927, | |
| "step": 144128 | |
| }, | |
| { | |
| "epoch": 4.186135513612246, | |
| "grad_norm": 1.0957989692687988, | |
| "learning_rate": 5.815575077556464e-06, | |
| "loss": 0.05084366351366043, | |
| "step": 144384 | |
| }, | |
| { | |
| "epoch": 4.1935577397002115, | |
| "grad_norm": 1.7821711301803589, | |
| "learning_rate": 5.808152851468499e-06, | |
| "loss": 0.0500972606241703, | |
| "step": 144640 | |
| }, | |
| { | |
| "epoch": 4.200979965788177, | |
| "grad_norm": 1.729944109916687, | |
| "learning_rate": 5.800730625380534e-06, | |
| "loss": 0.05059131607413292, | |
| "step": 144896 | |
| }, | |
| { | |
| "epoch": 4.208402191876141, | |
| "grad_norm": 1.7224652767181396, | |
| "learning_rate": 5.79330839929257e-06, | |
| "loss": 0.04996553435921669, | |
| "step": 145152 | |
| }, | |
| { | |
| "epoch": 4.2158244179641065, | |
| "grad_norm": 1.394722580909729, | |
| "learning_rate": 5.785886173204605e-06, | |
| "loss": 0.04998284578323364, | |
| "step": 145408 | |
| }, | |
| { | |
| "epoch": 4.223246644052072, | |
| "grad_norm": 1.2044475078582764, | |
| "learning_rate": 5.778492940187296e-06, | |
| "loss": 0.05008779466152191, | |
| "step": 145664 | |
| }, | |
| { | |
| "epoch": 4.230668870140036, | |
| "grad_norm": 1.5280312299728394, | |
| "learning_rate": 5.771070714099331e-06, | |
| "loss": 0.04947424307465553, | |
| "step": 145920 | |
| }, | |
| { | |
| "epoch": 4.238091096228001, | |
| "grad_norm": 2.591538667678833, | |
| "learning_rate": 5.763648488011365e-06, | |
| "loss": 0.05114565044641495, | |
| "step": 146176 | |
| }, | |
| { | |
| "epoch": 4.245513322315967, | |
| "grad_norm": 0.7985010147094727, | |
| "learning_rate": 5.756226261923401e-06, | |
| "loss": 0.05035054311156273, | |
| "step": 146432 | |
| }, | |
| { | |
| "epoch": 4.252935548403931, | |
| "grad_norm": 1.182391881942749, | |
| "learning_rate": 5.748804035835436e-06, | |
| "loss": 0.04990994557738304, | |
| "step": 146688 | |
| }, | |
| { | |
| "epoch": 4.260357774491896, | |
| "grad_norm": 0.974776029586792, | |
| "learning_rate": 5.74138180974747e-06, | |
| "loss": 0.04932263866066933, | |
| "step": 146944 | |
| }, | |
| { | |
| "epoch": 4.267780000579862, | |
| "grad_norm": 0.9236627221107483, | |
| "learning_rate": 5.733959583659507e-06, | |
| "loss": 0.04902668297290802, | |
| "step": 147200 | |
| }, | |
| { | |
| "epoch": 4.275202226667826, | |
| "grad_norm": 1.5337306261062622, | |
| "learning_rate": 5.7265373575715414e-06, | |
| "loss": 0.05148604139685631, | |
| "step": 147456 | |
| }, | |
| { | |
| "epoch": 4.282624452755791, | |
| "grad_norm": 0.8552817106246948, | |
| "learning_rate": 5.719144124554233e-06, | |
| "loss": 0.050737082958221436, | |
| "step": 147712 | |
| }, | |
| { | |
| "epoch": 4.290046678843757, | |
| "grad_norm": 1.3546576499938965, | |
| "learning_rate": 5.711750891536923e-06, | |
| "loss": 0.04890008643269539, | |
| "step": 147968 | |
| }, | |
| { | |
| "epoch": 4.297468904931721, | |
| "grad_norm": 1.5393311977386475, | |
| "learning_rate": 5.704328665448959e-06, | |
| "loss": 0.048157546669244766, | |
| "step": 148224 | |
| }, | |
| { | |
| "epoch": 4.304891131019686, | |
| "grad_norm": 1.0600520372390747, | |
| "learning_rate": 5.696906439360993e-06, | |
| "loss": 0.05047709867358208, | |
| "step": 148480 | |
| }, | |
| { | |
| "epoch": 4.3123133571076515, | |
| "grad_norm": 1.0730761289596558, | |
| "learning_rate": 5.689484213273028e-06, | |
| "loss": 0.049854230135679245, | |
| "step": 148736 | |
| }, | |
| { | |
| "epoch": 4.319735583195616, | |
| "grad_norm": 1.6848291158676147, | |
| "learning_rate": 5.682061987185064e-06, | |
| "loss": 0.050251930952072144, | |
| "step": 148992 | |
| }, | |
| { | |
| "epoch": 4.327157809283581, | |
| "grad_norm": 3.1243410110473633, | |
| "learning_rate": 5.674639761097098e-06, | |
| "loss": 0.04877758026123047, | |
| "step": 149248 | |
| }, | |
| { | |
| "epoch": 4.3345800353715465, | |
| "grad_norm": 1.2716270685195923, | |
| "learning_rate": 5.667217535009133e-06, | |
| "loss": 0.04907894879579544, | |
| "step": 149504 | |
| }, | |
| { | |
| "epoch": 4.342002261459511, | |
| "grad_norm": 0.9298461675643921, | |
| "learning_rate": 5.659795308921169e-06, | |
| "loss": 0.04849814996123314, | |
| "step": 149760 | |
| }, | |
| { | |
| "epoch": 4.349424487547476, | |
| "grad_norm": 1.7026667594909668, | |
| "learning_rate": 5.652373082833203e-06, | |
| "loss": 0.04887459799647331, | |
| "step": 150016 | |
| }, | |
| { | |
| "epoch": 4.356846713635441, | |
| "grad_norm": 1.5504307746887207, | |
| "learning_rate": 5.644950856745238e-06, | |
| "loss": 0.04786668345332146, | |
| "step": 150272 | |
| }, | |
| { | |
| "epoch": 4.364268939723406, | |
| "grad_norm": 1.0309391021728516, | |
| "learning_rate": 5.637528630657273e-06, | |
| "loss": 0.04870835691690445, | |
| "step": 150528 | |
| }, | |
| { | |
| "epoch": 4.371691165811371, | |
| "grad_norm": 0.9634639024734497, | |
| "learning_rate": 5.630106404569309e-06, | |
| "loss": 0.04876358062028885, | |
| "step": 150784 | |
| }, | |
| { | |
| "epoch": 4.379113391899336, | |
| "grad_norm": 0.8486874103546143, | |
| "learning_rate": 5.622684178481344e-06, | |
| "loss": 0.0493316575884819, | |
| "step": 151040 | |
| }, | |
| { | |
| "epoch": 4.386535617987301, | |
| "grad_norm": 0.8094356656074524, | |
| "learning_rate": 5.6152619523933785e-06, | |
| "loss": 0.048698797821998596, | |
| "step": 151296 | |
| }, | |
| { | |
| "epoch": 4.393957844075266, | |
| "grad_norm": 1.2043676376342773, | |
| "learning_rate": 5.607839726305413e-06, | |
| "loss": 0.04800259321928024, | |
| "step": 151552 | |
| }, | |
| { | |
| "epoch": 4.401380070163231, | |
| "grad_norm": 0.6494690775871277, | |
| "learning_rate": 5.600417500217449e-06, | |
| "loss": 0.049171607941389084, | |
| "step": 151808 | |
| }, | |
| { | |
| "epoch": 4.408802296251196, | |
| "grad_norm": 1.1811987161636353, | |
| "learning_rate": 5.59302426720014e-06, | |
| "loss": 0.04764533415436745, | |
| "step": 152064 | |
| }, | |
| { | |
| "epoch": 4.416224522339161, | |
| "grad_norm": 1.8065850734710693, | |
| "learning_rate": 5.585602041112175e-06, | |
| "loss": 0.04826079681515694, | |
| "step": 152320 | |
| }, | |
| { | |
| "epoch": 4.423646748427126, | |
| "grad_norm": 1.2280396223068237, | |
| "learning_rate": 5.5781798150242094e-06, | |
| "loss": 0.048462193459272385, | |
| "step": 152576 | |
| }, | |
| { | |
| "epoch": 4.431068974515091, | |
| "grad_norm": 0.8196685314178467, | |
| "learning_rate": 5.570757588936244e-06, | |
| "loss": 0.04874304309487343, | |
| "step": 152832 | |
| }, | |
| { | |
| "epoch": 4.438491200603056, | |
| "grad_norm": 0.90520840883255, | |
| "learning_rate": 5.56333536284828e-06, | |
| "loss": 0.04872792214155197, | |
| "step": 153088 | |
| }, | |
| { | |
| "epoch": 4.445913426691021, | |
| "grad_norm": 2.574571371078491, | |
| "learning_rate": 5.555942129830971e-06, | |
| "loss": 0.04761277139186859, | |
| "step": 153344 | |
| }, | |
| { | |
| "epoch": 4.453335652778986, | |
| "grad_norm": 0.7857698202133179, | |
| "learning_rate": 5.548519903743006e-06, | |
| "loss": 0.0485231950879097, | |
| "step": 153600 | |
| }, | |
| { | |
| "epoch": 4.460757878866951, | |
| "grad_norm": 1.9180569648742676, | |
| "learning_rate": 5.54109767765504e-06, | |
| "loss": 0.04640769585967064, | |
| "step": 153856 | |
| }, | |
| { | |
| "epoch": 4.468180104954916, | |
| "grad_norm": 0.9869174361228943, | |
| "learning_rate": 5.533675451567075e-06, | |
| "loss": 0.04795295000076294, | |
| "step": 154112 | |
| }, | |
| { | |
| "epoch": 4.4756023310428805, | |
| "grad_norm": 1.6127818822860718, | |
| "learning_rate": 5.5262532254791115e-06, | |
| "loss": 0.04709058627486229, | |
| "step": 154368 | |
| }, | |
| { | |
| "epoch": 4.483024557130846, | |
| "grad_norm": 1.5535714626312256, | |
| "learning_rate": 5.518830999391146e-06, | |
| "loss": 0.04727622866630554, | |
| "step": 154624 | |
| }, | |
| { | |
| "epoch": 4.490446783218811, | |
| "grad_norm": 1.0301692485809326, | |
| "learning_rate": 5.511408773303181e-06, | |
| "loss": 0.046475108712911606, | |
| "step": 154880 | |
| }, | |
| { | |
| "epoch": 4.4978690093067755, | |
| "grad_norm": 1.1295990943908691, | |
| "learning_rate": 5.503986547215216e-06, | |
| "loss": 0.046787649393081665, | |
| "step": 155136 | |
| }, | |
| { | |
| "epoch": 4.505291235394741, | |
| "grad_norm": 1.1239657402038574, | |
| "learning_rate": 5.496564321127251e-06, | |
| "loss": 0.04681747406721115, | |
| "step": 155392 | |
| }, | |
| { | |
| "epoch": 4.512713461482706, | |
| "grad_norm": 0.9050191044807434, | |
| "learning_rate": 5.489142095039286e-06, | |
| "loss": 0.047589726746082306, | |
| "step": 155648 | |
| }, | |
| { | |
| "epoch": 4.52013568757067, | |
| "grad_norm": 1.8931196928024292, | |
| "learning_rate": 5.4817198689513205e-06, | |
| "loss": 0.04757443442940712, | |
| "step": 155904 | |
| }, | |
| { | |
| "epoch": 4.527557913658636, | |
| "grad_norm": 2.9384677410125732, | |
| "learning_rate": 5.474297642863357e-06, | |
| "loss": 0.047171711921691895, | |
| "step": 156160 | |
| }, | |
| { | |
| "epoch": 4.534980139746601, | |
| "grad_norm": 1.7394589185714722, | |
| "learning_rate": 5.466875416775392e-06, | |
| "loss": 0.04649728164076805, | |
| "step": 156416 | |
| }, | |
| { | |
| "epoch": 4.542402365834565, | |
| "grad_norm": 1.6961169242858887, | |
| "learning_rate": 5.459453190687426e-06, | |
| "loss": 0.047085653990507126, | |
| "step": 156672 | |
| }, | |
| { | |
| "epoch": 4.549824591922531, | |
| "grad_norm": 1.1568101644515991, | |
| "learning_rate": 5.452030964599461e-06, | |
| "loss": 0.04730558767914772, | |
| "step": 156928 | |
| }, | |
| { | |
| "epoch": 4.557246818010496, | |
| "grad_norm": 1.0575653314590454, | |
| "learning_rate": 5.444608738511497e-06, | |
| "loss": 0.047147639095783234, | |
| "step": 157184 | |
| }, | |
| { | |
| "epoch": 4.56466904409846, | |
| "grad_norm": 2.008256196975708, | |
| "learning_rate": 5.437215505494188e-06, | |
| "loss": 0.04751652851700783, | |
| "step": 157440 | |
| }, | |
| { | |
| "epoch": 4.572091270186426, | |
| "grad_norm": 2.341398000717163, | |
| "learning_rate": 5.4297932794062226e-06, | |
| "loss": 0.04728484898805618, | |
| "step": 157696 | |
| }, | |
| { | |
| "epoch": 4.579513496274391, | |
| "grad_norm": 1.0035334825515747, | |
| "learning_rate": 5.422371053318257e-06, | |
| "loss": 0.04672228917479515, | |
| "step": 157952 | |
| }, | |
| { | |
| "epoch": 4.586935722362355, | |
| "grad_norm": 1.460576057434082, | |
| "learning_rate": 5.414948827230292e-06, | |
| "loss": 0.04826118424534798, | |
| "step": 158208 | |
| }, | |
| { | |
| "epoch": 4.5943579484503205, | |
| "grad_norm": 0.8920812010765076, | |
| "learning_rate": 5.4075266011423275e-06, | |
| "loss": 0.0474756695330143, | |
| "step": 158464 | |
| }, | |
| { | |
| "epoch": 4.601780174538286, | |
| "grad_norm": 1.5901042222976685, | |
| "learning_rate": 5.400104375054362e-06, | |
| "loss": 0.04701026901602745, | |
| "step": 158720 | |
| }, | |
| { | |
| "epoch": 4.60920240062625, | |
| "grad_norm": 0.6921675205230713, | |
| "learning_rate": 5.392682148966397e-06, | |
| "loss": 0.047285985201597214, | |
| "step": 158976 | |
| }, | |
| { | |
| "epoch": 4.6166246267142155, | |
| "grad_norm": 1.0079972743988037, | |
| "learning_rate": 5.385259922878433e-06, | |
| "loss": 0.0462353453040123, | |
| "step": 159232 | |
| }, | |
| { | |
| "epoch": 4.624046852802181, | |
| "grad_norm": 3.8479926586151123, | |
| "learning_rate": 5.377837696790468e-06, | |
| "loss": 0.04624490439891815, | |
| "step": 159488 | |
| }, | |
| { | |
| "epoch": 4.631469078890145, | |
| "grad_norm": 0.9857171773910522, | |
| "learning_rate": 5.370415470702503e-06, | |
| "loss": 0.046180061995983124, | |
| "step": 159744 | |
| }, | |
| { | |
| "epoch": 4.63889130497811, | |
| "grad_norm": 1.1568992137908936, | |
| "learning_rate": 5.363022237685194e-06, | |
| "loss": 0.046975065022706985, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 4.646313531066076, | |
| "grad_norm": 1.3709701299667358, | |
| "learning_rate": 5.355600011597229e-06, | |
| "loss": 0.0451301634311676, | |
| "step": 160256 | |
| }, | |
| { | |
| "epoch": 4.65373575715404, | |
| "grad_norm": 1.5765212774276733, | |
| "learning_rate": 5.348177785509264e-06, | |
| "loss": 0.046722594648599625, | |
| "step": 160512 | |
| }, | |
| { | |
| "epoch": 4.661157983242005, | |
| "grad_norm": 0.942916750907898, | |
| "learning_rate": 5.340755559421299e-06, | |
| "loss": 0.04559171572327614, | |
| "step": 160768 | |
| }, | |
| { | |
| "epoch": 4.66858020932997, | |
| "grad_norm": 0.8100948929786682, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.045222412794828415, | |
| "step": 161024 | |
| }, | |
| { | |
| "epoch": 4.676002435417935, | |
| "grad_norm": 0.9296344518661499, | |
| "learning_rate": 5.325911107245368e-06, | |
| "loss": 0.04632991924881935, | |
| "step": 161280 | |
| }, | |
| { | |
| "epoch": 4.6834246615059, | |
| "grad_norm": 1.6158431768417358, | |
| "learning_rate": 5.318488881157404e-06, | |
| "loss": 0.045601487159729004, | |
| "step": 161536 | |
| }, | |
| { | |
| "epoch": 4.690846887593865, | |
| "grad_norm": 1.0350348949432373, | |
| "learning_rate": 5.311066655069439e-06, | |
| "loss": 0.04574718698859215, | |
| "step": 161792 | |
| }, | |
| { | |
| "epoch": 4.69826911368183, | |
| "grad_norm": 1.5401016473770142, | |
| "learning_rate": 5.303644428981473e-06, | |
| "loss": 0.04511517658829689, | |
| "step": 162048 | |
| }, | |
| { | |
| "epoch": 4.705691339769795, | |
| "grad_norm": 2.0879998207092285, | |
| "learning_rate": 5.296251195964165e-06, | |
| "loss": 0.04421548545360565, | |
| "step": 162304 | |
| }, | |
| { | |
| "epoch": 4.71311356585776, | |
| "grad_norm": 2.1445138454437256, | |
| "learning_rate": 5.288828969876199e-06, | |
| "loss": 0.04451997950673103, | |
| "step": 162560 | |
| }, | |
| { | |
| "epoch": 4.720535791945725, | |
| "grad_norm": 1.3072458505630493, | |
| "learning_rate": 5.281406743788236e-06, | |
| "loss": 0.04533161222934723, | |
| "step": 162816 | |
| }, | |
| { | |
| "epoch": 4.72795801803369, | |
| "grad_norm": 1.2464230060577393, | |
| "learning_rate": 5.27398451770027e-06, | |
| "loss": 0.045559678226709366, | |
| "step": 163072 | |
| }, | |
| { | |
| "epoch": 4.735380244121655, | |
| "grad_norm": 0.8314167857170105, | |
| "learning_rate": 5.266562291612305e-06, | |
| "loss": 0.04433640465140343, | |
| "step": 163328 | |
| }, | |
| { | |
| "epoch": 4.74280247020962, | |
| "grad_norm": 1.6224439144134521, | |
| "learning_rate": 5.25914006552434e-06, | |
| "loss": 0.04455113410949707, | |
| "step": 163584 | |
| }, | |
| { | |
| "epoch": 4.750224696297585, | |
| "grad_norm": 1.887376070022583, | |
| "learning_rate": 5.251717839436375e-06, | |
| "loss": 0.04445907846093178, | |
| "step": 163840 | |
| }, | |
| { | |
| "epoch": 4.7576469223855495, | |
| "grad_norm": 1.4797005653381348, | |
| "learning_rate": 5.24429561334841e-06, | |
| "loss": 0.04413611441850662, | |
| "step": 164096 | |
| }, | |
| { | |
| "epoch": 4.765069148473515, | |
| "grad_norm": Infinity, | |
| "learning_rate": 5.236873387260445e-06, | |
| "loss": 0.04555521532893181, | |
| "step": 164352 | |
| }, | |
| { | |
| "epoch": 4.77249137456148, | |
| "grad_norm": 1.3807029724121094, | |
| "learning_rate": 5.229480154243136e-06, | |
| "loss": 0.04356437548995018, | |
| "step": 164608 | |
| }, | |
| { | |
| "epoch": 4.7799136006494445, | |
| "grad_norm": 1.562363624572754, | |
| "learning_rate": 5.222057928155171e-06, | |
| "loss": 0.04474165290594101, | |
| "step": 164864 | |
| }, | |
| { | |
| "epoch": 4.78733582673741, | |
| "grad_norm": 1.1772706508636475, | |
| "learning_rate": 5.214635702067207e-06, | |
| "loss": 0.04473555088043213, | |
| "step": 165120 | |
| }, | |
| { | |
| "epoch": 4.794758052825375, | |
| "grad_norm": 1.3282051086425781, | |
| "learning_rate": 5.207213475979242e-06, | |
| "loss": 0.044975169003009796, | |
| "step": 165376 | |
| }, | |
| { | |
| "epoch": 4.802180278913339, | |
| "grad_norm": 1.65360689163208, | |
| "learning_rate": 5.1997912498912765e-06, | |
| "loss": 0.04405369237065315, | |
| "step": 165632 | |
| }, | |
| { | |
| "epoch": 4.809602505001305, | |
| "grad_norm": 2.0444693565368652, | |
| "learning_rate": 5.192369023803312e-06, | |
| "loss": 0.04358972609043121, | |
| "step": 165888 | |
| }, | |
| { | |
| "epoch": 4.81702473108927, | |
| "grad_norm": 1.6881554126739502, | |
| "learning_rate": 5.184946797715347e-06, | |
| "loss": 0.043333280831575394, | |
| "step": 166144 | |
| }, | |
| { | |
| "epoch": 4.824446957177234, | |
| "grad_norm": 0.8001645803451538, | |
| "learning_rate": 5.1775245716273815e-06, | |
| "loss": 0.044071778655052185, | |
| "step": 166400 | |
| }, | |
| { | |
| "epoch": 4.8318691832652, | |
| "grad_norm": 1.155084252357483, | |
| "learning_rate": 5.170131338610073e-06, | |
| "loss": 0.04443291947245598, | |
| "step": 166656 | |
| }, | |
| { | |
| "epoch": 4.839291409353165, | |
| "grad_norm": 1.50560462474823, | |
| "learning_rate": 5.1627091125221075e-06, | |
| "loss": 0.04399026930332184, | |
| "step": 166912 | |
| }, | |
| { | |
| "epoch": 4.846713635441129, | |
| "grad_norm": 1.0352333784103394, | |
| "learning_rate": 5.155286886434143e-06, | |
| "loss": 0.043082624673843384, | |
| "step": 167168 | |
| }, | |
| { | |
| "epoch": 4.854135861529095, | |
| "grad_norm": 1.7933018207550049, | |
| "learning_rate": 5.147864660346178e-06, | |
| "loss": 0.04398785158991814, | |
| "step": 167424 | |
| }, | |
| { | |
| "epoch": 4.86155808761706, | |
| "grad_norm": 1.0038179159164429, | |
| "learning_rate": 5.140442434258212e-06, | |
| "loss": 0.043893154710531235, | |
| "step": 167680 | |
| }, | |
| { | |
| "epoch": 4.868980313705024, | |
| "grad_norm": 2.1641430854797363, | |
| "learning_rate": 5.133020208170247e-06, | |
| "loss": 0.04407702013850212, | |
| "step": 167936 | |
| }, | |
| { | |
| "epoch": 4.8764025397929895, | |
| "grad_norm": 1.0531773567199707, | |
| "learning_rate": 5.1255979820822835e-06, | |
| "loss": 0.042828816920518875, | |
| "step": 168192 | |
| }, | |
| { | |
| "epoch": 4.883824765880955, | |
| "grad_norm": 0.7979609966278076, | |
| "learning_rate": 5.118175755994318e-06, | |
| "loss": 0.04335318133234978, | |
| "step": 168448 | |
| }, | |
| { | |
| "epoch": 4.891246991968919, | |
| "grad_norm": 1.9395774602890015, | |
| "learning_rate": 5.110753529906353e-06, | |
| "loss": 0.04351218044757843, | |
| "step": 168704 | |
| }, | |
| { | |
| "epoch": 4.8986692180568845, | |
| "grad_norm": 1.2753045558929443, | |
| "learning_rate": 5.103331303818388e-06, | |
| "loss": 0.04289284721016884, | |
| "step": 168960 | |
| }, | |
| { | |
| "epoch": 4.90609144414485, | |
| "grad_norm": 0.857566237449646, | |
| "learning_rate": 5.095909077730423e-06, | |
| "loss": 0.04341334477066994, | |
| "step": 169216 | |
| }, | |
| { | |
| "epoch": 4.913513670232814, | |
| "grad_norm": 0.9497765302658081, | |
| "learning_rate": 5.0885158447131144e-06, | |
| "loss": 0.043380167335271835, | |
| "step": 169472 | |
| }, | |
| { | |
| "epoch": 4.920935896320779, | |
| "grad_norm": 1.0917929410934448, | |
| "learning_rate": 5.081093618625149e-06, | |
| "loss": 0.042631786316633224, | |
| "step": 169728 | |
| }, | |
| { | |
| "epoch": 4.928358122408744, | |
| "grad_norm": 1.9655131101608276, | |
| "learning_rate": 5.073671392537184e-06, | |
| "loss": 0.04326947405934334, | |
| "step": 169984 | |
| }, | |
| { | |
| "epoch": 4.935780348496709, | |
| "grad_norm": 1.1260302066802979, | |
| "learning_rate": 5.0662491664492186e-06, | |
| "loss": 0.04309643805027008, | |
| "step": 170240 | |
| }, | |
| { | |
| "epoch": 4.943202574584674, | |
| "grad_norm": 1.1136366128921509, | |
| "learning_rate": 5.058826940361254e-06, | |
| "loss": 0.042600467801094055, | |
| "step": 170496 | |
| }, | |
| { | |
| "epoch": 4.950624800672639, | |
| "grad_norm": 1.2051705121994019, | |
| "learning_rate": 5.051404714273289e-06, | |
| "loss": 0.04228004813194275, | |
| "step": 170752 | |
| }, | |
| { | |
| "epoch": 4.958047026760604, | |
| "grad_norm": 1.7227786779403687, | |
| "learning_rate": 5.0439824881853235e-06, | |
| "loss": 0.042368240654468536, | |
| "step": 171008 | |
| }, | |
| { | |
| "epoch": 4.965469252848569, | |
| "grad_norm": 2.274092674255371, | |
| "learning_rate": 5.036560262097358e-06, | |
| "loss": 0.043530821800231934, | |
| "step": 171264 | |
| }, | |
| { | |
| "epoch": 4.972891478936534, | |
| "grad_norm": 1.0979293584823608, | |
| "learning_rate": 5.0291670290800495e-06, | |
| "loss": 0.042770832777023315, | |
| "step": 171520 | |
| }, | |
| { | |
| "epoch": 4.980313705024499, | |
| "grad_norm": 1.0030477046966553, | |
| "learning_rate": 5.021744802992086e-06, | |
| "loss": 0.04295011982321739, | |
| "step": 171776 | |
| }, | |
| { | |
| "epoch": 4.987735931112464, | |
| "grad_norm": 1.213944911956787, | |
| "learning_rate": 5.014322576904121e-06, | |
| "loss": 0.04209183156490326, | |
| "step": 172032 | |
| }, | |
| { | |
| "epoch": 4.995158157200429, | |
| "grad_norm": 1.1684976816177368, | |
| "learning_rate": 5.006900350816155e-06, | |
| "loss": 0.04208473488688469, | |
| "step": 172288 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_f1_B-DISEASE": 0.8731543713839041, | |
| "eval_f1_B-PROCEDURE": 0.8580242773791161, | |
| "eval_f1_B-SYMPTOM": 0.858197558805111, | |
| "eval_f1_I-DISEASE": 0.9029563216711708, | |
| "eval_f1_I-PROCEDURE": 0.8963639576160926, | |
| "eval_f1_I-SYMPTOM": 0.8813308521848306, | |
| "eval_f1_O": 0.947994131071765, | |
| "eval_f1_macro": 0.8882887814445699, | |
| "eval_f1_micro": 0.9219783367609227, | |
| "eval_loss": 0.0677226185798645, | |
| "eval_precision_B-DISEASE": 0.8782828418590898, | |
| "eval_precision_B-PROCEDURE": 0.8684967959161508, | |
| "eval_precision_B-SYMPTOM": 0.8667968663308108, | |
| "eval_precision_I-DISEASE": 0.8955728011156288, | |
| "eval_precision_I-PROCEDURE": 0.8945200209453178, | |
| "eval_precision_I-SYMPTOM": 0.8788809229255787, | |
| "eval_precision_O": 0.9545386928634556, | |
| "eval_precision_macro": 0.8910127059937188, | |
| "eval_precision_micro": 0.9246487510080916, | |
| "eval_rauc_macro": 0.9341987747075685, | |
| "eval_rauc_micro": 0.9531943224029301, | |
| "eval_recall_B-DISEASE": 0.868085445560338, | |
| "eval_recall_B-PROCEDURE": 0.847801309407056, | |
| "eval_recall_B-SYMPTOM": 0.8497671989904704, | |
| "eval_recall_I-DISEASE": 0.9104626006904488, | |
| "eval_recall_I-PROCEDURE": 0.8982155120611212, | |
| "eval_recall_I-SYMPTOM": 0.8837944782517462, | |
| "eval_recall_O": 0.9415387005573862, | |
| "eval_recall_macro": 0.8856664636455094, | |
| "eval_recall_micro": 0.9193233025738639, | |
| "eval_roc_auc_B-DISEASE": 0.9327433663174488, | |
| "eval_roc_auc_B-PROCEDURE": 0.922510834740169, | |
| "eval_roc_auc_B-SYMPTOM": 0.9231529958369201, | |
| "eval_roc_auc_I-DISEASE": 0.9478521851863642, | |
| "eval_roc_auc_I-PROCEDURE": 0.9429141319400265, | |
| "eval_roc_auc_I-SYMPTOM": 0.9311568515810746, | |
| "eval_roc_auc_O": 0.9390610573509757, | |
| "eval_runtime": 50.9164, | |
| "eval_samples_per_second": 277.906, | |
| "eval_steps_per_second": 34.743, | |
| "step": 172455 | |
| }, | |
| { | |
| "epoch": 5.002580383288394, | |
| "grad_norm": 0.7581572532653809, | |
| "learning_rate": 4.99947812472819e-06, | |
| "loss": 0.0411066971719265, | |
| "step": 172544 | |
| }, | |
| { | |
| "epoch": 5.010002609376359, | |
| "grad_norm": 0.6313815712928772, | |
| "learning_rate": 4.9920558986402255e-06, | |
| "loss": 0.03819066286087036, | |
| "step": 172800 | |
| }, | |
| { | |
| "epoch": 5.017424835464324, | |
| "grad_norm": 2.0061378479003906, | |
| "learning_rate": 4.98463367255226e-06, | |
| "loss": 0.03691718354821205, | |
| "step": 173056 | |
| }, | |
| { | |
| "epoch": 5.024847061552289, | |
| "grad_norm": 1.3767658472061157, | |
| "learning_rate": 4.977211446464296e-06, | |
| "loss": 0.03767155110836029, | |
| "step": 173312 | |
| }, | |
| { | |
| "epoch": 5.032269287640254, | |
| "grad_norm": 0.6843507885932922, | |
| "learning_rate": 4.9697892203763305e-06, | |
| "loss": 0.03721943870186806, | |
| "step": 173568 | |
| }, | |
| { | |
| "epoch": 5.0396915137282186, | |
| "grad_norm": 0.8531635999679565, | |
| "learning_rate": 4.962366994288366e-06, | |
| "loss": 0.03769301623106003, | |
| "step": 173824 | |
| }, | |
| { | |
| "epoch": 5.047113739816184, | |
| "grad_norm": 0.7804009914398193, | |
| "learning_rate": 4.954944768200401e-06, | |
| "loss": 0.03719908744096756, | |
| "step": 174080 | |
| }, | |
| { | |
| "epoch": 5.054535965904149, | |
| "grad_norm": 1.0283087491989136, | |
| "learning_rate": 4.9475225421124355e-06, | |
| "loss": 0.03786902874708176, | |
| "step": 174336 | |
| }, | |
| { | |
| "epoch": 5.0619581919921135, | |
| "grad_norm": 1.5266740322113037, | |
| "learning_rate": 4.94010031602447e-06, | |
| "loss": 0.03749080002307892, | |
| "step": 174592 | |
| }, | |
| { | |
| "epoch": 5.069380418080079, | |
| "grad_norm": 2.2684950828552246, | |
| "learning_rate": 4.932678089936506e-06, | |
| "loss": 0.03684950992465019, | |
| "step": 174848 | |
| }, | |
| { | |
| "epoch": 5.076802644168044, | |
| "grad_norm": 1.4241979122161865, | |
| "learning_rate": 4.92525586384854e-06, | |
| "loss": 0.037433601915836334, | |
| "step": 175104 | |
| }, | |
| { | |
| "epoch": 5.084224870256008, | |
| "grad_norm": 1.1710699796676636, | |
| "learning_rate": 4.917833637760576e-06, | |
| "loss": 0.03730001673102379, | |
| "step": 175360 | |
| }, | |
| { | |
| "epoch": 5.091647096343974, | |
| "grad_norm": 0.8361418843269348, | |
| "learning_rate": 4.910469397813923e-06, | |
| "loss": 0.03734878450632095, | |
| "step": 175616 | |
| }, | |
| { | |
| "epoch": 5.099069322431939, | |
| "grad_norm": 0.9384424686431885, | |
| "learning_rate": 4.9030471717259585e-06, | |
| "loss": 0.037683479487895966, | |
| "step": 175872 | |
| }, | |
| { | |
| "epoch": 5.106491548519903, | |
| "grad_norm": 1.1333366632461548, | |
| "learning_rate": 4.895624945637993e-06, | |
| "loss": 0.0370732806622982, | |
| "step": 176128 | |
| }, | |
| { | |
| "epoch": 5.113913774607869, | |
| "grad_norm": 1.8009032011032104, | |
| "learning_rate": 4.888202719550028e-06, | |
| "loss": 0.03719591349363327, | |
| "step": 176384 | |
| }, | |
| { | |
| "epoch": 5.121336000695834, | |
| "grad_norm": 2.4545230865478516, | |
| "learning_rate": 4.880780493462063e-06, | |
| "loss": 0.036990392953157425, | |
| "step": 176640 | |
| }, | |
| { | |
| "epoch": 5.128758226783798, | |
| "grad_norm": 2.2979793548583984, | |
| "learning_rate": 4.873358267374098e-06, | |
| "loss": 0.037442296743392944, | |
| "step": 176896 | |
| }, | |
| { | |
| "epoch": 5.136180452871764, | |
| "grad_norm": 1.7477954626083374, | |
| "learning_rate": 4.865936041286133e-06, | |
| "loss": 0.03654196485877037, | |
| "step": 177152 | |
| }, | |
| { | |
| "epoch": 5.143602678959729, | |
| "grad_norm": 1.8604315519332886, | |
| "learning_rate": 4.858513815198168e-06, | |
| "loss": 0.03761319816112518, | |
| "step": 177408 | |
| }, | |
| { | |
| "epoch": 5.151024905047693, | |
| "grad_norm": 1.1665540933609009, | |
| "learning_rate": 4.851091589110203e-06, | |
| "loss": 0.037620872259140015, | |
| "step": 177664 | |
| }, | |
| { | |
| "epoch": 5.158447131135659, | |
| "grad_norm": 1.84141206741333, | |
| "learning_rate": 4.843669363022238e-06, | |
| "loss": 0.0379265621304512, | |
| "step": 177920 | |
| }, | |
| { | |
| "epoch": 5.165869357223624, | |
| "grad_norm": 1.058616280555725, | |
| "learning_rate": 4.8362471369342725e-06, | |
| "loss": 0.037140730768442154, | |
| "step": 178176 | |
| }, | |
| { | |
| "epoch": 5.173291583311588, | |
| "grad_norm": 4.370513916015625, | |
| "learning_rate": 4.828824910846308e-06, | |
| "loss": 0.038546182215213776, | |
| "step": 178432 | |
| }, | |
| { | |
| "epoch": 5.1807138093995535, | |
| "grad_norm": 0.8379314541816711, | |
| "learning_rate": 4.821402684758344e-06, | |
| "loss": 0.03713466227054596, | |
| "step": 178688 | |
| }, | |
| { | |
| "epoch": 5.188136035487519, | |
| "grad_norm": 1.1568081378936768, | |
| "learning_rate": 4.813980458670378e-06, | |
| "loss": 0.03810407966375351, | |
| "step": 178944 | |
| }, | |
| { | |
| "epoch": 5.195558261575483, | |
| "grad_norm": 0.6176631450653076, | |
| "learning_rate": 4.80658722565307e-06, | |
| "loss": 0.03869181126356125, | |
| "step": 179200 | |
| }, | |
| { | |
| "epoch": 5.2029804876634485, | |
| "grad_norm": 1.0596495866775513, | |
| "learning_rate": 4.799164999565104e-06, | |
| "loss": 0.03698258474469185, | |
| "step": 179456 | |
| }, | |
| { | |
| "epoch": 5.210402713751414, | |
| "grad_norm": 1.302703619003296, | |
| "learning_rate": 4.791742773477139e-06, | |
| "loss": 0.0362229198217392, | |
| "step": 179712 | |
| }, | |
| { | |
| "epoch": 5.217824939839378, | |
| "grad_norm": 1.2289581298828125, | |
| "learning_rate": 4.7843205473891746e-06, | |
| "loss": 0.036259524524211884, | |
| "step": 179968 | |
| }, | |
| { | |
| "epoch": 5.225247165927343, | |
| "grad_norm": 1.0403485298156738, | |
| "learning_rate": 4.776898321301209e-06, | |
| "loss": 0.03635624051094055, | |
| "step": 180224 | |
| }, | |
| { | |
| "epoch": 5.232669392015309, | |
| "grad_norm": 1.6828593015670776, | |
| "learning_rate": 4.769476095213245e-06, | |
| "loss": 0.03610437735915184, | |
| "step": 180480 | |
| }, | |
| { | |
| "epoch": 5.240091618103273, | |
| "grad_norm": 1.2846966981887817, | |
| "learning_rate": 4.7620538691252795e-06, | |
| "loss": 0.036061983555555344, | |
| "step": 180736 | |
| }, | |
| { | |
| "epoch": 5.247513844191238, | |
| "grad_norm": 1.221166968345642, | |
| "learning_rate": 4.754631643037314e-06, | |
| "loss": 0.037047479301691055, | |
| "step": 180992 | |
| }, | |
| { | |
| "epoch": 5.254936070279204, | |
| "grad_norm": 0.8924992680549622, | |
| "learning_rate": 4.747209416949349e-06, | |
| "loss": 0.03579121455550194, | |
| "step": 181248 | |
| }, | |
| { | |
| "epoch": 5.262358296367168, | |
| "grad_norm": 1.257304310798645, | |
| "learning_rate": 4.73981618393204e-06, | |
| "loss": 0.03585585206747055, | |
| "step": 181504 | |
| }, | |
| { | |
| "epoch": 5.269780522455133, | |
| "grad_norm": 1.389647364616394, | |
| "learning_rate": 4.732393957844076e-06, | |
| "loss": 0.0362885482609272, | |
| "step": 181760 | |
| }, | |
| { | |
| "epoch": 5.277202748543099, | |
| "grad_norm": 0.7663612961769104, | |
| "learning_rate": 4.7249717317561104e-06, | |
| "loss": 0.036874283105134964, | |
| "step": 182016 | |
| }, | |
| { | |
| "epoch": 5.284624974631063, | |
| "grad_norm": 0.7243006825447083, | |
| "learning_rate": 4.717549505668146e-06, | |
| "loss": 0.03536940738558769, | |
| "step": 182272 | |
| }, | |
| { | |
| "epoch": 5.292047200719028, | |
| "grad_norm": 1.031685709953308, | |
| "learning_rate": 4.710127279580181e-06, | |
| "loss": 0.03676433116197586, | |
| "step": 182528 | |
| }, | |
| { | |
| "epoch": 5.2994694268069935, | |
| "grad_norm": 1.2616617679595947, | |
| "learning_rate": 4.702705053492216e-06, | |
| "loss": 0.03628234937787056, | |
| "step": 182784 | |
| }, | |
| { | |
| "epoch": 5.306891652894958, | |
| "grad_norm": 0.8668111562728882, | |
| "learning_rate": 4.695282827404251e-06, | |
| "loss": 0.03764399513602257, | |
| "step": 183040 | |
| }, | |
| { | |
| "epoch": 5.314313878982923, | |
| "grad_norm": 1.1500765085220337, | |
| "learning_rate": 4.687860601316286e-06, | |
| "loss": 0.03546790033578873, | |
| "step": 183296 | |
| }, | |
| { | |
| "epoch": 5.3217361050708885, | |
| "grad_norm": 1.7504640817642212, | |
| "learning_rate": 4.68043837522832e-06, | |
| "loss": 0.036827899515628815, | |
| "step": 183552 | |
| }, | |
| { | |
| "epoch": 5.329158331158853, | |
| "grad_norm": 1.4775344133377075, | |
| "learning_rate": 4.673016149140356e-06, | |
| "loss": 0.03753214702010155, | |
| "step": 183808 | |
| }, | |
| { | |
| "epoch": 5.336580557246818, | |
| "grad_norm": 1.0644928216934204, | |
| "learning_rate": 4.6655939230523915e-06, | |
| "loss": 0.035319630056619644, | |
| "step": 184064 | |
| }, | |
| { | |
| "epoch": 5.344002783334783, | |
| "grad_norm": 2.237816333770752, | |
| "learning_rate": 4.658171696964426e-06, | |
| "loss": 0.03639177232980728, | |
| "step": 184320 | |
| }, | |
| { | |
| "epoch": 5.351425009422748, | |
| "grad_norm": 1.5501213073730469, | |
| "learning_rate": 4.650749470876461e-06, | |
| "loss": 0.03628689795732498, | |
| "step": 184576 | |
| }, | |
| { | |
| "epoch": 5.358847235510713, | |
| "grad_norm": 0.8848547339439392, | |
| "learning_rate": 4.6433272447884956e-06, | |
| "loss": 0.03511285409331322, | |
| "step": 184832 | |
| }, | |
| { | |
| "epoch": 5.366269461598678, | |
| "grad_norm": 1.4660401344299316, | |
| "learning_rate": 4.635905018700531e-06, | |
| "loss": 0.0370330885052681, | |
| "step": 185088 | |
| }, | |
| { | |
| "epoch": 5.373691687686643, | |
| "grad_norm": 0.6388676762580872, | |
| "learning_rate": 4.628482792612566e-06, | |
| "loss": 0.03555865213274956, | |
| "step": 185344 | |
| }, | |
| { | |
| "epoch": 5.381113913774608, | |
| "grad_norm": 0.9646828770637512, | |
| "learning_rate": 4.621089559595257e-06, | |
| "loss": 0.0356537364423275, | |
| "step": 185600 | |
| }, | |
| { | |
| "epoch": 5.388536139862573, | |
| "grad_norm": 1.1582335233688354, | |
| "learning_rate": 4.613696326577948e-06, | |
| "loss": 0.03515341505408287, | |
| "step": 185856 | |
| }, | |
| { | |
| "epoch": 5.395958365950538, | |
| "grad_norm": 0.8179342150688171, | |
| "learning_rate": 4.606274100489983e-06, | |
| "loss": 0.03622395917773247, | |
| "step": 186112 | |
| }, | |
| { | |
| "epoch": 5.403380592038503, | |
| "grad_norm": 1.2221698760986328, | |
| "learning_rate": 4.598851874402019e-06, | |
| "loss": 0.03511522337794304, | |
| "step": 186368 | |
| }, | |
| { | |
| "epoch": 5.410802818126468, | |
| "grad_norm": 1.491208553314209, | |
| "learning_rate": 4.591429648314053e-06, | |
| "loss": 0.035874124616384506, | |
| "step": 186624 | |
| }, | |
| { | |
| "epoch": 5.418225044214433, | |
| "grad_norm": 1.4371721744537354, | |
| "learning_rate": 4.584007422226088e-06, | |
| "loss": 0.035513415932655334, | |
| "step": 186880 | |
| }, | |
| { | |
| "epoch": 5.425647270302398, | |
| "grad_norm": 0.6826869249343872, | |
| "learning_rate": 4.5765851961381236e-06, | |
| "loss": 0.034688860177993774, | |
| "step": 187136 | |
| }, | |
| { | |
| "epoch": 5.433069496390362, | |
| "grad_norm": 1.0652885437011719, | |
| "learning_rate": 4.569162970050158e-06, | |
| "loss": 0.035885218530893326, | |
| "step": 187392 | |
| }, | |
| { | |
| "epoch": 5.440491722478328, | |
| "grad_norm": 1.6044352054595947, | |
| "learning_rate": 4.561740743962194e-06, | |
| "loss": 0.03499657288193703, | |
| "step": 187648 | |
| }, | |
| { | |
| "epoch": 5.447913948566293, | |
| "grad_norm": 0.9734591245651245, | |
| "learning_rate": 4.5543185178742285e-06, | |
| "loss": 0.03455566614866257, | |
| "step": 187904 | |
| }, | |
| { | |
| "epoch": 5.455336174654257, | |
| "grad_norm": 2.1935832500457764, | |
| "learning_rate": 4.546896291786263e-06, | |
| "loss": 0.0354565866291523, | |
| "step": 188160 | |
| }, | |
| { | |
| "epoch": 5.4627584007422225, | |
| "grad_norm": 1.7379024028778076, | |
| "learning_rate": 4.539474065698298e-06, | |
| "loss": 0.03592940419912338, | |
| "step": 188416 | |
| }, | |
| { | |
| "epoch": 5.470180626830188, | |
| "grad_norm": 1.095860242843628, | |
| "learning_rate": 4.5320518396103335e-06, | |
| "loss": 0.03530087321996689, | |
| "step": 188672 | |
| }, | |
| { | |
| "epoch": 5.477602852918152, | |
| "grad_norm": 0.9119752645492554, | |
| "learning_rate": 4.524629613522368e-06, | |
| "loss": 0.035207945853471756, | |
| "step": 188928 | |
| }, | |
| { | |
| "epoch": 5.4850250790061175, | |
| "grad_norm": 1.4463428258895874, | |
| "learning_rate": 4.5172363805050595e-06, | |
| "loss": 0.03535320237278938, | |
| "step": 189184 | |
| }, | |
| { | |
| "epoch": 5.492447305094083, | |
| "grad_norm": 1.434726357460022, | |
| "learning_rate": 4.509814154417095e-06, | |
| "loss": 0.033964719623327255, | |
| "step": 189440 | |
| }, | |
| { | |
| "epoch": 5.499869531182047, | |
| "grad_norm": 0.744862973690033, | |
| "learning_rate": 4.50239192832913e-06, | |
| "loss": 0.03516576066613197, | |
| "step": 189696 | |
| }, | |
| { | |
| "epoch": 5.507291757270012, | |
| "grad_norm": 1.066475510597229, | |
| "learning_rate": 4.494969702241164e-06, | |
| "loss": 0.03472698852419853, | |
| "step": 189952 | |
| }, | |
| { | |
| "epoch": 5.514713983357978, | |
| "grad_norm": 0.9533625245094299, | |
| "learning_rate": 4.487547476153199e-06, | |
| "loss": 0.035540252923965454, | |
| "step": 190208 | |
| }, | |
| { | |
| "epoch": 5.522136209445942, | |
| "grad_norm": 1.1177531480789185, | |
| "learning_rate": 4.480125250065235e-06, | |
| "loss": 0.03571905568242073, | |
| "step": 190464 | |
| }, | |
| { | |
| "epoch": 5.529558435533907, | |
| "grad_norm": 0.9426385164260864, | |
| "learning_rate": 4.47270302397727e-06, | |
| "loss": 0.03423643484711647, | |
| "step": 190720 | |
| }, | |
| { | |
| "epoch": 5.536980661621873, | |
| "grad_norm": 1.4343687295913696, | |
| "learning_rate": 4.465280797889305e-06, | |
| "loss": 0.034453149884939194, | |
| "step": 190976 | |
| }, | |
| { | |
| "epoch": 5.544402887709837, | |
| "grad_norm": 3.8593392372131348, | |
| "learning_rate": 4.4578585718013405e-06, | |
| "loss": 0.03434739634394646, | |
| "step": 191232 | |
| }, | |
| { | |
| "epoch": 5.551825113797802, | |
| "grad_norm": 1.1529148817062378, | |
| "learning_rate": 4.450436345713375e-06, | |
| "loss": 0.03335895761847496, | |
| "step": 191488 | |
| }, | |
| { | |
| "epoch": 5.559247339885768, | |
| "grad_norm": 1.0085877180099487, | |
| "learning_rate": 4.44301411962541e-06, | |
| "loss": 0.03488295525312424, | |
| "step": 191744 | |
| }, | |
| { | |
| "epoch": 5.566669565973732, | |
| "grad_norm": 0.9415827393531799, | |
| "learning_rate": 4.435591893537445e-06, | |
| "loss": 0.0333930142223835, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 5.574091792061697, | |
| "grad_norm": 1.1781333684921265, | |
| "learning_rate": 4.42816966744948e-06, | |
| "loss": 0.034888722002506256, | |
| "step": 192256 | |
| }, | |
| { | |
| "epoch": 5.5815140181496625, | |
| "grad_norm": 1.331027865409851, | |
| "learning_rate": 4.420747441361515e-06, | |
| "loss": 0.034972723573446274, | |
| "step": 192512 | |
| }, | |
| { | |
| "epoch": 5.588936244237627, | |
| "grad_norm": 1.1153818368911743, | |
| "learning_rate": 4.41332521527355e-06, | |
| "loss": 0.03404976800084114, | |
| "step": 192768 | |
| }, | |
| { | |
| "epoch": 5.596358470325592, | |
| "grad_norm": 0.7376364469528198, | |
| "learning_rate": 4.405902989185585e-06, | |
| "loss": 0.03406954184174538, | |
| "step": 193024 | |
| }, | |
| { | |
| "epoch": 5.6037806964135575, | |
| "grad_norm": 0.7925898432731628, | |
| "learning_rate": 4.398509756168276e-06, | |
| "loss": 0.03464552387595177, | |
| "step": 193280 | |
| }, | |
| { | |
| "epoch": 5.611202922501522, | |
| "grad_norm": 1.0866154432296753, | |
| "learning_rate": 4.391087530080311e-06, | |
| "loss": 0.03376219421625137, | |
| "step": 193536 | |
| }, | |
| { | |
| "epoch": 5.618625148589487, | |
| "grad_norm": 0.7970457077026367, | |
| "learning_rate": 4.383665303992346e-06, | |
| "loss": 0.03417901694774628, | |
| "step": 193792 | |
| }, | |
| { | |
| "epoch": 5.626047374677452, | |
| "grad_norm": 1.266381025314331, | |
| "learning_rate": 4.376272070975037e-06, | |
| "loss": 0.03481767326593399, | |
| "step": 194048 | |
| }, | |
| { | |
| "epoch": 5.633469600765417, | |
| "grad_norm": 0.7626895904541016, | |
| "learning_rate": 4.368849844887073e-06, | |
| "loss": 0.034649841487407684, | |
| "step": 194304 | |
| }, | |
| { | |
| "epoch": 5.640891826853382, | |
| "grad_norm": 0.9965762495994568, | |
| "learning_rate": 4.361427618799107e-06, | |
| "loss": 0.034473009407520294, | |
| "step": 194560 | |
| }, | |
| { | |
| "epoch": 5.648314052941347, | |
| "grad_norm": 0.8567109107971191, | |
| "learning_rate": 4.354005392711143e-06, | |
| "loss": 0.03406520560383797, | |
| "step": 194816 | |
| }, | |
| { | |
| "epoch": 5.655736279029312, | |
| "grad_norm": 1.3910399675369263, | |
| "learning_rate": 4.3465831666231775e-06, | |
| "loss": 0.03407694026827812, | |
| "step": 195072 | |
| }, | |
| { | |
| "epoch": 5.663158505117277, | |
| "grad_norm": 1.269702672958374, | |
| "learning_rate": 4.339160940535212e-06, | |
| "loss": 0.03484127297997475, | |
| "step": 195328 | |
| }, | |
| { | |
| "epoch": 5.670580731205242, | |
| "grad_norm": 3.456882953643799, | |
| "learning_rate": 4.331738714447247e-06, | |
| "loss": 0.033184461295604706, | |
| "step": 195584 | |
| }, | |
| { | |
| "epoch": 5.678002957293207, | |
| "grad_norm": 0.6877565979957581, | |
| "learning_rate": 4.3243164883592825e-06, | |
| "loss": 0.033996909856796265, | |
| "step": 195840 | |
| }, | |
| { | |
| "epoch": 5.685425183381172, | |
| "grad_norm": 2.7158353328704834, | |
| "learning_rate": 4.316894262271317e-06, | |
| "loss": 0.03413985297083855, | |
| "step": 196096 | |
| }, | |
| { | |
| "epoch": 5.692847409469137, | |
| "grad_norm": 1.5668667554855347, | |
| "learning_rate": 4.309472036183353e-06, | |
| "loss": 0.03448374941945076, | |
| "step": 196352 | |
| }, | |
| { | |
| "epoch": 5.700269635557102, | |
| "grad_norm": 1.197697639465332, | |
| "learning_rate": 4.3020498100953875e-06, | |
| "loss": 0.032984409481287, | |
| "step": 196608 | |
| }, | |
| { | |
| "epoch": 5.707691861645067, | |
| "grad_norm": 1.063459873199463, | |
| "learning_rate": 4.294656577078079e-06, | |
| "loss": 0.0347076877951622, | |
| "step": 196864 | |
| }, | |
| { | |
| "epoch": 5.715114087733031, | |
| "grad_norm": 1.3296229839324951, | |
| "learning_rate": 4.2872343509901134e-06, | |
| "loss": 0.033475086092948914, | |
| "step": 197120 | |
| }, | |
| { | |
| "epoch": 5.722536313820997, | |
| "grad_norm": 0.8644685745239258, | |
| "learning_rate": 4.279812124902148e-06, | |
| "loss": 0.033407215029001236, | |
| "step": 197376 | |
| }, | |
| { | |
| "epoch": 5.729958539908962, | |
| "grad_norm": 0.6593635082244873, | |
| "learning_rate": 4.272389898814184e-06, | |
| "loss": 0.03437068685889244, | |
| "step": 197632 | |
| }, | |
| { | |
| "epoch": 5.737380765996926, | |
| "grad_norm": 3.1430885791778564, | |
| "learning_rate": 4.264967672726219e-06, | |
| "loss": 0.03338238596916199, | |
| "step": 197888 | |
| }, | |
| { | |
| "epoch": 5.7448029920848915, | |
| "grad_norm": 1.9421474933624268, | |
| "learning_rate": 4.257545446638254e-06, | |
| "loss": 0.03321695327758789, | |
| "step": 198144 | |
| }, | |
| { | |
| "epoch": 5.752225218172857, | |
| "grad_norm": 0.7477284073829651, | |
| "learning_rate": 4.250123220550289e-06, | |
| "loss": 0.03227618336677551, | |
| "step": 198400 | |
| }, | |
| { | |
| "epoch": 5.759647444260821, | |
| "grad_norm": 1.8507879972457886, | |
| "learning_rate": 4.242700994462323e-06, | |
| "loss": 0.033165134489536285, | |
| "step": 198656 | |
| }, | |
| { | |
| "epoch": 5.7670696703487865, | |
| "grad_norm": 1.9801945686340332, | |
| "learning_rate": 4.235278768374359e-06, | |
| "loss": 0.032966770231723785, | |
| "step": 198912 | |
| }, | |
| { | |
| "epoch": 5.774491896436752, | |
| "grad_norm": 1.2919081449508667, | |
| "learning_rate": 4.227856542286394e-06, | |
| "loss": 0.0328759141266346, | |
| "step": 199168 | |
| }, | |
| { | |
| "epoch": 5.781914122524716, | |
| "grad_norm": 0.830310046672821, | |
| "learning_rate": 4.220434316198429e-06, | |
| "loss": 0.03241224214434624, | |
| "step": 199424 | |
| }, | |
| { | |
| "epoch": 5.789336348612681, | |
| "grad_norm": 0.971379280090332, | |
| "learning_rate": 4.213012090110464e-06, | |
| "loss": 0.03393639251589775, | |
| "step": 199680 | |
| }, | |
| { | |
| "epoch": 5.796758574700647, | |
| "grad_norm": 0.6849918961524963, | |
| "learning_rate": 4.205589864022499e-06, | |
| "loss": 0.032369766384363174, | |
| "step": 199936 | |
| }, | |
| { | |
| "epoch": 5.804180800788611, | |
| "grad_norm": 0.9221972823143005, | |
| "learning_rate": 4.198167637934534e-06, | |
| "loss": 0.03348960727453232, | |
| "step": 200192 | |
| }, | |
| { | |
| "epoch": 5.811603026876576, | |
| "grad_norm": 1.2533525228500366, | |
| "learning_rate": 4.190745411846569e-06, | |
| "loss": 0.032671619206666946, | |
| "step": 200448 | |
| }, | |
| { | |
| "epoch": 5.819025252964542, | |
| "grad_norm": 1.268965721130371, | |
| "learning_rate": 4.183323185758604e-06, | |
| "loss": 0.03352795168757439, | |
| "step": 200704 | |
| }, | |
| { | |
| "epoch": 5.826447479052506, | |
| "grad_norm": 1.1504443883895874, | |
| "learning_rate": 4.175900959670639e-06, | |
| "loss": 0.03346399590373039, | |
| "step": 200960 | |
| }, | |
| { | |
| "epoch": 5.833869705140471, | |
| "grad_norm": 0.8690232038497925, | |
| "learning_rate": 4.168478733582675e-06, | |
| "loss": 0.03255928307771683, | |
| "step": 201216 | |
| }, | |
| { | |
| "epoch": 5.841291931228437, | |
| "grad_norm": 0.738191545009613, | |
| "learning_rate": 4.161056507494709e-06, | |
| "loss": 0.03213713690638542, | |
| "step": 201472 | |
| }, | |
| { | |
| "epoch": 5.848714157316401, | |
| "grad_norm": 1.3269877433776855, | |
| "learning_rate": 4.153634281406744e-06, | |
| "loss": 0.032123155891895294, | |
| "step": 201728 | |
| }, | |
| { | |
| "epoch": 5.856136383404366, | |
| "grad_norm": 0.9794167876243591, | |
| "learning_rate": 4.146212055318779e-06, | |
| "loss": 0.0331299714744091, | |
| "step": 201984 | |
| }, | |
| { | |
| "epoch": 5.8635586094923315, | |
| "grad_norm": 1.3606939315795898, | |
| "learning_rate": 4.138789829230814e-06, | |
| "loss": 0.03226563334465027, | |
| "step": 202240 | |
| }, | |
| { | |
| "epoch": 5.870980835580296, | |
| "grad_norm": 1.2537027597427368, | |
| "learning_rate": 4.131367603142849e-06, | |
| "loss": 0.03364162519574165, | |
| "step": 202496 | |
| }, | |
| { | |
| "epoch": 5.878403061668261, | |
| "grad_norm": 0.7907251119613647, | |
| "learning_rate": 4.12397437012554e-06, | |
| "loss": 0.03190918266773224, | |
| "step": 202752 | |
| }, | |
| { | |
| "epoch": 5.8858252877562265, | |
| "grad_norm": 0.8717405796051025, | |
| "learning_rate": 4.116552144037576e-06, | |
| "loss": 0.03183775395154953, | |
| "step": 203008 | |
| }, | |
| { | |
| "epoch": 5.893247513844191, | |
| "grad_norm": 1.6494139432907104, | |
| "learning_rate": 4.1091299179496105e-06, | |
| "loss": 0.031699661165475845, | |
| "step": 203264 | |
| }, | |
| { | |
| "epoch": 5.900669739932156, | |
| "grad_norm": 0.9705306887626648, | |
| "learning_rate": 4.101707691861645e-06, | |
| "loss": 0.031915683299303055, | |
| "step": 203520 | |
| }, | |
| { | |
| "epoch": 5.908091966020121, | |
| "grad_norm": 0.9495534896850586, | |
| "learning_rate": 4.09428546577368e-06, | |
| "loss": 0.03163834661245346, | |
| "step": 203776 | |
| }, | |
| { | |
| "epoch": 5.915514192108086, | |
| "grad_norm": 1.3784449100494385, | |
| "learning_rate": 4.0868632396857155e-06, | |
| "loss": 0.03247217833995819, | |
| "step": 204032 | |
| }, | |
| { | |
| "epoch": 5.922936418196051, | |
| "grad_norm": 1.424973726272583, | |
| "learning_rate": 4.07944101359775e-06, | |
| "loss": 0.03180554881691933, | |
| "step": 204288 | |
| }, | |
| { | |
| "epoch": 5.930358644284016, | |
| "grad_norm": 0.8632015585899353, | |
| "learning_rate": 4.072018787509786e-06, | |
| "loss": 0.03168933093547821, | |
| "step": 204544 | |
| }, | |
| { | |
| "epoch": 5.937780870371981, | |
| "grad_norm": 1.1267317533493042, | |
| "learning_rate": 4.0645965614218204e-06, | |
| "loss": 0.032534148544073105, | |
| "step": 204800 | |
| }, | |
| { | |
| "epoch": 5.945203096459946, | |
| "grad_norm": 1.1765679121017456, | |
| "learning_rate": 4.057174335333855e-06, | |
| "loss": 0.031553205102682114, | |
| "step": 205056 | |
| }, | |
| { | |
| "epoch": 5.952625322547911, | |
| "grad_norm": 1.0186113119125366, | |
| "learning_rate": 4.049781102316546e-06, | |
| "loss": 0.032411832362413406, | |
| "step": 205312 | |
| }, | |
| { | |
| "epoch": 5.960047548635876, | |
| "grad_norm": 0.7090544700622559, | |
| "learning_rate": 4.042358876228581e-06, | |
| "loss": 0.03235161677002907, | |
| "step": 205568 | |
| }, | |
| { | |
| "epoch": 5.967469774723841, | |
| "grad_norm": 1.1012688875198364, | |
| "learning_rate": 4.034936650140617e-06, | |
| "loss": 0.03152674809098244, | |
| "step": 205824 | |
| }, | |
| { | |
| "epoch": 5.974892000811806, | |
| "grad_norm": 1.1395506858825684, | |
| "learning_rate": 4.027514424052651e-06, | |
| "loss": 0.032106056809425354, | |
| "step": 206080 | |
| }, | |
| { | |
| "epoch": 5.982314226899771, | |
| "grad_norm": 3.751807451248169, | |
| "learning_rate": 4.020092197964687e-06, | |
| "loss": 0.032239604741334915, | |
| "step": 206336 | |
| }, | |
| { | |
| "epoch": 5.989736452987736, | |
| "grad_norm": 0.7881655097007751, | |
| "learning_rate": 4.012669971876722e-06, | |
| "loss": 0.03217990696430206, | |
| "step": 206592 | |
| }, | |
| { | |
| "epoch": 5.997158679075701, | |
| "grad_norm": 0.9459748864173889, | |
| "learning_rate": 4.005247745788757e-06, | |
| "loss": 0.03224950283765793, | |
| "step": 206848 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_f1_B-DISEASE": 0.8805179832969154, | |
| "eval_f1_B-PROCEDURE": 0.8657260263042409, | |
| "eval_f1_B-SYMPTOM": 0.8665456745311555, | |
| "eval_f1_I-DISEASE": 0.9087752510887719, | |
| "eval_f1_I-PROCEDURE": 0.9016105755664426, | |
| "eval_f1_I-SYMPTOM": 0.8859550476995532, | |
| "eval_f1_O": 0.9498061892599832, | |
| "eval_f1_macro": 0.8941338211067232, | |
| "eval_f1_micro": 0.9255500049949587, | |
| "eval_loss": 0.07590102404356003, | |
| "eval_precision_B-DISEASE": 0.8886062934603209, | |
| "eval_precision_B-PROCEDURE": 0.8686949559647719, | |
| "eval_precision_B-SYMPTOM": 0.8762707716948813, | |
| "eval_precision_I-DISEASE": 0.8985807879911446, | |
| "eval_precision_I-PROCEDURE": 0.9032247642944449, | |
| "eval_precision_I-SYMPTOM": 0.8922854145763021, | |
| "eval_precision_O": 0.9516515065159123, | |
| "eval_precision_macro": 0.8970449277853968, | |
| "eval_precision_micro": 0.9269212748963036, | |
| "eval_rauc_macro": 0.9370242986879962, | |
| "eval_rauc_micro": 0.9558014170514155, | |
| "eval_recall_B-DISEASE": 0.872575588500983, | |
| "eval_recall_B-PROCEDURE": 0.8627773212818406, | |
| "eval_recall_B-SYMPTOM": 0.8570340716243854, | |
| "eval_recall_I-DISEASE": 0.9192036823935558, | |
| "eval_recall_I-PROCEDURE": 0.9000021461069619, | |
| "eval_recall_I-SYMPTOM": 0.8797138703198537, | |
| "eval_recall_O": 0.9479680145465147, | |
| "eval_recall_macro": 0.8913249563962993, | |
| "eval_recall_micro": 0.9241827863620482, | |
| "eval_roc_auc_B-DISEASE": 0.9351063784914326, | |
| "eval_roc_auc_B-PROCEDURE": 0.9299767436019514, | |
| "eval_roc_auc_B-SYMPTOM": 0.9269133007992723, | |
| "eval_roc_auc_I-DISEASE": 0.952390694970384, | |
| "eval_roc_auc_I-PROCEDURE": 0.9443621486940681, | |
| "eval_roc_auc_I-SYMPTOM": 0.9304921393376546, | |
| "eval_roc_auc_O": 0.9399286849212097, | |
| "eval_runtime": 53.6119, | |
| "eval_samples_per_second": 263.934, | |
| "eval_steps_per_second": 32.996, | |
| "step": 206946 | |
| }, | |
| { | |
| "epoch": 6.004580905163666, | |
| "grad_norm": 1.3526554107666016, | |
| "learning_rate": 3.997825519700792e-06, | |
| "loss": 0.028902210295200348, | |
| "step": 207104 | |
| }, | |
| { | |
| "epoch": 6.012003131251631, | |
| "grad_norm": 1.0079988241195679, | |
| "learning_rate": 3.9904032936128266e-06, | |
| "loss": 0.028314389288425446, | |
| "step": 207360 | |
| }, | |
| { | |
| "epoch": 6.019425357339596, | |
| "grad_norm": 0.9535010457038879, | |
| "learning_rate": 3.983010060595518e-06, | |
| "loss": 0.027740446850657463, | |
| "step": 207616 | |
| }, | |
| { | |
| "epoch": 6.0268475834275606, | |
| "grad_norm": 0.5832136869430542, | |
| "learning_rate": 3.975587834507553e-06, | |
| "loss": 0.02724306657910347, | |
| "step": 207872 | |
| }, | |
| { | |
| "epoch": 6.034269809515526, | |
| "grad_norm": 2.147702693939209, | |
| "learning_rate": 3.968165608419588e-06, | |
| "loss": 0.027774613350629807, | |
| "step": 208128 | |
| }, | |
| { | |
| "epoch": 6.041692035603491, | |
| "grad_norm": 1.0929224491119385, | |
| "learning_rate": 3.960743382331624e-06, | |
| "loss": 0.028861574828624725, | |
| "step": 208384 | |
| }, | |
| { | |
| "epoch": 6.0491142616914555, | |
| "grad_norm": 0.3395084738731384, | |
| "learning_rate": 3.953321156243658e-06, | |
| "loss": 0.027403783053159714, | |
| "step": 208640 | |
| }, | |
| { | |
| "epoch": 6.056536487779421, | |
| "grad_norm": 1.6763341426849365, | |
| "learning_rate": 3.945898930155693e-06, | |
| "loss": 0.027653271332383156, | |
| "step": 208896 | |
| }, | |
| { | |
| "epoch": 6.063958713867386, | |
| "grad_norm": 0.7062895894050598, | |
| "learning_rate": 3.938476704067728e-06, | |
| "loss": 0.027056170627474785, | |
| "step": 209152 | |
| }, | |
| { | |
| "epoch": 6.07138093995535, | |
| "grad_norm": 1.0973756313323975, | |
| "learning_rate": 3.931054477979763e-06, | |
| "loss": 0.027962762862443924, | |
| "step": 209408 | |
| }, | |
| { | |
| "epoch": 6.078803166043316, | |
| "grad_norm": 0.7082591652870178, | |
| "learning_rate": 3.923632251891798e-06, | |
| "loss": 0.02705645188689232, | |
| "step": 209664 | |
| }, | |
| { | |
| "epoch": 6.086225392131281, | |
| "grad_norm": 1.0970408916473389, | |
| "learning_rate": 3.916239018874489e-06, | |
| "loss": 0.028795413672924042, | |
| "step": 209920 | |
| }, | |
| { | |
| "epoch": 6.093647618219245, | |
| "grad_norm": 0.7350139617919922, | |
| "learning_rate": 3.908816792786525e-06, | |
| "loss": 0.02743448317050934, | |
| "step": 210176 | |
| }, | |
| { | |
| "epoch": 6.101069844307211, | |
| "grad_norm": 0.5603310465812683, | |
| "learning_rate": 3.9013945666985595e-06, | |
| "loss": 0.02784096822142601, | |
| "step": 210432 | |
| }, | |
| { | |
| "epoch": 6.108492070395176, | |
| "grad_norm": 0.9388706088066101, | |
| "learning_rate": 3.893972340610594e-06, | |
| "loss": 0.026850486174225807, | |
| "step": 210688 | |
| }, | |
| { | |
| "epoch": 6.11591429648314, | |
| "grad_norm": 1.0053739547729492, | |
| "learning_rate": 3.886550114522629e-06, | |
| "loss": 0.027635516598820686, | |
| "step": 210944 | |
| }, | |
| { | |
| "epoch": 6.123336522571106, | |
| "grad_norm": 0.8339763283729553, | |
| "learning_rate": 3.8791278884346645e-06, | |
| "loss": 0.02769089862704277, | |
| "step": 211200 | |
| }, | |
| { | |
| "epoch": 6.130758748659071, | |
| "grad_norm": 1.2035611867904663, | |
| "learning_rate": 3.871705662346699e-06, | |
| "loss": 0.027635926380753517, | |
| "step": 211456 | |
| }, | |
| { | |
| "epoch": 6.138180974747035, | |
| "grad_norm": 0.7306374907493591, | |
| "learning_rate": 3.864283436258735e-06, | |
| "loss": 0.026829054579138756, | |
| "step": 211712 | |
| }, | |
| { | |
| "epoch": 6.145603200835001, | |
| "grad_norm": 5.125953197479248, | |
| "learning_rate": 3.856890203241426e-06, | |
| "loss": 0.02787015214562416, | |
| "step": 211968 | |
| }, | |
| { | |
| "epoch": 6.153025426922966, | |
| "grad_norm": 1.7391374111175537, | |
| "learning_rate": 3.849467977153461e-06, | |
| "loss": 0.02728964202105999, | |
| "step": 212224 | |
| }, | |
| { | |
| "epoch": 6.16044765301093, | |
| "grad_norm": 1.943947196006775, | |
| "learning_rate": 3.842045751065495e-06, | |
| "loss": 0.027082830667495728, | |
| "step": 212480 | |
| }, | |
| { | |
| "epoch": 6.1678698790988955, | |
| "grad_norm": 1.370179295539856, | |
| "learning_rate": 3.83462352497753e-06, | |
| "loss": 0.027254845947027206, | |
| "step": 212736 | |
| }, | |
| { | |
| "epoch": 6.175292105186861, | |
| "grad_norm": 1.6606528759002686, | |
| "learning_rate": 3.827201298889566e-06, | |
| "loss": 0.02708587795495987, | |
| "step": 212992 | |
| }, | |
| { | |
| "epoch": 6.182714331274825, | |
| "grad_norm": 1.5031845569610596, | |
| "learning_rate": 3.819779072801601e-06, | |
| "loss": 0.027441246435046196, | |
| "step": 213248 | |
| }, | |
| { | |
| "epoch": 6.1901365573627904, | |
| "grad_norm": 1.34352445602417, | |
| "learning_rate": 3.8123568467136355e-06, | |
| "loss": 0.02759948931634426, | |
| "step": 213504 | |
| }, | |
| { | |
| "epoch": 6.197558783450756, | |
| "grad_norm": 0.7863252758979797, | |
| "learning_rate": 3.804934620625671e-06, | |
| "loss": 0.027000058442354202, | |
| "step": 213760 | |
| }, | |
| { | |
| "epoch": 6.20498100953872, | |
| "grad_norm": 1.3386611938476562, | |
| "learning_rate": 3.7975123945377058e-06, | |
| "loss": 0.02707917056977749, | |
| "step": 214016 | |
| }, | |
| { | |
| "epoch": 6.212403235626685, | |
| "grad_norm": 1.3500096797943115, | |
| "learning_rate": 3.790119161520397e-06, | |
| "loss": 0.027128679677844048, | |
| "step": 214272 | |
| }, | |
| { | |
| "epoch": 6.21982546171465, | |
| "grad_norm": 1.288244605064392, | |
| "learning_rate": 3.782696935432432e-06, | |
| "loss": 0.02706947736442089, | |
| "step": 214528 | |
| }, | |
| { | |
| "epoch": 6.227247687802615, | |
| "grad_norm": 0.9172043204307556, | |
| "learning_rate": 3.775274709344467e-06, | |
| "loss": 0.027603566646575928, | |
| "step": 214784 | |
| }, | |
| { | |
| "epoch": 6.23466991389058, | |
| "grad_norm": 0.8866100907325745, | |
| "learning_rate": 3.7678524832565024e-06, | |
| "loss": 0.029069341719150543, | |
| "step": 215040 | |
| }, | |
| { | |
| "epoch": 6.242092139978545, | |
| "grad_norm": 0.5434070229530334, | |
| "learning_rate": 3.760430257168537e-06, | |
| "loss": 0.027310721576213837, | |
| "step": 215296 | |
| }, | |
| { | |
| "epoch": 6.24951436606651, | |
| "grad_norm": 1.119490146636963, | |
| "learning_rate": 3.7530370241512284e-06, | |
| "loss": 0.02869304083287716, | |
| "step": 215552 | |
| }, | |
| { | |
| "epoch": 6.256936592154475, | |
| "grad_norm": 1.2084109783172607, | |
| "learning_rate": 3.7456147980632635e-06, | |
| "loss": 0.0274447500705719, | |
| "step": 215808 | |
| }, | |
| { | |
| "epoch": 6.26435881824244, | |
| "grad_norm": 1.7319245338439941, | |
| "learning_rate": 3.738192571975298e-06, | |
| "loss": 0.02777915634214878, | |
| "step": 216064 | |
| }, | |
| { | |
| "epoch": 6.271781044330405, | |
| "grad_norm": 0.6577815413475037, | |
| "learning_rate": 3.7307703458873333e-06, | |
| "loss": 0.026969220489263535, | |
| "step": 216320 | |
| }, | |
| { | |
| "epoch": 6.27920327041837, | |
| "grad_norm": 1.1062045097351074, | |
| "learning_rate": 3.723348119799368e-06, | |
| "loss": 0.02718285098671913, | |
| "step": 216576 | |
| }, | |
| { | |
| "epoch": 6.286625496506335, | |
| "grad_norm": 0.8376836776733398, | |
| "learning_rate": 3.7159258937114036e-06, | |
| "loss": 0.027209943160414696, | |
| "step": 216832 | |
| }, | |
| { | |
| "epoch": 6.2940477225943, | |
| "grad_norm": 1.664854645729065, | |
| "learning_rate": 3.7085036676234383e-06, | |
| "loss": 0.02681136690080166, | |
| "step": 217088 | |
| }, | |
| { | |
| "epoch": 6.301469948682265, | |
| "grad_norm": 0.8556686639785767, | |
| "learning_rate": 3.7010814415354734e-06, | |
| "loss": 0.026085693389177322, | |
| "step": 217344 | |
| }, | |
| { | |
| "epoch": 6.30889217477023, | |
| "grad_norm": 1.363673448562622, | |
| "learning_rate": 3.693659215447508e-06, | |
| "loss": 0.027581091970205307, | |
| "step": 217600 | |
| }, | |
| { | |
| "epoch": 6.316314400858195, | |
| "grad_norm": 0.9876484870910645, | |
| "learning_rate": 3.6862369893595432e-06, | |
| "loss": 0.027108732610940933, | |
| "step": 217856 | |
| }, | |
| { | |
| "epoch": 6.32373662694616, | |
| "grad_norm": 0.8539023399353027, | |
| "learning_rate": 3.6788147632715784e-06, | |
| "loss": 0.0267562884837389, | |
| "step": 218112 | |
| }, | |
| { | |
| "epoch": 6.3311588530341245, | |
| "grad_norm": 1.0268065929412842, | |
| "learning_rate": 3.6713925371836135e-06, | |
| "loss": 0.02615993842482567, | |
| "step": 218368 | |
| }, | |
| { | |
| "epoch": 6.33858107912209, | |
| "grad_norm": 0.6833095550537109, | |
| "learning_rate": 3.6639703110956486e-06, | |
| "loss": 0.027816491201519966, | |
| "step": 218624 | |
| }, | |
| { | |
| "epoch": 6.346003305210055, | |
| "grad_norm": 2.2063663005828857, | |
| "learning_rate": 3.6565480850076833e-06, | |
| "loss": 0.02738560363650322, | |
| "step": 218880 | |
| }, | |
| { | |
| "epoch": 6.3534255312980195, | |
| "grad_norm": 1.5379475355148315, | |
| "learning_rate": 3.649125858919719e-06, | |
| "loss": 0.026971235871315002, | |
| "step": 219136 | |
| }, | |
| { | |
| "epoch": 6.360847757385985, | |
| "grad_norm": 0.8156259655952454, | |
| "learning_rate": 3.6417036328317536e-06, | |
| "loss": 0.026984762400388718, | |
| "step": 219392 | |
| }, | |
| { | |
| "epoch": 6.36826998347395, | |
| "grad_norm": 0.6960588097572327, | |
| "learning_rate": 3.6342814067437887e-06, | |
| "loss": 0.02737814001739025, | |
| "step": 219648 | |
| }, | |
| { | |
| "epoch": 6.375692209561914, | |
| "grad_norm": 1.1413209438323975, | |
| "learning_rate": 3.6268591806558234e-06, | |
| "loss": 0.026520228013396263, | |
| "step": 219904 | |
| }, | |
| { | |
| "epoch": 6.38311443564988, | |
| "grad_norm": 0.5902450084686279, | |
| "learning_rate": 3.6194369545678585e-06, | |
| "loss": 0.02662413939833641, | |
| "step": 220160 | |
| }, | |
| { | |
| "epoch": 6.390536661737845, | |
| "grad_norm": 0.8722751140594482, | |
| "learning_rate": 3.6120147284798932e-06, | |
| "loss": 0.027573632076382637, | |
| "step": 220416 | |
| }, | |
| { | |
| "epoch": 6.397958887825809, | |
| "grad_norm": 0.7500815987586975, | |
| "learning_rate": 3.6046214954625845e-06, | |
| "loss": 0.025736436247825623, | |
| "step": 220672 | |
| }, | |
| { | |
| "epoch": 6.405381113913775, | |
| "grad_norm": 1.2666388750076294, | |
| "learning_rate": 3.59719926937462e-06, | |
| "loss": 0.02646820992231369, | |
| "step": 220928 | |
| }, | |
| { | |
| "epoch": 6.41280334000174, | |
| "grad_norm": 0.8173425793647766, | |
| "learning_rate": 3.5897770432866548e-06, | |
| "loss": 0.025433262810111046, | |
| "step": 221184 | |
| }, | |
| { | |
| "epoch": 6.420225566089704, | |
| "grad_norm": 0.7605726718902588, | |
| "learning_rate": 3.58235481719869e-06, | |
| "loss": 0.026508865877985954, | |
| "step": 221440 | |
| }, | |
| { | |
| "epoch": 6.42764779217767, | |
| "grad_norm": 1.0714449882507324, | |
| "learning_rate": 3.5749325911107246e-06, | |
| "loss": 0.026653369888663292, | |
| "step": 221696 | |
| }, | |
| { | |
| "epoch": 6.435070018265635, | |
| "grad_norm": 0.9441837668418884, | |
| "learning_rate": 3.56751036502276e-06, | |
| "loss": 0.026864144951105118, | |
| "step": 221952 | |
| }, | |
| { | |
| "epoch": 6.442492244353599, | |
| "grad_norm": 0.9714749455451965, | |
| "learning_rate": 3.560088138934795e-06, | |
| "loss": 0.02659711241722107, | |
| "step": 222208 | |
| }, | |
| { | |
| "epoch": 6.4499144704415645, | |
| "grad_norm": 1.506098985671997, | |
| "learning_rate": 3.55266591284683e-06, | |
| "loss": 0.027149107307195663, | |
| "step": 222464 | |
| }, | |
| { | |
| "epoch": 6.45733669652953, | |
| "grad_norm": 1.2610422372817993, | |
| "learning_rate": 3.5452436867588647e-06, | |
| "loss": 0.026045454666018486, | |
| "step": 222720 | |
| }, | |
| { | |
| "epoch": 6.464758922617494, | |
| "grad_norm": 1.1605359315872192, | |
| "learning_rate": 3.5378214606709e-06, | |
| "loss": 0.025891808792948723, | |
| "step": 222976 | |
| }, | |
| { | |
| "epoch": 6.4721811487054595, | |
| "grad_norm": 1.030522108078003, | |
| "learning_rate": 3.5303992345829354e-06, | |
| "loss": 0.02475164830684662, | |
| "step": 223232 | |
| }, | |
| { | |
| "epoch": 6.479603374793425, | |
| "grad_norm": 1.5318316221237183, | |
| "learning_rate": 3.52297700849497e-06, | |
| "loss": 0.026338962838053703, | |
| "step": 223488 | |
| }, | |
| { | |
| "epoch": 6.487025600881389, | |
| "grad_norm": 2.0347232818603516, | |
| "learning_rate": 3.515554782407005e-06, | |
| "loss": 0.026244593784213066, | |
| "step": 223744 | |
| }, | |
| { | |
| "epoch": 6.494447826969354, | |
| "grad_norm": 1.3294179439544678, | |
| "learning_rate": 3.50813255631904e-06, | |
| "loss": 0.02654501609504223, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 6.50187005305732, | |
| "grad_norm": 1.4940099716186523, | |
| "learning_rate": 3.500710330231075e-06, | |
| "loss": 0.026277676224708557, | |
| "step": 224256 | |
| }, | |
| { | |
| "epoch": 6.509292279145284, | |
| "grad_norm": 2.8128721714019775, | |
| "learning_rate": 3.4933170972137663e-06, | |
| "loss": 0.025952383875846863, | |
| "step": 224512 | |
| }, | |
| { | |
| "epoch": 6.516714505233249, | |
| "grad_norm": 1.3695175647735596, | |
| "learning_rate": 3.485894871125801e-06, | |
| "loss": 0.025985803455114365, | |
| "step": 224768 | |
| }, | |
| { | |
| "epoch": 6.524136731321214, | |
| "grad_norm": 0.7925586104393005, | |
| "learning_rate": 3.4784726450378365e-06, | |
| "loss": 0.026154454797506332, | |
| "step": 225024 | |
| }, | |
| { | |
| "epoch": 6.531558957409179, | |
| "grad_norm": 0.8483281135559082, | |
| "learning_rate": 3.4710504189498712e-06, | |
| "loss": 0.026133989915251732, | |
| "step": 225280 | |
| }, | |
| { | |
| "epoch": 6.538981183497144, | |
| "grad_norm": 0.8360992074012756, | |
| "learning_rate": 3.4636281928619064e-06, | |
| "loss": 0.025212224572896957, | |
| "step": 225536 | |
| }, | |
| { | |
| "epoch": 6.546403409585109, | |
| "grad_norm": 1.0589946508407593, | |
| "learning_rate": 3.456205966773941e-06, | |
| "loss": 0.025941988453269005, | |
| "step": 225792 | |
| }, | |
| { | |
| "epoch": 6.553825635673074, | |
| "grad_norm": 1.3036407232284546, | |
| "learning_rate": 3.4487837406859766e-06, | |
| "loss": 0.02647586353123188, | |
| "step": 226048 | |
| }, | |
| { | |
| "epoch": 6.561247861761039, | |
| "grad_norm": 1.1977921724319458, | |
| "learning_rate": 3.4413615145980113e-06, | |
| "loss": 0.02536041848361492, | |
| "step": 226304 | |
| }, | |
| { | |
| "epoch": 6.568670087849004, | |
| "grad_norm": 1.0261441469192505, | |
| "learning_rate": 3.4339392885100465e-06, | |
| "loss": 0.025931160897016525, | |
| "step": 226560 | |
| }, | |
| { | |
| "epoch": 6.576092313936969, | |
| "grad_norm": 1.670589804649353, | |
| "learning_rate": 3.426517062422081e-06, | |
| "loss": 0.02533942461013794, | |
| "step": 226816 | |
| }, | |
| { | |
| "epoch": 6.583514540024934, | |
| "grad_norm": 0.9306561350822449, | |
| "learning_rate": 3.4190948363341163e-06, | |
| "loss": 0.02631131187081337, | |
| "step": 227072 | |
| }, | |
| { | |
| "epoch": 6.590936766112899, | |
| "grad_norm": 1.3584060668945312, | |
| "learning_rate": 3.411672610246152e-06, | |
| "loss": 0.026502396911382675, | |
| "step": 227328 | |
| }, | |
| { | |
| "epoch": 6.598358992200864, | |
| "grad_norm": 0.9885308146476746, | |
| "learning_rate": 3.4042503841581865e-06, | |
| "loss": 0.025215620175004005, | |
| "step": 227584 | |
| }, | |
| { | |
| "epoch": 6.605781218288829, | |
| "grad_norm": 3.159776449203491, | |
| "learning_rate": 3.3968281580702217e-06, | |
| "loss": 0.024496247991919518, | |
| "step": 227840 | |
| }, | |
| { | |
| "epoch": 6.6132034443767935, | |
| "grad_norm": 1.2466294765472412, | |
| "learning_rate": 3.3894349250529125e-06, | |
| "loss": 0.025045378133654594, | |
| "step": 228096 | |
| }, | |
| { | |
| "epoch": 6.620625670464759, | |
| "grad_norm": 0.8900142908096313, | |
| "learning_rate": 3.3820126989649476e-06, | |
| "loss": 0.026400478556752205, | |
| "step": 228352 | |
| }, | |
| { | |
| "epoch": 6.628047896552724, | |
| "grad_norm": 1.8306187391281128, | |
| "learning_rate": 3.3745904728769828e-06, | |
| "loss": 0.025547971948981285, | |
| "step": 228608 | |
| }, | |
| { | |
| "epoch": 6.6354701226406885, | |
| "grad_norm": 0.4828698933124542, | |
| "learning_rate": 3.3671682467890175e-06, | |
| "loss": 0.024781465530395508, | |
| "step": 228864 | |
| }, | |
| { | |
| "epoch": 6.642892348728654, | |
| "grad_norm": 0.6588764786720276, | |
| "learning_rate": 3.359746020701053e-06, | |
| "loss": 0.026618916541337967, | |
| "step": 229120 | |
| }, | |
| { | |
| "epoch": 6.650314574816619, | |
| "grad_norm": 2.0483646392822266, | |
| "learning_rate": 3.3523237946130877e-06, | |
| "loss": 0.025891058146953583, | |
| "step": 229376 | |
| }, | |
| { | |
| "epoch": 6.657736800904583, | |
| "grad_norm": 0.9484197497367859, | |
| "learning_rate": 3.344901568525123e-06, | |
| "loss": 0.02605433017015457, | |
| "step": 229632 | |
| }, | |
| { | |
| "epoch": 6.665159026992549, | |
| "grad_norm": 0.4335108697414398, | |
| "learning_rate": 3.3374793424371576e-06, | |
| "loss": 0.025832410901784897, | |
| "step": 229888 | |
| }, | |
| { | |
| "epoch": 6.672581253080514, | |
| "grad_norm": 1.0268943309783936, | |
| "learning_rate": 3.330057116349193e-06, | |
| "loss": 0.02432437613606453, | |
| "step": 230144 | |
| }, | |
| { | |
| "epoch": 6.680003479168478, | |
| "grad_norm": 1.2779659032821655, | |
| "learning_rate": 3.322634890261228e-06, | |
| "loss": 0.02511880360543728, | |
| "step": 230400 | |
| }, | |
| { | |
| "epoch": 6.687425705256444, | |
| "grad_norm": 1.468088150024414, | |
| "learning_rate": 3.315212664173263e-06, | |
| "loss": 0.02572665363550186, | |
| "step": 230656 | |
| }, | |
| { | |
| "epoch": 6.694847931344409, | |
| "grad_norm": 1.0313997268676758, | |
| "learning_rate": 3.307819431155954e-06, | |
| "loss": 0.025244878605008125, | |
| "step": 230912 | |
| }, | |
| { | |
| "epoch": 6.702270157432373, | |
| "grad_norm": 0.9503042697906494, | |
| "learning_rate": 3.3004261981386455e-06, | |
| "loss": 0.025702176615595818, | |
| "step": 231168 | |
| }, | |
| { | |
| "epoch": 6.709692383520339, | |
| "grad_norm": 3.60322642326355, | |
| "learning_rate": 3.29300397205068e-06, | |
| "loss": 0.024175329133868217, | |
| "step": 231424 | |
| }, | |
| { | |
| "epoch": 6.717114609608304, | |
| "grad_norm": 0.6954478025436401, | |
| "learning_rate": 3.2855817459627153e-06, | |
| "loss": 0.025860166177153587, | |
| "step": 231680 | |
| }, | |
| { | |
| "epoch": 6.724536835696268, | |
| "grad_norm": 1.0973129272460938, | |
| "learning_rate": 3.27815951987475e-06, | |
| "loss": 0.025737272575497627, | |
| "step": 231936 | |
| }, | |
| { | |
| "epoch": 6.7319590617842335, | |
| "grad_norm": 2.1501033306121826, | |
| "learning_rate": 3.2707372937867856e-06, | |
| "loss": 0.02551724761724472, | |
| "step": 232192 | |
| }, | |
| { | |
| "epoch": 6.739381287872199, | |
| "grad_norm": 1.346191167831421, | |
| "learning_rate": 3.2633150676988203e-06, | |
| "loss": 0.024873437359929085, | |
| "step": 232448 | |
| }, | |
| { | |
| "epoch": 6.746803513960163, | |
| "grad_norm": 1.3468310832977295, | |
| "learning_rate": 3.2558928416108554e-06, | |
| "loss": 0.024750404059886932, | |
| "step": 232704 | |
| }, | |
| { | |
| "epoch": 6.7542257400481285, | |
| "grad_norm": 0.7212878465652466, | |
| "learning_rate": 3.24847061552289e-06, | |
| "loss": 0.02438187226653099, | |
| "step": 232960 | |
| }, | |
| { | |
| "epoch": 6.761647966136094, | |
| "grad_norm": 0.7745961546897888, | |
| "learning_rate": 3.2410483894349252e-06, | |
| "loss": 0.024748967960476875, | |
| "step": 233216 | |
| }, | |
| { | |
| "epoch": 6.769070192224058, | |
| "grad_norm": 1.1349756717681885, | |
| "learning_rate": 3.23362616334696e-06, | |
| "loss": 0.025096165016293526, | |
| "step": 233472 | |
| }, | |
| { | |
| "epoch": 6.776492418312023, | |
| "grad_norm": 1.4615886211395264, | |
| "learning_rate": 3.2262039372589955e-06, | |
| "loss": 0.024579860270023346, | |
| "step": 233728 | |
| }, | |
| { | |
| "epoch": 6.783914644399989, | |
| "grad_norm": 0.7326033115386963, | |
| "learning_rate": 3.21878171117103e-06, | |
| "loss": 0.02599087916314602, | |
| "step": 233984 | |
| }, | |
| { | |
| "epoch": 6.791336870487953, | |
| "grad_norm": 0.9617591500282288, | |
| "learning_rate": 3.2113594850830653e-06, | |
| "loss": 0.025381196290254593, | |
| "step": 234240 | |
| }, | |
| { | |
| "epoch": 6.798759096575918, | |
| "grad_norm": 0.754433810710907, | |
| "learning_rate": 3.203937258995101e-06, | |
| "loss": 0.024315977469086647, | |
| "step": 234496 | |
| }, | |
| { | |
| "epoch": 6.806181322663884, | |
| "grad_norm": 0.8312054872512817, | |
| "learning_rate": 3.1965150329071356e-06, | |
| "loss": 0.024831917136907578, | |
| "step": 234752 | |
| }, | |
| { | |
| "epoch": 6.813603548751848, | |
| "grad_norm": 0.9501178860664368, | |
| "learning_rate": 3.1890928068191707e-06, | |
| "loss": 0.025097619742155075, | |
| "step": 235008 | |
| }, | |
| { | |
| "epoch": 6.821025774839813, | |
| "grad_norm": 1.5179344415664673, | |
| "learning_rate": 3.1816705807312054e-06, | |
| "loss": 0.025188904255628586, | |
| "step": 235264 | |
| }, | |
| { | |
| "epoch": 6.828448000927779, | |
| "grad_norm": 0.7843024134635925, | |
| "learning_rate": 3.1742483546432405e-06, | |
| "loss": 0.02561766281723976, | |
| "step": 235520 | |
| }, | |
| { | |
| "epoch": 6.835870227015743, | |
| "grad_norm": 1.9117580652236938, | |
| "learning_rate": 3.1668261285552752e-06, | |
| "loss": 0.024742627516388893, | |
| "step": 235776 | |
| }, | |
| { | |
| "epoch": 6.843292453103708, | |
| "grad_norm": 0.8089895844459534, | |
| "learning_rate": 3.1594328955379665e-06, | |
| "loss": 0.025844300165772438, | |
| "step": 236032 | |
| }, | |
| { | |
| "epoch": 6.8507146791916735, | |
| "grad_norm": 0.8744811415672302, | |
| "learning_rate": 3.152010669450002e-06, | |
| "loss": 0.023944813758134842, | |
| "step": 236288 | |
| }, | |
| { | |
| "epoch": 6.858136905279638, | |
| "grad_norm": 1.154313564300537, | |
| "learning_rate": 3.1445884433620367e-06, | |
| "loss": 0.024763792753219604, | |
| "step": 236544 | |
| }, | |
| { | |
| "epoch": 6.865559131367603, | |
| "grad_norm": 1.172471523284912, | |
| "learning_rate": 3.137166217274072e-06, | |
| "loss": 0.024448392912745476, | |
| "step": 236800 | |
| }, | |
| { | |
| "epoch": 6.8729813574555685, | |
| "grad_norm": 1.6809154748916626, | |
| "learning_rate": 3.1297439911861066e-06, | |
| "loss": 0.024603597819805145, | |
| "step": 237056 | |
| }, | |
| { | |
| "epoch": 6.880403583543533, | |
| "grad_norm": 1.0935890674591064, | |
| "learning_rate": 3.1223217650981417e-06, | |
| "loss": 0.02462896518409252, | |
| "step": 237312 | |
| }, | |
| { | |
| "epoch": 6.887825809631498, | |
| "grad_norm": 2.866032361984253, | |
| "learning_rate": 3.1148995390101764e-06, | |
| "loss": 0.024592090398073196, | |
| "step": 237568 | |
| }, | |
| { | |
| "epoch": 6.895248035719463, | |
| "grad_norm": 0.8147240281105042, | |
| "learning_rate": 3.107477312922212e-06, | |
| "loss": 0.024236580356955528, | |
| "step": 237824 | |
| }, | |
| { | |
| "epoch": 6.902670261807428, | |
| "grad_norm": 0.8196598291397095, | |
| "learning_rate": 3.1000550868342467e-06, | |
| "loss": 0.023431329056620598, | |
| "step": 238080 | |
| }, | |
| { | |
| "epoch": 6.910092487895393, | |
| "grad_norm": 0.8307262659072876, | |
| "learning_rate": 3.092632860746282e-06, | |
| "loss": 0.024313384667038918, | |
| "step": 238336 | |
| }, | |
| { | |
| "epoch": 6.917514713983358, | |
| "grad_norm": 0.7427026033401489, | |
| "learning_rate": 3.0852106346583173e-06, | |
| "loss": 0.023867307230830193, | |
| "step": 238592 | |
| }, | |
| { | |
| "epoch": 6.924936940071323, | |
| "grad_norm": 0.6474505066871643, | |
| "learning_rate": 3.077788408570352e-06, | |
| "loss": 0.02389070764183998, | |
| "step": 238848 | |
| }, | |
| { | |
| "epoch": 6.932359166159288, | |
| "grad_norm": 1.5184189081192017, | |
| "learning_rate": 3.070366182482387e-06, | |
| "loss": 0.024250920861959457, | |
| "step": 239104 | |
| }, | |
| { | |
| "epoch": 6.939781392247253, | |
| "grad_norm": 0.5383797883987427, | |
| "learning_rate": 3.062943956394422e-06, | |
| "loss": 0.023970460519194603, | |
| "step": 239360 | |
| }, | |
| { | |
| "epoch": 6.947203618335218, | |
| "grad_norm": 1.1781668663024902, | |
| "learning_rate": 3.055521730306457e-06, | |
| "loss": 0.024724142625927925, | |
| "step": 239616 | |
| }, | |
| { | |
| "epoch": 6.954625844423183, | |
| "grad_norm": 0.9853819012641907, | |
| "learning_rate": 3.0480995042184917e-06, | |
| "loss": 0.023510104045271873, | |
| "step": 239872 | |
| }, | |
| { | |
| "epoch": 6.962048070511148, | |
| "grad_norm": 0.7056636214256287, | |
| "learning_rate": 3.040706271201183e-06, | |
| "loss": 0.024097319692373276, | |
| "step": 240128 | |
| }, | |
| { | |
| "epoch": 6.969470296599113, | |
| "grad_norm": 1.941941499710083, | |
| "learning_rate": 3.0333130381838742e-06, | |
| "loss": 0.024600131437182426, | |
| "step": 240384 | |
| }, | |
| { | |
| "epoch": 6.976892522687078, | |
| "grad_norm": 1.7744487524032593, | |
| "learning_rate": 3.025890812095909e-06, | |
| "loss": 0.024313684552907944, | |
| "step": 240640 | |
| }, | |
| { | |
| "epoch": 6.984314748775043, | |
| "grad_norm": 1.317827820777893, | |
| "learning_rate": 3.0184685860079445e-06, | |
| "loss": 0.02359217405319214, | |
| "step": 240896 | |
| }, | |
| { | |
| "epoch": 6.991736974863008, | |
| "grad_norm": 0.4502719044685364, | |
| "learning_rate": 3.0110463599199796e-06, | |
| "loss": 0.024374496191740036, | |
| "step": 241152 | |
| }, | |
| { | |
| "epoch": 6.999159200950973, | |
| "grad_norm": 1.0133689641952515, | |
| "learning_rate": 3.0036241338320143e-06, | |
| "loss": 0.024222668260335922, | |
| "step": 241408 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_f1_B-DISEASE": 0.8870616943398228, | |
| "eval_f1_B-PROCEDURE": 0.8694186802542362, | |
| "eval_f1_B-SYMPTOM": 0.8698900277951727, | |
| "eval_f1_I-DISEASE": 0.9134480497003042, | |
| "eval_f1_I-PROCEDURE": 0.9037418992178928, | |
| "eval_f1_I-SYMPTOM": 0.890687621020399, | |
| "eval_f1_O": 0.9515695119041204, | |
| "eval_f1_macro": 0.8979739263188498, | |
| "eval_f1_micro": 0.9282589179424787, | |
| "eval_loss": 0.07676578313112259, | |
| "eval_precision_B-DISEASE": 0.8855584388073929, | |
| "eval_precision_B-PROCEDURE": 0.8704697629928791, | |
| "eval_precision_B-SYMPTOM": 0.878586804553827, | |
| "eval_precision_I-DISEASE": 0.903953809124591, | |
| "eval_precision_I-PROCEDURE": 0.9104888859846876, | |
| "eval_precision_I-SYMPTOM": 0.8919863507906921, | |
| "eval_precision_O": 0.9547111452633693, | |
| "eval_precision_macro": 0.8993935996453484, | |
| "eval_precision_micro": 0.9299372465394238, | |
| "eval_rauc_macro": 0.9400983542920119, | |
| "eval_rauc_micro": 0.9572668537479893, | |
| "eval_recall_B-DISEASE": 0.8885700621712099, | |
| "eval_recall_B-PROCEDURE": 0.8683701327961407, | |
| "eval_recall_B-SYMPTOM": 0.8613637352595622, | |
| "eval_recall_I-DISEASE": 0.9231438434982738, | |
| "eval_recall_I-PROCEDURE": 0.8970941711734913, | |
| "eval_recall_I-SYMPTOM": 0.8893926676459126, | |
| "eval_recall_O": 0.9484484868479099, | |
| "eval_recall_macro": 0.8966261570560716, | |
| "eval_recall_micro": 0.926586636446837, | |
| "eval_roc_auc_B-DISEASE": 0.9430447884548903, | |
| "eval_roc_auc_B-PROCEDURE": 0.932786063234458, | |
| "eval_roc_auc_B-SYMPTOM": 0.929104371359431, | |
| "eval_roc_auc_I-DISEASE": 0.9547543027359496, | |
| "eval_roc_auc_I-PROCEDURE": 0.9433897585617623, | |
| "eval_roc_auc_I-SYMPTOM": 0.9351990344501839, | |
| "eval_roc_auc_O": 0.9424101612474091, | |
| "eval_runtime": 50.7023, | |
| "eval_samples_per_second": 279.08, | |
| "eval_steps_per_second": 34.89, | |
| "step": 241437 | |
| }, | |
| { | |
| "epoch": 7.006581427038937, | |
| "grad_norm": 2.0348429679870605, | |
| "learning_rate": 2.9962019077440494e-06, | |
| "loss": 0.021857967600226402, | |
| "step": 241664 | |
| }, | |
| { | |
| "epoch": 7.0140036531269025, | |
| "grad_norm": 1.3302487134933472, | |
| "learning_rate": 2.988779681656084e-06, | |
| "loss": 0.0204441100358963, | |
| "step": 241920 | |
| }, | |
| { | |
| "epoch": 7.021425879214868, | |
| "grad_norm": 0.7470573782920837, | |
| "learning_rate": 2.9813574555681197e-06, | |
| "loss": 0.021242015063762665, | |
| "step": 242176 | |
| }, | |
| { | |
| "epoch": 7.028848105302832, | |
| "grad_norm": 0.7607685327529907, | |
| "learning_rate": 2.9739352294801544e-06, | |
| "loss": 0.020995140075683594, | |
| "step": 242432 | |
| }, | |
| { | |
| "epoch": 7.0362703313907975, | |
| "grad_norm": 0.6476452946662903, | |
| "learning_rate": 2.9665419964628457e-06, | |
| "loss": 0.02166188880801201, | |
| "step": 242688 | |
| }, | |
| { | |
| "epoch": 7.043692557478763, | |
| "grad_norm": 0.7162955403327942, | |
| "learning_rate": 2.959119770374881e-06, | |
| "loss": 0.02120421826839447, | |
| "step": 242944 | |
| }, | |
| { | |
| "epoch": 7.051114783566727, | |
| "grad_norm": 0.7249234318733215, | |
| "learning_rate": 2.9516975442869155e-06, | |
| "loss": 0.02193913795053959, | |
| "step": 243200 | |
| }, | |
| { | |
| "epoch": 7.058537009654692, | |
| "grad_norm": 0.3952397406101227, | |
| "learning_rate": 2.944275318198951e-06, | |
| "loss": 0.02079729177057743, | |
| "step": 243456 | |
| }, | |
| { | |
| "epoch": 7.065959235742658, | |
| "grad_norm": 0.9519045948982239, | |
| "learning_rate": 2.9368530921109858e-06, | |
| "loss": 0.020243000239133835, | |
| "step": 243712 | |
| }, | |
| { | |
| "epoch": 7.073381461830622, | |
| "grad_norm": 0.9144233465194702, | |
| "learning_rate": 2.929430866023021e-06, | |
| "loss": 0.021480903029441833, | |
| "step": 243968 | |
| }, | |
| { | |
| "epoch": 7.080803687918587, | |
| "grad_norm": 1.0967696905136108, | |
| "learning_rate": 2.9220086399350556e-06, | |
| "loss": 0.020933549851179123, | |
| "step": 244224 | |
| }, | |
| { | |
| "epoch": 7.088225914006553, | |
| "grad_norm": 1.8657640218734741, | |
| "learning_rate": 2.9145864138470907e-06, | |
| "loss": 0.021546700969338417, | |
| "step": 244480 | |
| }, | |
| { | |
| "epoch": 7.095648140094517, | |
| "grad_norm": 0.6910285353660583, | |
| "learning_rate": 2.907193180829782e-06, | |
| "loss": 0.02136489935219288, | |
| "step": 244736 | |
| }, | |
| { | |
| "epoch": 7.103070366182482, | |
| "grad_norm": 0.6766055822372437, | |
| "learning_rate": 2.8997709547418167e-06, | |
| "loss": 0.020617298781871796, | |
| "step": 244992 | |
| }, | |
| { | |
| "epoch": 7.110492592270448, | |
| "grad_norm": 0.7210803031921387, | |
| "learning_rate": 2.892377721724508e-06, | |
| "loss": 0.021551325917243958, | |
| "step": 245248 | |
| }, | |
| { | |
| "epoch": 7.117914818358412, | |
| "grad_norm": 2.4113802909851074, | |
| "learning_rate": 2.8849554956365435e-06, | |
| "loss": 0.020926212891936302, | |
| "step": 245504 | |
| }, | |
| { | |
| "epoch": 7.125337044446377, | |
| "grad_norm": 1.4565765857696533, | |
| "learning_rate": 2.877533269548578e-06, | |
| "loss": 0.02048218622803688, | |
| "step": 245760 | |
| }, | |
| { | |
| "epoch": 7.1327592705343426, | |
| "grad_norm": 0.8086670637130737, | |
| "learning_rate": 2.8701110434606133e-06, | |
| "loss": 0.022243766114115715, | |
| "step": 246016 | |
| }, | |
| { | |
| "epoch": 7.140181496622307, | |
| "grad_norm": 0.9381563663482666, | |
| "learning_rate": 2.862688817372648e-06, | |
| "loss": 0.02127542719244957, | |
| "step": 246272 | |
| }, | |
| { | |
| "epoch": 7.147603722710272, | |
| "grad_norm": 0.7935812473297119, | |
| "learning_rate": 2.855266591284683e-06, | |
| "loss": 0.02121017314493656, | |
| "step": 246528 | |
| }, | |
| { | |
| "epoch": 7.1550259487982375, | |
| "grad_norm": 1.0938653945922852, | |
| "learning_rate": 2.847844365196718e-06, | |
| "loss": 0.020473578944802284, | |
| "step": 246784 | |
| }, | |
| { | |
| "epoch": 7.162448174886202, | |
| "grad_norm": 1.3731411695480347, | |
| "learning_rate": 2.8404221391087534e-06, | |
| "loss": 0.021220071241259575, | |
| "step": 247040 | |
| }, | |
| { | |
| "epoch": 7.169870400974167, | |
| "grad_norm": 0.5543429851531982, | |
| "learning_rate": 2.832999913020788e-06, | |
| "loss": 0.02135831117630005, | |
| "step": 247296 | |
| }, | |
| { | |
| "epoch": 7.1772926270621324, | |
| "grad_norm": 2.9699106216430664, | |
| "learning_rate": 2.8255776869328232e-06, | |
| "loss": 0.020661460235714912, | |
| "step": 247552 | |
| }, | |
| { | |
| "epoch": 7.184714853150097, | |
| "grad_norm": 0.7410014867782593, | |
| "learning_rate": 2.8181554608448584e-06, | |
| "loss": 0.020882125943899155, | |
| "step": 247808 | |
| }, | |
| { | |
| "epoch": 7.192137079238062, | |
| "grad_norm": 0.8216644525527954, | |
| "learning_rate": 2.8107332347568935e-06, | |
| "loss": 0.02100284770131111, | |
| "step": 248064 | |
| }, | |
| { | |
| "epoch": 7.199559305326027, | |
| "grad_norm": 0.547845721244812, | |
| "learning_rate": 2.8033110086689286e-06, | |
| "loss": 0.020722176879644394, | |
| "step": 248320 | |
| }, | |
| { | |
| "epoch": 7.206981531413992, | |
| "grad_norm": 0.9745422005653381, | |
| "learning_rate": 2.7958887825809633e-06, | |
| "loss": 0.020939286798238754, | |
| "step": 248576 | |
| }, | |
| { | |
| "epoch": 7.214403757501957, | |
| "grad_norm": 0.846403181552887, | |
| "learning_rate": 2.7884665564929985e-06, | |
| "loss": 0.019545886665582657, | |
| "step": 248832 | |
| }, | |
| { | |
| "epoch": 7.221825983589922, | |
| "grad_norm": 0.6990591287612915, | |
| "learning_rate": 2.781044330405033e-06, | |
| "loss": 0.020903486758470535, | |
| "step": 249088 | |
| }, | |
| { | |
| "epoch": 7.229248209677887, | |
| "grad_norm": 1.048094630241394, | |
| "learning_rate": 2.7736221043170687e-06, | |
| "loss": 0.02116040512919426, | |
| "step": 249344 | |
| }, | |
| { | |
| "epoch": 7.236670435765852, | |
| "grad_norm": 0.5728402733802795, | |
| "learning_rate": 2.7661998782291034e-06, | |
| "loss": 0.020977187901735306, | |
| "step": 249600 | |
| }, | |
| { | |
| "epoch": 7.244092661853817, | |
| "grad_norm": 1.5667152404785156, | |
| "learning_rate": 2.7587776521411385e-06, | |
| "loss": 0.020316850394010544, | |
| "step": 249856 | |
| }, | |
| { | |
| "epoch": 7.251514887941782, | |
| "grad_norm": 1.3079615831375122, | |
| "learning_rate": 2.7513554260531733e-06, | |
| "loss": 0.02036348357796669, | |
| "step": 250112 | |
| }, | |
| { | |
| "epoch": 7.258937114029747, | |
| "grad_norm": 0.8075518012046814, | |
| "learning_rate": 2.7439331999652084e-06, | |
| "loss": 0.021776551380753517, | |
| "step": 250368 | |
| }, | |
| { | |
| "epoch": 7.266359340117712, | |
| "grad_norm": 0.9148851037025452, | |
| "learning_rate": 2.736510973877243e-06, | |
| "loss": 0.020299196243286133, | |
| "step": 250624 | |
| }, | |
| { | |
| "epoch": 7.273781566205677, | |
| "grad_norm": 2.997012138366699, | |
| "learning_rate": 2.7290887477892786e-06, | |
| "loss": 0.019802654162049294, | |
| "step": 250880 | |
| }, | |
| { | |
| "epoch": 7.281203792293642, | |
| "grad_norm": 1.3588472604751587, | |
| "learning_rate": 2.7216665217013138e-06, | |
| "loss": 0.02064085192978382, | |
| "step": 251136 | |
| }, | |
| { | |
| "epoch": 7.288626018381607, | |
| "grad_norm": 0.4068481922149658, | |
| "learning_rate": 2.7142732886840046e-06, | |
| "loss": 0.0209208894520998, | |
| "step": 251392 | |
| }, | |
| { | |
| "epoch": 7.296048244469572, | |
| "grad_norm": 0.6390463709831238, | |
| "learning_rate": 2.7068510625960397e-06, | |
| "loss": 0.02011655457317829, | |
| "step": 251648 | |
| }, | |
| { | |
| "epoch": 7.303470470557537, | |
| "grad_norm": 1.4816343784332275, | |
| "learning_rate": 2.6994288365080744e-06, | |
| "loss": 0.020773885771632195, | |
| "step": 251904 | |
| }, | |
| { | |
| "epoch": 7.310892696645502, | |
| "grad_norm": 1.3310258388519287, | |
| "learning_rate": 2.69200661042011e-06, | |
| "loss": 0.02009221352636814, | |
| "step": 252160 | |
| }, | |
| { | |
| "epoch": 7.3183149227334665, | |
| "grad_norm": 0.9011105895042419, | |
| "learning_rate": 2.684584384332145e-06, | |
| "loss": 0.02129613235592842, | |
| "step": 252416 | |
| }, | |
| { | |
| "epoch": 7.325737148821432, | |
| "grad_norm": 2.8727986812591553, | |
| "learning_rate": 2.67716215824418e-06, | |
| "loss": 0.020425565540790558, | |
| "step": 252672 | |
| }, | |
| { | |
| "epoch": 7.333159374909397, | |
| "grad_norm": 1.9416484832763672, | |
| "learning_rate": 2.669739932156215e-06, | |
| "loss": 0.020380154252052307, | |
| "step": 252928 | |
| }, | |
| { | |
| "epoch": 7.3405816009973615, | |
| "grad_norm": 0.8640945553779602, | |
| "learning_rate": 2.6623177060682496e-06, | |
| "loss": 0.020084215328097343, | |
| "step": 253184 | |
| }, | |
| { | |
| "epoch": 7.348003827085327, | |
| "grad_norm": 1.1281476020812988, | |
| "learning_rate": 2.654895479980285e-06, | |
| "loss": 0.020082183182239532, | |
| "step": 253440 | |
| }, | |
| { | |
| "epoch": 7.355426053173291, | |
| "grad_norm": 0.7797291874885559, | |
| "learning_rate": 2.6475022469629765e-06, | |
| "loss": 0.019974490627646446, | |
| "step": 253696 | |
| }, | |
| { | |
| "epoch": 7.362848279261256, | |
| "grad_norm": 1.3273131847381592, | |
| "learning_rate": 2.640080020875011e-06, | |
| "loss": 0.020340699702501297, | |
| "step": 253952 | |
| }, | |
| { | |
| "epoch": 7.370270505349222, | |
| "grad_norm": 0.7194586992263794, | |
| "learning_rate": 2.6326577947870463e-06, | |
| "loss": 0.02018049731850624, | |
| "step": 254208 | |
| }, | |
| { | |
| "epoch": 7.377692731437186, | |
| "grad_norm": 1.6127992868423462, | |
| "learning_rate": 2.625235568699081e-06, | |
| "loss": 0.020147008821368217, | |
| "step": 254464 | |
| }, | |
| { | |
| "epoch": 7.385114957525151, | |
| "grad_norm": 1.2921494245529175, | |
| "learning_rate": 2.617813342611116e-06, | |
| "loss": 0.020299099385738373, | |
| "step": 254720 | |
| }, | |
| { | |
| "epoch": 7.392537183613117, | |
| "grad_norm": 0.7169179916381836, | |
| "learning_rate": 2.610391116523151e-06, | |
| "loss": 0.02025839127600193, | |
| "step": 254976 | |
| }, | |
| { | |
| "epoch": 7.399959409701081, | |
| "grad_norm": 0.6367201805114746, | |
| "learning_rate": 2.6029688904351864e-06, | |
| "loss": 0.020546825602650642, | |
| "step": 255232 | |
| }, | |
| { | |
| "epoch": 7.407381635789046, | |
| "grad_norm": 1.7125663757324219, | |
| "learning_rate": 2.595546664347221e-06, | |
| "loss": 0.020340323448181152, | |
| "step": 255488 | |
| }, | |
| { | |
| "epoch": 7.414803861877012, | |
| "grad_norm": 0.6656335592269897, | |
| "learning_rate": 2.5881534313299123e-06, | |
| "loss": 0.019990479573607445, | |
| "step": 255744 | |
| }, | |
| { | |
| "epoch": 7.422226087964976, | |
| "grad_norm": 0.7568262219429016, | |
| "learning_rate": 2.5807312052419475e-06, | |
| "loss": 0.020358338952064514, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 7.429648314052941, | |
| "grad_norm": 0.9109243750572205, | |
| "learning_rate": 2.573308979153982e-06, | |
| "loss": 0.0201952513307333, | |
| "step": 256256 | |
| }, | |
| { | |
| "epoch": 7.4370705401409065, | |
| "grad_norm": 0.947914183139801, | |
| "learning_rate": 2.5658867530660177e-06, | |
| "loss": 0.02002812549471855, | |
| "step": 256512 | |
| }, | |
| { | |
| "epoch": 7.444492766228871, | |
| "grad_norm": 0.9793144464492798, | |
| "learning_rate": 2.5584645269780524e-06, | |
| "loss": 0.01997041516005993, | |
| "step": 256768 | |
| }, | |
| { | |
| "epoch": 7.451914992316836, | |
| "grad_norm": 1.3479286432266235, | |
| "learning_rate": 2.5510712939607437e-06, | |
| "loss": 0.01979486271739006, | |
| "step": 257024 | |
| }, | |
| { | |
| "epoch": 7.4593372184048015, | |
| "grad_norm": 1.0751733779907227, | |
| "learning_rate": 2.543649067872779e-06, | |
| "loss": 0.0201573446393013, | |
| "step": 257280 | |
| }, | |
| { | |
| "epoch": 7.466759444492766, | |
| "grad_norm": 0.8343433737754822, | |
| "learning_rate": 2.5362268417848135e-06, | |
| "loss": 0.019964421167969704, | |
| "step": 257536 | |
| }, | |
| { | |
| "epoch": 7.474181670580731, | |
| "grad_norm": 0.7085336446762085, | |
| "learning_rate": 2.5288046156968487e-06, | |
| "loss": 0.01973593793809414, | |
| "step": 257792 | |
| }, | |
| { | |
| "epoch": 7.481603896668696, | |
| "grad_norm": 1.0218849182128906, | |
| "learning_rate": 2.5213823896088834e-06, | |
| "loss": 0.019713174551725388, | |
| "step": 258048 | |
| }, | |
| { | |
| "epoch": 7.489026122756661, | |
| "grad_norm": 1.6293119192123413, | |
| "learning_rate": 2.513960163520919e-06, | |
| "loss": 0.020315375179052353, | |
| "step": 258304 | |
| }, | |
| { | |
| "epoch": 7.496448348844626, | |
| "grad_norm": 1.086714267730713, | |
| "learning_rate": 2.5065379374329536e-06, | |
| "loss": 0.020414341241121292, | |
| "step": 258560 | |
| }, | |
| { | |
| "epoch": 7.503870574932591, | |
| "grad_norm": 0.5886796116828918, | |
| "learning_rate": 2.4991157113449887e-06, | |
| "loss": 0.02043706364929676, | |
| "step": 258816 | |
| }, | |
| { | |
| "epoch": 7.511292801020556, | |
| "grad_norm": 0.7196995615959167, | |
| "learning_rate": 2.491693485257024e-06, | |
| "loss": 0.019802773371338844, | |
| "step": 259072 | |
| }, | |
| { | |
| "epoch": 7.518715027108521, | |
| "grad_norm": 1.3985934257507324, | |
| "learning_rate": 2.4842712591690586e-06, | |
| "loss": 0.018815403804183006, | |
| "step": 259328 | |
| }, | |
| { | |
| "epoch": 7.526137253196486, | |
| "grad_norm": 0.5434567928314209, | |
| "learning_rate": 2.4768490330810937e-06, | |
| "loss": 0.020260611549019814, | |
| "step": 259584 | |
| }, | |
| { | |
| "epoch": 7.533559479284451, | |
| "grad_norm": 2.1898112297058105, | |
| "learning_rate": 2.469426806993129e-06, | |
| "loss": 0.0197544377297163, | |
| "step": 259840 | |
| }, | |
| { | |
| "epoch": 7.540981705372416, | |
| "grad_norm": 1.802711009979248, | |
| "learning_rate": 2.462004580905164e-06, | |
| "loss": 0.020619960501790047, | |
| "step": 260096 | |
| }, | |
| { | |
| "epoch": 7.548403931460381, | |
| "grad_norm": 2.526963233947754, | |
| "learning_rate": 2.4545823548171987e-06, | |
| "loss": 0.019032476469874382, | |
| "step": 260352 | |
| }, | |
| { | |
| "epoch": 7.555826157548346, | |
| "grad_norm": 1.4277547597885132, | |
| "learning_rate": 2.4471601287292342e-06, | |
| "loss": 0.01940942369401455, | |
| "step": 260608 | |
| }, | |
| { | |
| "epoch": 7.563248383636311, | |
| "grad_norm": 1.0160654783248901, | |
| "learning_rate": 2.439737902641269e-06, | |
| "loss": 0.01931748166680336, | |
| "step": 260864 | |
| }, | |
| { | |
| "epoch": 7.570670609724276, | |
| "grad_norm": 0.5958607196807861, | |
| "learning_rate": 2.43234466962396e-06, | |
| "loss": 0.01939370110630989, | |
| "step": 261120 | |
| }, | |
| { | |
| "epoch": 7.578092835812241, | |
| "grad_norm": 0.6600317358970642, | |
| "learning_rate": 2.424922443535995e-06, | |
| "loss": 0.020635899156332016, | |
| "step": 261376 | |
| }, | |
| { | |
| "epoch": 7.585515061900206, | |
| "grad_norm": 0.6846224665641785, | |
| "learning_rate": 2.41750021744803e-06, | |
| "loss": 0.019544269889593124, | |
| "step": 261632 | |
| }, | |
| { | |
| "epoch": 7.592937287988171, | |
| "grad_norm": 0.7630472779273987, | |
| "learning_rate": 2.410077991360065e-06, | |
| "loss": 0.02025151625275612, | |
| "step": 261888 | |
| }, | |
| { | |
| "epoch": 7.6003595140761355, | |
| "grad_norm": 0.5054865479469299, | |
| "learning_rate": 2.4026557652721003e-06, | |
| "loss": 0.020576078444719315, | |
| "step": 262144 | |
| }, | |
| { | |
| "epoch": 7.607781740164101, | |
| "grad_norm": 1.665007472038269, | |
| "learning_rate": 2.3952335391841354e-06, | |
| "loss": 0.019354388117790222, | |
| "step": 262400 | |
| }, | |
| { | |
| "epoch": 7.615203966252066, | |
| "grad_norm": 0.896648108959198, | |
| "learning_rate": 2.38781131309617e-06, | |
| "loss": 0.019793247804045677, | |
| "step": 262656 | |
| }, | |
| { | |
| "epoch": 7.6226261923400305, | |
| "grad_norm": 1.2570286989212036, | |
| "learning_rate": 2.3803890870082052e-06, | |
| "loss": 0.019577788189053535, | |
| "step": 262912 | |
| }, | |
| { | |
| "epoch": 7.630048418427996, | |
| "grad_norm": 1.4413504600524902, | |
| "learning_rate": 2.3730248470615526e-06, | |
| "loss": 0.01890435814857483, | |
| "step": 263168 | |
| }, | |
| { | |
| "epoch": 7.637470644515961, | |
| "grad_norm": 0.7741318941116333, | |
| "learning_rate": 2.3656026209735873e-06, | |
| "loss": 0.019931374117732048, | |
| "step": 263424 | |
| }, | |
| { | |
| "epoch": 7.644892870603925, | |
| "grad_norm": 1.1268854141235352, | |
| "learning_rate": 2.3581803948856225e-06, | |
| "loss": 0.01981361024081707, | |
| "step": 263680 | |
| }, | |
| { | |
| "epoch": 7.652315096691891, | |
| "grad_norm": 0.4975253939628601, | |
| "learning_rate": 2.3507581687976576e-06, | |
| "loss": 0.01936083845794201, | |
| "step": 263936 | |
| }, | |
| { | |
| "epoch": 7.659737322779856, | |
| "grad_norm": 0.787571907043457, | |
| "learning_rate": 2.3433359427096923e-06, | |
| "loss": 0.01990235038101673, | |
| "step": 264192 | |
| }, | |
| { | |
| "epoch": 7.66715954886782, | |
| "grad_norm": 1.4735548496246338, | |
| "learning_rate": 2.335913716621728e-06, | |
| "loss": 0.018744993954896927, | |
| "step": 264448 | |
| }, | |
| { | |
| "epoch": 7.674581774955786, | |
| "grad_norm": 0.5953457355499268, | |
| "learning_rate": 2.3284914905337625e-06, | |
| "loss": 0.019753796979784966, | |
| "step": 264704 | |
| }, | |
| { | |
| "epoch": 7.682004001043751, | |
| "grad_norm": 0.9420062303543091, | |
| "learning_rate": 2.3210692644457977e-06, | |
| "loss": 0.01961057260632515, | |
| "step": 264960 | |
| }, | |
| { | |
| "epoch": 7.689426227131715, | |
| "grad_norm": 0.7770072221755981, | |
| "learning_rate": 2.313647038357833e-06, | |
| "loss": 0.019495196640491486, | |
| "step": 265216 | |
| }, | |
| { | |
| "epoch": 7.696848453219681, | |
| "grad_norm": 0.618382453918457, | |
| "learning_rate": 2.306224812269868e-06, | |
| "loss": 0.0193632785230875, | |
| "step": 265472 | |
| }, | |
| { | |
| "epoch": 7.704270679307646, | |
| "grad_norm": 0.8640962243080139, | |
| "learning_rate": 2.2988025861819026e-06, | |
| "loss": 0.01896437630057335, | |
| "step": 265728 | |
| }, | |
| { | |
| "epoch": 7.71169290539561, | |
| "grad_norm": 1.004714012145996, | |
| "learning_rate": 2.2913803600939378e-06, | |
| "loss": 0.019315877929329872, | |
| "step": 265984 | |
| }, | |
| { | |
| "epoch": 7.7191151314835755, | |
| "grad_norm": 0.7368209958076477, | |
| "learning_rate": 2.283958134005973e-06, | |
| "loss": 0.019432280212640762, | |
| "step": 266240 | |
| }, | |
| { | |
| "epoch": 7.726537357571541, | |
| "grad_norm": 0.7293087244033813, | |
| "learning_rate": 2.2765359079180076e-06, | |
| "loss": 0.019598202779889107, | |
| "step": 266496 | |
| }, | |
| { | |
| "epoch": 7.733959583659505, | |
| "grad_norm": 0.8341611623764038, | |
| "learning_rate": 2.2691136818300427e-06, | |
| "loss": 0.018693827092647552, | |
| "step": 266752 | |
| }, | |
| { | |
| "epoch": 7.7413818097474705, | |
| "grad_norm": 1.1728421449661255, | |
| "learning_rate": 2.261691455742078e-06, | |
| "loss": 0.019444908946752548, | |
| "step": 267008 | |
| }, | |
| { | |
| "epoch": 7.748804035835436, | |
| "grad_norm": 0.5628138780593872, | |
| "learning_rate": 2.254298222724769e-06, | |
| "loss": 0.019465288147330284, | |
| "step": 267264 | |
| }, | |
| { | |
| "epoch": 7.7562262619234, | |
| "grad_norm": 3.7096822261810303, | |
| "learning_rate": 2.2469049897074604e-06, | |
| "loss": 0.019718138501048088, | |
| "step": 267520 | |
| }, | |
| { | |
| "epoch": 7.763648488011365, | |
| "grad_norm": 1.058203101158142, | |
| "learning_rate": 2.239482763619495e-06, | |
| "loss": 0.019666267558932304, | |
| "step": 267776 | |
| }, | |
| { | |
| "epoch": 7.771070714099331, | |
| "grad_norm": 2.215019702911377, | |
| "learning_rate": 2.23206053753153e-06, | |
| "loss": 0.018484462052583694, | |
| "step": 268032 | |
| }, | |
| { | |
| "epoch": 7.778492940187295, | |
| "grad_norm": 1.5556467771530151, | |
| "learning_rate": 2.2246383114435653e-06, | |
| "loss": 0.01945563592016697, | |
| "step": 268288 | |
| }, | |
| { | |
| "epoch": 7.78591516627526, | |
| "grad_norm": 2.0950751304626465, | |
| "learning_rate": 2.2172160853556e-06, | |
| "loss": 0.018351901322603226, | |
| "step": 268544 | |
| }, | |
| { | |
| "epoch": 7.793337392363226, | |
| "grad_norm": 0.7168514728546143, | |
| "learning_rate": 2.209793859267635e-06, | |
| "loss": 0.018952462822198868, | |
| "step": 268800 | |
| }, | |
| { | |
| "epoch": 7.80075961845119, | |
| "grad_norm": 1.6901471614837646, | |
| "learning_rate": 2.2023716331796703e-06, | |
| "loss": 0.019636450335383415, | |
| "step": 269056 | |
| }, | |
| { | |
| "epoch": 7.808181844539155, | |
| "grad_norm": 0.29729220271110535, | |
| "learning_rate": 2.194949407091705e-06, | |
| "loss": 0.018051423132419586, | |
| "step": 269312 | |
| }, | |
| { | |
| "epoch": 7.815604070627121, | |
| "grad_norm": 0.7447888255119324, | |
| "learning_rate": 2.18752718100374e-06, | |
| "loss": 0.019316522404551506, | |
| "step": 269568 | |
| }, | |
| { | |
| "epoch": 7.823026296715085, | |
| "grad_norm": 1.053483247756958, | |
| "learning_rate": 2.1801049549157752e-06, | |
| "loss": 0.018997350707650185, | |
| "step": 269824 | |
| }, | |
| { | |
| "epoch": 7.83044852280305, | |
| "grad_norm": 3.5569405555725098, | |
| "learning_rate": 2.1726827288278104e-06, | |
| "loss": 0.018772602081298828, | |
| "step": 270080 | |
| }, | |
| { | |
| "epoch": 7.8378707488910155, | |
| "grad_norm": 0.7023041248321533, | |
| "learning_rate": 2.1652605027398455e-06, | |
| "loss": 0.018561918288469315, | |
| "step": 270336 | |
| }, | |
| { | |
| "epoch": 7.84529297497898, | |
| "grad_norm": 0.6641720533370972, | |
| "learning_rate": 2.1578672697225363e-06, | |
| "loss": 0.018886465579271317, | |
| "step": 270592 | |
| }, | |
| { | |
| "epoch": 7.852715201066945, | |
| "grad_norm": 1.0777596235275269, | |
| "learning_rate": 2.1504450436345715e-06, | |
| "loss": 0.018519580364227295, | |
| "step": 270848 | |
| }, | |
| { | |
| "epoch": 7.86013742715491, | |
| "grad_norm": 1.0486457347869873, | |
| "learning_rate": 2.1430228175466066e-06, | |
| "loss": 0.018756285309791565, | |
| "step": 271104 | |
| }, | |
| { | |
| "epoch": 7.867559653242875, | |
| "grad_norm": 1.1527291536331177, | |
| "learning_rate": 2.1356005914586417e-06, | |
| "loss": 0.01867642067372799, | |
| "step": 271360 | |
| }, | |
| { | |
| "epoch": 7.87498187933084, | |
| "grad_norm": 0.513008177280426, | |
| "learning_rate": 2.128178365370677e-06, | |
| "loss": 0.018500298261642456, | |
| "step": 271616 | |
| }, | |
| { | |
| "epoch": 7.8824041054188045, | |
| "grad_norm": 0.49376487731933594, | |
| "learning_rate": 2.1207561392827116e-06, | |
| "loss": 0.01862124353647232, | |
| "step": 271872 | |
| }, | |
| { | |
| "epoch": 7.88982633150677, | |
| "grad_norm": 1.1185438632965088, | |
| "learning_rate": 2.113362906265403e-06, | |
| "loss": 0.019213099032640457, | |
| "step": 272128 | |
| }, | |
| { | |
| "epoch": 7.897248557594735, | |
| "grad_norm": 0.6070836782455444, | |
| "learning_rate": 2.1059406801774375e-06, | |
| "loss": 0.018839983269572258, | |
| "step": 272384 | |
| }, | |
| { | |
| "epoch": 7.9046707836826995, | |
| "grad_norm": 0.6914726495742798, | |
| "learning_rate": 2.098518454089473e-06, | |
| "loss": 0.01875445805490017, | |
| "step": 272640 | |
| }, | |
| { | |
| "epoch": 7.912093009770665, | |
| "grad_norm": 1.3377681970596313, | |
| "learning_rate": 2.0910962280015078e-06, | |
| "loss": 0.01854925975203514, | |
| "step": 272896 | |
| }, | |
| { | |
| "epoch": 7.91951523585863, | |
| "grad_norm": 0.8639470934867859, | |
| "learning_rate": 2.083674001913543e-06, | |
| "loss": 0.01848948746919632, | |
| "step": 273152 | |
| }, | |
| { | |
| "epoch": 7.926937461946594, | |
| "grad_norm": 0.5949268937110901, | |
| "learning_rate": 2.076251775825578e-06, | |
| "loss": 0.018381042405962944, | |
| "step": 273408 | |
| }, | |
| { | |
| "epoch": 7.93435968803456, | |
| "grad_norm": 1.5548309087753296, | |
| "learning_rate": 2.0688295497376127e-06, | |
| "loss": 0.01857753098011017, | |
| "step": 273664 | |
| }, | |
| { | |
| "epoch": 7.941781914122525, | |
| "grad_norm": 0.9342926740646362, | |
| "learning_rate": 2.061407323649648e-06, | |
| "loss": 0.018817652016878128, | |
| "step": 273920 | |
| }, | |
| { | |
| "epoch": 7.949204140210489, | |
| "grad_norm": 0.729648768901825, | |
| "learning_rate": 2.053985097561683e-06, | |
| "loss": 0.018660228699445724, | |
| "step": 274176 | |
| }, | |
| { | |
| "epoch": 7.956626366298455, | |
| "grad_norm": 2.3486342430114746, | |
| "learning_rate": 2.0465628714737177e-06, | |
| "loss": 0.018646761775016785, | |
| "step": 274432 | |
| }, | |
| { | |
| "epoch": 7.96404859238642, | |
| "grad_norm": 1.1601566076278687, | |
| "learning_rate": 2.039140645385753e-06, | |
| "loss": 0.018127109855413437, | |
| "step": 274688 | |
| }, | |
| { | |
| "epoch": 7.971470818474384, | |
| "grad_norm": 0.9393084049224854, | |
| "learning_rate": 2.031718419297788e-06, | |
| "loss": 0.018819065764546394, | |
| "step": 274944 | |
| }, | |
| { | |
| "epoch": 7.97889304456235, | |
| "grad_norm": 0.7464454770088196, | |
| "learning_rate": 2.024296193209823e-06, | |
| "loss": 0.017582645639777184, | |
| "step": 275200 | |
| }, | |
| { | |
| "epoch": 7.986315270650315, | |
| "grad_norm": 1.2547037601470947, | |
| "learning_rate": 2.0169029601925143e-06, | |
| "loss": 0.020482897758483887, | |
| "step": 275456 | |
| }, | |
| { | |
| "epoch": 7.993737496738279, | |
| "grad_norm": 0.5836815237998962, | |
| "learning_rate": 2.009480734104549e-06, | |
| "loss": 0.018345054239034653, | |
| "step": 275712 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_f1_B-DISEASE": 0.8881473110416805, | |
| "eval_f1_B-PROCEDURE": 0.8697541331072488, | |
| "eval_f1_B-SYMPTOM": 0.8758161342623023, | |
| "eval_f1_I-DISEASE": 0.9152045845272206, | |
| "eval_f1_I-PROCEDURE": 0.9065541208843036, | |
| "eval_f1_I-SYMPTOM": 0.8956227815606521, | |
| "eval_f1_O": 0.9526831207591596, | |
| "eval_f1_macro": 0.9005403123060811, | |
| "eval_f1_micro": 0.9303222840720738, | |
| "eval_loss": 0.08025326579809189, | |
| "eval_precision_B-DISEASE": 0.8888681906485351, | |
| "eval_precision_B-PROCEDURE": 0.8693625698471968, | |
| "eval_precision_B-SYMPTOM": 0.8819999117426416, | |
| "eval_precision_I-DISEASE": 0.9115479452054794, | |
| "eval_precision_I-PROCEDURE": 0.9107735629382937, | |
| "eval_precision_I-SYMPTOM": 0.8996899459295951, | |
| "eval_precision_O": 0.9539307692307692, | |
| "eval_precision_macro": 0.90231041364893, | |
| "eval_precision_micro": 0.9317878712485543, | |
| "eval_rauc_macro": 0.9412876564522803, | |
| "eval_rauc_micro": 0.9585606455844751, | |
| "eval_recall_B-DISEASE": 0.8874275997661938, | |
| "eval_recall_B-PROCEDURE": 0.8701460492485488, | |
| "eval_recall_B-SYMPTOM": 0.869718463078195, | |
| "eval_recall_I-DISEASE": 0.9188906789413118, | |
| "eval_recall_I-PROCEDURE": 0.9023735942999399, | |
| "eval_recall_I-SYMPTOM": 0.8915922239875296, | |
| "eval_recall_O": 0.9514387316298465, | |
| "eval_recall_macro": 0.8987981915645094, | |
| "eval_recall_micro": 0.9288613000304706, | |
| "eval_roc_auc_B-DISEASE": 0.9425154534634151, | |
| "eval_roc_auc_B-PROCEDURE": 0.9336573769102026, | |
| "eval_roc_auc_B-SYMPTOM": 0.9333172008218918, | |
| "eval_roc_auc_I-DISEASE": 0.9532477657738869, | |
| "eval_roc_auc_I-PROCEDURE": 0.9460172341015058, | |
| "eval_roc_auc_I-SYMPTOM": 0.9370300603947014, | |
| "eval_roc_auc_O": 0.9432285037003589, | |
| "eval_runtime": 50.7497, | |
| "eval_samples_per_second": 278.819, | |
| "eval_steps_per_second": 34.857, | |
| "step": 275928 | |
| }, | |
| { | |
| "epoch": 8.001159722826245, | |
| "grad_norm": 0.9876375794410706, | |
| "learning_rate": 2.002058508016584e-06, | |
| "loss": 0.017906054854393005, | |
| "step": 275968 | |
| }, | |
| { | |
| "epoch": 8.008581948914209, | |
| "grad_norm": 0.6705123782157898, | |
| "learning_rate": 1.9946362819286193e-06, | |
| "loss": 0.016327261924743652, | |
| "step": 276224 | |
| }, | |
| { | |
| "epoch": 8.016004175002175, | |
| "grad_norm": 0.3969533145427704, | |
| "learning_rate": 1.987214055840654e-06, | |
| "loss": 0.015958691015839577, | |
| "step": 276480 | |
| }, | |
| { | |
| "epoch": 8.02342640109014, | |
| "grad_norm": 2.4101226329803467, | |
| "learning_rate": 1.979791829752689e-06, | |
| "loss": 0.016231857240200043, | |
| "step": 276736 | |
| }, | |
| { | |
| "epoch": 8.030848627178104, | |
| "grad_norm": 0.7168958783149719, | |
| "learning_rate": 1.9723696036647243e-06, | |
| "loss": 0.015635322779417038, | |
| "step": 276992 | |
| }, | |
| { | |
| "epoch": 8.03827085326607, | |
| "grad_norm": 0.8767186999320984, | |
| "learning_rate": 1.9649473775767594e-06, | |
| "loss": 0.016447046771645546, | |
| "step": 277248 | |
| }, | |
| { | |
| "epoch": 8.045693079354034, | |
| "grad_norm": 0.36944112181663513, | |
| "learning_rate": 1.9575251514887945e-06, | |
| "loss": 0.016049547120928764, | |
| "step": 277504 | |
| }, | |
| { | |
| "epoch": 8.053115305441999, | |
| "grad_norm": 1.8824539184570312, | |
| "learning_rate": 1.9501029254008292e-06, | |
| "loss": 0.016436420381069183, | |
| "step": 277760 | |
| }, | |
| { | |
| "epoch": 8.060537531529965, | |
| "grad_norm": 0.5750571489334106, | |
| "learning_rate": 1.9426806993128643e-06, | |
| "loss": 0.01575016975402832, | |
| "step": 278016 | |
| }, | |
| { | |
| "epoch": 8.06795975761793, | |
| "grad_norm": 5.41447114944458, | |
| "learning_rate": 1.9352584732248995e-06, | |
| "loss": 0.01653452403843403, | |
| "step": 278272 | |
| }, | |
| { | |
| "epoch": 8.075381983705894, | |
| "grad_norm": 1.5512313842773438, | |
| "learning_rate": 1.9278362471369346e-06, | |
| "loss": 0.016168829053640366, | |
| "step": 278528 | |
| }, | |
| { | |
| "epoch": 8.08280420979386, | |
| "grad_norm": 0.5696364045143127, | |
| "learning_rate": 1.9204140210489693e-06, | |
| "loss": 0.01604221574962139, | |
| "step": 278784 | |
| }, | |
| { | |
| "epoch": 8.090226435881824, | |
| "grad_norm": 0.7741561532020569, | |
| "learning_rate": 1.9129917949610044e-06, | |
| "loss": 0.015817606821656227, | |
| "step": 279040 | |
| }, | |
| { | |
| "epoch": 8.097648661969789, | |
| "grad_norm": 0.5559753179550171, | |
| "learning_rate": 1.9055695688730394e-06, | |
| "loss": 0.01641227677464485, | |
| "step": 279296 | |
| }, | |
| { | |
| "epoch": 8.105070888057755, | |
| "grad_norm": 0.8511784672737122, | |
| "learning_rate": 1.8981473427850745e-06, | |
| "loss": 0.015824679285287857, | |
| "step": 279552 | |
| }, | |
| { | |
| "epoch": 8.11249311414572, | |
| "grad_norm": 0.5478092432022095, | |
| "learning_rate": 1.8907251166971096e-06, | |
| "loss": 0.016275722533464432, | |
| "step": 279808 | |
| }, | |
| { | |
| "epoch": 8.119915340233684, | |
| "grad_norm": 0.6021304726600647, | |
| "learning_rate": 1.8833028906091447e-06, | |
| "loss": 0.016562415286898613, | |
| "step": 280064 | |
| }, | |
| { | |
| "epoch": 8.12733756632165, | |
| "grad_norm": 0.6858309507369995, | |
| "learning_rate": 1.8758806645211797e-06, | |
| "loss": 0.016449617221951485, | |
| "step": 280320 | |
| }, | |
| { | |
| "epoch": 8.134759792409614, | |
| "grad_norm": 1.4423892498016357, | |
| "learning_rate": 1.8684874315038707e-06, | |
| "loss": 0.016584884375333786, | |
| "step": 280576 | |
| }, | |
| { | |
| "epoch": 8.142182018497579, | |
| "grad_norm": 1.4599642753601074, | |
| "learning_rate": 1.8610652054159056e-06, | |
| "loss": 0.01652132160961628, | |
| "step": 280832 | |
| }, | |
| { | |
| "epoch": 8.149604244585545, | |
| "grad_norm": 0.7201634049415588, | |
| "learning_rate": 1.853642979327941e-06, | |
| "loss": 0.01638868823647499, | |
| "step": 281088 | |
| }, | |
| { | |
| "epoch": 8.15702647067351, | |
| "grad_norm": 1.5236265659332275, | |
| "learning_rate": 1.8462207532399759e-06, | |
| "loss": 0.016754554584622383, | |
| "step": 281344 | |
| }, | |
| { | |
| "epoch": 8.164448696761474, | |
| "grad_norm": 0.6590694785118103, | |
| "learning_rate": 1.838798527152011e-06, | |
| "loss": 0.01605048030614853, | |
| "step": 281600 | |
| }, | |
| { | |
| "epoch": 8.17187092284944, | |
| "grad_norm": 5.759639263153076, | |
| "learning_rate": 1.831376301064046e-06, | |
| "loss": 0.01651783101260662, | |
| "step": 281856 | |
| }, | |
| { | |
| "epoch": 8.179293148937404, | |
| "grad_norm": 0.9102259874343872, | |
| "learning_rate": 1.8239540749760808e-06, | |
| "loss": 0.016150884330272675, | |
| "step": 282112 | |
| }, | |
| { | |
| "epoch": 8.186715375025369, | |
| "grad_norm": 0.964730978012085, | |
| "learning_rate": 1.816531848888116e-06, | |
| "loss": 0.01596345379948616, | |
| "step": 282368 | |
| }, | |
| { | |
| "epoch": 8.194137601113335, | |
| "grad_norm": 0.5655167698860168, | |
| "learning_rate": 1.8091096228001509e-06, | |
| "loss": 0.015554811805486679, | |
| "step": 282624 | |
| }, | |
| { | |
| "epoch": 8.201559827201299, | |
| "grad_norm": 0.8061972260475159, | |
| "learning_rate": 1.801687396712186e-06, | |
| "loss": 0.016146361827850342, | |
| "step": 282880 | |
| }, | |
| { | |
| "epoch": 8.208982053289263, | |
| "grad_norm": 0.8786393404006958, | |
| "learning_rate": 1.794265170624221e-06, | |
| "loss": 0.01596941240131855, | |
| "step": 283136 | |
| }, | |
| { | |
| "epoch": 8.21640427937723, | |
| "grad_norm": 0.7836214900016785, | |
| "learning_rate": 1.7868429445362558e-06, | |
| "loss": 0.01632365956902504, | |
| "step": 283392 | |
| }, | |
| { | |
| "epoch": 8.223826505465194, | |
| "grad_norm": 0.6920164227485657, | |
| "learning_rate": 1.779420718448291e-06, | |
| "loss": 0.016017304733395576, | |
| "step": 283648 | |
| }, | |
| { | |
| "epoch": 8.231248731553158, | |
| "grad_norm": 1.0860404968261719, | |
| "learning_rate": 1.7719984923603259e-06, | |
| "loss": 0.016032757237553596, | |
| "step": 283904 | |
| }, | |
| { | |
| "epoch": 8.238670957641125, | |
| "grad_norm": 0.4290538728237152, | |
| "learning_rate": 1.7645762662723612e-06, | |
| "loss": 0.016645820811390877, | |
| "step": 284160 | |
| }, | |
| { | |
| "epoch": 8.246093183729089, | |
| "grad_norm": 0.3660452365875244, | |
| "learning_rate": 1.7571540401843961e-06, | |
| "loss": 0.01602095365524292, | |
| "step": 284416 | |
| }, | |
| { | |
| "epoch": 8.253515409817053, | |
| "grad_norm": 0.6826481819152832, | |
| "learning_rate": 1.7497608071670872e-06, | |
| "loss": 0.016106512397527695, | |
| "step": 284672 | |
| }, | |
| { | |
| "epoch": 8.26093763590502, | |
| "grad_norm": 0.8434770107269287, | |
| "learning_rate": 1.742338581079122e-06, | |
| "loss": 0.016240239143371582, | |
| "step": 284928 | |
| }, | |
| { | |
| "epoch": 8.268359861992984, | |
| "grad_norm": 1.2117093801498413, | |
| "learning_rate": 1.7349163549911572e-06, | |
| "loss": 0.01664714701473713, | |
| "step": 285184 | |
| }, | |
| { | |
| "epoch": 8.275782088080948, | |
| "grad_norm": 0.48151081800460815, | |
| "learning_rate": 1.7274941289031924e-06, | |
| "loss": 0.015604979358613491, | |
| "step": 285440 | |
| }, | |
| { | |
| "epoch": 8.283204314168914, | |
| "grad_norm": 0.7376235127449036, | |
| "learning_rate": 1.7200719028152275e-06, | |
| "loss": 0.01568613015115261, | |
| "step": 285696 | |
| }, | |
| { | |
| "epoch": 8.290626540256879, | |
| "grad_norm": 1.147287368774414, | |
| "learning_rate": 1.7126496767272624e-06, | |
| "loss": 0.015849363058805466, | |
| "step": 285952 | |
| }, | |
| { | |
| "epoch": 8.298048766344843, | |
| "grad_norm": 1.2356882095336914, | |
| "learning_rate": 1.7052274506392973e-06, | |
| "loss": 0.015893224626779556, | |
| "step": 286208 | |
| }, | |
| { | |
| "epoch": 8.30547099243281, | |
| "grad_norm": 1.2852275371551514, | |
| "learning_rate": 1.6978052245513324e-06, | |
| "loss": 0.016255663707852364, | |
| "step": 286464 | |
| }, | |
| { | |
| "epoch": 8.312893218520774, | |
| "grad_norm": 0.48054972290992737, | |
| "learning_rate": 1.6903829984633674e-06, | |
| "loss": 0.016588209196925163, | |
| "step": 286720 | |
| }, | |
| { | |
| "epoch": 8.320315444608738, | |
| "grad_norm": 1.5743788480758667, | |
| "learning_rate": 1.6829897654460586e-06, | |
| "loss": 0.015848536044359207, | |
| "step": 286976 | |
| }, | |
| { | |
| "epoch": 8.327737670696704, | |
| "grad_norm": 1.1871750354766846, | |
| "learning_rate": 1.6755675393580935e-06, | |
| "loss": 0.01542384922504425, | |
| "step": 287232 | |
| }, | |
| { | |
| "epoch": 8.335159896784669, | |
| "grad_norm": 0.7835791110992432, | |
| "learning_rate": 1.6681453132701287e-06, | |
| "loss": 0.01607108861207962, | |
| "step": 287488 | |
| }, | |
| { | |
| "epoch": 8.342582122872633, | |
| "grad_norm": 1.7914983034133911, | |
| "learning_rate": 1.6607230871821636e-06, | |
| "loss": 0.015760155394673347, | |
| "step": 287744 | |
| }, | |
| { | |
| "epoch": 8.3500043489606, | |
| "grad_norm": 1.527265191078186, | |
| "learning_rate": 1.6533008610941987e-06, | |
| "loss": 0.01624034158885479, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 8.357426575048564, | |
| "grad_norm": 0.6113833785057068, | |
| "learning_rate": 1.6458786350062336e-06, | |
| "loss": 0.01606636680662632, | |
| "step": 288256 | |
| }, | |
| { | |
| "epoch": 8.364848801136528, | |
| "grad_norm": 0.5767993927001953, | |
| "learning_rate": 1.6384564089182685e-06, | |
| "loss": 0.015884777531027794, | |
| "step": 288512 | |
| }, | |
| { | |
| "epoch": 8.372271027224492, | |
| "grad_norm": 0.702173113822937, | |
| "learning_rate": 1.6310341828303037e-06, | |
| "loss": 0.015483900904655457, | |
| "step": 288768 | |
| }, | |
| { | |
| "epoch": 8.379693253312459, | |
| "grad_norm": 1.584822416305542, | |
| "learning_rate": 1.6236119567423386e-06, | |
| "loss": 0.015037352219223976, | |
| "step": 289024 | |
| }, | |
| { | |
| "epoch": 8.387115479400423, | |
| "grad_norm": 0.518094539642334, | |
| "learning_rate": 1.6161897306543737e-06, | |
| "loss": 0.016294999048113823, | |
| "step": 289280 | |
| }, | |
| { | |
| "epoch": 8.39453770548839, | |
| "grad_norm": 0.694807767868042, | |
| "learning_rate": 1.6087964976370648e-06, | |
| "loss": 0.015848014503717422, | |
| "step": 289536 | |
| }, | |
| { | |
| "epoch": 8.401959931576354, | |
| "grad_norm": 0.7278647422790527, | |
| "learning_rate": 1.6013742715490999e-06, | |
| "loss": 0.016026966273784637, | |
| "step": 289792 | |
| }, | |
| { | |
| "epoch": 8.409382157664318, | |
| "grad_norm": 0.5409690737724304, | |
| "learning_rate": 1.5939520454611348e-06, | |
| "loss": 0.01611691154539585, | |
| "step": 290048 | |
| }, | |
| { | |
| "epoch": 8.416804383752282, | |
| "grad_norm": 0.9079444408416748, | |
| "learning_rate": 1.58652981937317e-06, | |
| "loss": 0.01547956932336092, | |
| "step": 290304 | |
| }, | |
| { | |
| "epoch": 8.424226609840249, | |
| "grad_norm": 0.8612942099571228, | |
| "learning_rate": 1.5791075932852048e-06, | |
| "loss": 0.014944296330213547, | |
| "step": 290560 | |
| }, | |
| { | |
| "epoch": 8.431648835928213, | |
| "grad_norm": 1.4216639995574951, | |
| "learning_rate": 1.5717143602678961e-06, | |
| "loss": 0.016151465475559235, | |
| "step": 290816 | |
| }, | |
| { | |
| "epoch": 8.439071062016177, | |
| "grad_norm": 0.5649553537368774, | |
| "learning_rate": 1.564292134179931e-06, | |
| "loss": 0.016019219532608986, | |
| "step": 291072 | |
| }, | |
| { | |
| "epoch": 8.446493288104143, | |
| "grad_norm": 4.499303817749023, | |
| "learning_rate": 1.5568699080919662e-06, | |
| "loss": 0.015267412178218365, | |
| "step": 291328 | |
| }, | |
| { | |
| "epoch": 8.453915514192108, | |
| "grad_norm": 0.7536742687225342, | |
| "learning_rate": 1.549447682004001e-06, | |
| "loss": 0.015383939258754253, | |
| "step": 291584 | |
| }, | |
| { | |
| "epoch": 8.461337740280072, | |
| "grad_norm": 0.4496996998786926, | |
| "learning_rate": 1.542025455916036e-06, | |
| "loss": 0.01572938822209835, | |
| "step": 291840 | |
| }, | |
| { | |
| "epoch": 8.468759966368038, | |
| "grad_norm": 1.2914137840270996, | |
| "learning_rate": 1.5346032298280711e-06, | |
| "loss": 0.014957373961806297, | |
| "step": 292096 | |
| }, | |
| { | |
| "epoch": 8.476182192456003, | |
| "grad_norm": 0.8087487816810608, | |
| "learning_rate": 1.5271810037401065e-06, | |
| "loss": 0.015750816091895103, | |
| "step": 292352 | |
| }, | |
| { | |
| "epoch": 8.483604418543967, | |
| "grad_norm": 0.6612663269042969, | |
| "learning_rate": 1.5197587776521414e-06, | |
| "loss": 0.016216235235333443, | |
| "step": 292608 | |
| }, | |
| { | |
| "epoch": 8.491026644631933, | |
| "grad_norm": 0.6684610843658447, | |
| "learning_rate": 1.5123365515641763e-06, | |
| "loss": 0.015619264915585518, | |
| "step": 292864 | |
| }, | |
| { | |
| "epoch": 8.498448870719898, | |
| "grad_norm": 0.6170671582221985, | |
| "learning_rate": 1.5049433185468673e-06, | |
| "loss": 0.015738721936941147, | |
| "step": 293120 | |
| }, | |
| { | |
| "epoch": 8.505871096807862, | |
| "grad_norm": 1.4450145959854126, | |
| "learning_rate": 1.4975210924589023e-06, | |
| "loss": 0.015060283243656158, | |
| "step": 293376 | |
| }, | |
| { | |
| "epoch": 8.513293322895828, | |
| "grad_norm": 1.1573145389556885, | |
| "learning_rate": 1.4900988663709376e-06, | |
| "loss": 0.015352794900536537, | |
| "step": 293632 | |
| }, | |
| { | |
| "epoch": 8.520715548983793, | |
| "grad_norm": 0.8706235885620117, | |
| "learning_rate": 1.4826766402829725e-06, | |
| "loss": 0.016430571675300598, | |
| "step": 293888 | |
| }, | |
| { | |
| "epoch": 8.528137775071757, | |
| "grad_norm": 0.9027552008628845, | |
| "learning_rate": 1.4752834072656636e-06, | |
| "loss": 0.016407093033194542, | |
| "step": 294144 | |
| }, | |
| { | |
| "epoch": 8.535560001159723, | |
| "grad_norm": 1.0352160930633545, | |
| "learning_rate": 1.4678611811776985e-06, | |
| "loss": 0.015707889571785927, | |
| "step": 294400 | |
| }, | |
| { | |
| "epoch": 8.542982227247688, | |
| "grad_norm": 0.7208117246627808, | |
| "learning_rate": 1.4604389550897336e-06, | |
| "loss": 0.014801407232880592, | |
| "step": 294656 | |
| }, | |
| { | |
| "epoch": 8.550404453335652, | |
| "grad_norm": 0.96685391664505, | |
| "learning_rate": 1.4530167290017687e-06, | |
| "loss": 0.01515493355691433, | |
| "step": 294912 | |
| }, | |
| { | |
| "epoch": 8.557826679423618, | |
| "grad_norm": 0.4770279824733734, | |
| "learning_rate": 1.4455945029138039e-06, | |
| "loss": 0.015542468056082726, | |
| "step": 295168 | |
| }, | |
| { | |
| "epoch": 8.565248905511583, | |
| "grad_norm": 0.8233969807624817, | |
| "learning_rate": 1.4381722768258388e-06, | |
| "loss": 0.015357165597379208, | |
| "step": 295424 | |
| }, | |
| { | |
| "epoch": 8.572671131599547, | |
| "grad_norm": 1.6331242322921753, | |
| "learning_rate": 1.430750050737874e-06, | |
| "loss": 0.015083154663443565, | |
| "step": 295680 | |
| }, | |
| { | |
| "epoch": 8.580093357687513, | |
| "grad_norm": 1.0932587385177612, | |
| "learning_rate": 1.4233278246499088e-06, | |
| "loss": 0.01545856986194849, | |
| "step": 295936 | |
| }, | |
| { | |
| "epoch": 8.587515583775478, | |
| "grad_norm": 0.8885460495948792, | |
| "learning_rate": 1.4159055985619437e-06, | |
| "loss": 0.015700651332736015, | |
| "step": 296192 | |
| }, | |
| { | |
| "epoch": 8.594937809863442, | |
| "grad_norm": 0.6702226400375366, | |
| "learning_rate": 1.4084833724739789e-06, | |
| "loss": 0.01488122995942831, | |
| "step": 296448 | |
| }, | |
| { | |
| "epoch": 8.602360035951408, | |
| "grad_norm": 1.4412716627120972, | |
| "learning_rate": 1.4010611463860138e-06, | |
| "loss": 0.015476124361157417, | |
| "step": 296704 | |
| }, | |
| { | |
| "epoch": 8.609782262039372, | |
| "grad_norm": 0.9831516146659851, | |
| "learning_rate": 1.393638920298049e-06, | |
| "loss": 0.014927126467227936, | |
| "step": 296960 | |
| }, | |
| { | |
| "epoch": 8.617204488127337, | |
| "grad_norm": 0.647412121295929, | |
| "learning_rate": 1.3862166942100838e-06, | |
| "loss": 0.015709539875388145, | |
| "step": 297216 | |
| }, | |
| { | |
| "epoch": 8.624626714215303, | |
| "grad_norm": 1.1216012239456177, | |
| "learning_rate": 1.3787944681221187e-06, | |
| "loss": 0.016202064231038094, | |
| "step": 297472 | |
| }, | |
| { | |
| "epoch": 8.632048940303267, | |
| "grad_norm": 1.1198736429214478, | |
| "learning_rate": 1.3713722420341539e-06, | |
| "loss": 0.015115165151655674, | |
| "step": 297728 | |
| }, | |
| { | |
| "epoch": 8.639471166391232, | |
| "grad_norm": 1.232311725616455, | |
| "learning_rate": 1.363950015946189e-06, | |
| "loss": 0.015352166257798672, | |
| "step": 297984 | |
| }, | |
| { | |
| "epoch": 8.646893392479198, | |
| "grad_norm": 1.0231200456619263, | |
| "learning_rate": 1.3565277898582241e-06, | |
| "loss": 0.014888345263898373, | |
| "step": 298240 | |
| }, | |
| { | |
| "epoch": 8.654315618567162, | |
| "grad_norm": 0.3901139795780182, | |
| "learning_rate": 1.349134556840915e-06, | |
| "loss": 0.015830621123313904, | |
| "step": 298496 | |
| }, | |
| { | |
| "epoch": 8.661737844655127, | |
| "grad_norm": 1.5548733472824097, | |
| "learning_rate": 1.34171233075295e-06, | |
| "loss": 0.014992362819612026, | |
| "step": 298752 | |
| }, | |
| { | |
| "epoch": 8.669160070743093, | |
| "grad_norm": 0.3579477369785309, | |
| "learning_rate": 1.334290104664985e-06, | |
| "loss": 0.01570323295891285, | |
| "step": 299008 | |
| }, | |
| { | |
| "epoch": 8.676582296831057, | |
| "grad_norm": 0.6490548849105835, | |
| "learning_rate": 1.3268678785770203e-06, | |
| "loss": 0.015143617987632751, | |
| "step": 299264 | |
| }, | |
| { | |
| "epoch": 8.684004522919022, | |
| "grad_norm": 0.6476681232452393, | |
| "learning_rate": 1.3194456524890553e-06, | |
| "loss": 0.01427437737584114, | |
| "step": 299520 | |
| }, | |
| { | |
| "epoch": 8.691426749006988, | |
| "grad_norm": 0.826097846031189, | |
| "learning_rate": 1.3120234264010904e-06, | |
| "loss": 0.01496600080281496, | |
| "step": 299776 | |
| }, | |
| { | |
| "epoch": 8.698848975094952, | |
| "grad_norm": 0.5644974708557129, | |
| "learning_rate": 1.3046012003131253e-06, | |
| "loss": 0.0155008714646101, | |
| "step": 300032 | |
| }, | |
| { | |
| "epoch": 8.706271201182917, | |
| "grad_norm": 0.22873087227344513, | |
| "learning_rate": 1.2971789742251602e-06, | |
| "loss": 0.014675834216177464, | |
| "step": 300288 | |
| }, | |
| { | |
| "epoch": 8.713693427270883, | |
| "grad_norm": 0.9060187935829163, | |
| "learning_rate": 1.2897567481371953e-06, | |
| "loss": 0.015209571458399296, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 8.721115653358847, | |
| "grad_norm": 0.6566962599754333, | |
| "learning_rate": 1.2823635151198866e-06, | |
| "loss": 0.015152347274124622, | |
| "step": 300800 | |
| }, | |
| { | |
| "epoch": 8.728537879446812, | |
| "grad_norm": 0.7252996563911438, | |
| "learning_rate": 1.2749412890319215e-06, | |
| "loss": 0.015171556733548641, | |
| "step": 301056 | |
| }, | |
| { | |
| "epoch": 8.735960105534778, | |
| "grad_norm": 0.5341917872428894, | |
| "learning_rate": 1.2675190629439566e-06, | |
| "loss": 0.014903398230671883, | |
| "step": 301312 | |
| }, | |
| { | |
| "epoch": 8.743382331622742, | |
| "grad_norm": 0.8702526688575745, | |
| "learning_rate": 1.2600968368559916e-06, | |
| "loss": 0.014965674839913845, | |
| "step": 301568 | |
| }, | |
| { | |
| "epoch": 8.750804557710707, | |
| "grad_norm": 1.5633689165115356, | |
| "learning_rate": 1.2526746107680265e-06, | |
| "loss": 0.015623964369297028, | |
| "step": 301824 | |
| }, | |
| { | |
| "epoch": 8.758226783798673, | |
| "grad_norm": 1.444109559059143, | |
| "learning_rate": 1.2452523846800616e-06, | |
| "loss": 0.014967157505452633, | |
| "step": 302080 | |
| }, | |
| { | |
| "epoch": 8.765649009886637, | |
| "grad_norm": 0.8176071643829346, | |
| "learning_rate": 1.2378301585920965e-06, | |
| "loss": 0.014755125157535076, | |
| "step": 302336 | |
| }, | |
| { | |
| "epoch": 8.773071235974601, | |
| "grad_norm": 1.027308464050293, | |
| "learning_rate": 1.2304079325041317e-06, | |
| "loss": 0.015455594286322594, | |
| "step": 302592 | |
| }, | |
| { | |
| "epoch": 8.780493462062568, | |
| "grad_norm": 0.599094569683075, | |
| "learning_rate": 1.2229857064161668e-06, | |
| "loss": 0.014417259022593498, | |
| "step": 302848 | |
| }, | |
| { | |
| "epoch": 8.787915688150532, | |
| "grad_norm": 0.8513904809951782, | |
| "learning_rate": 1.2155634803282017e-06, | |
| "loss": 0.01419772394001484, | |
| "step": 303104 | |
| }, | |
| { | |
| "epoch": 8.795337914238496, | |
| "grad_norm": 0.8022279739379883, | |
| "learning_rate": 1.2081412542402366e-06, | |
| "loss": 0.014917213469743729, | |
| "step": 303360 | |
| }, | |
| { | |
| "epoch": 8.802760140326463, | |
| "grad_norm": 0.937586784362793, | |
| "learning_rate": 1.2007190281522717e-06, | |
| "loss": 0.014808700419962406, | |
| "step": 303616 | |
| }, | |
| { | |
| "epoch": 8.810182366414427, | |
| "grad_norm": 2.8443167209625244, | |
| "learning_rate": 1.1932968020643067e-06, | |
| "loss": 0.01453117560595274, | |
| "step": 303872 | |
| }, | |
| { | |
| "epoch": 8.817604592502391, | |
| "grad_norm": 2.0082223415374756, | |
| "learning_rate": 1.1858745759763418e-06, | |
| "loss": 0.014921676367521286, | |
| "step": 304128 | |
| }, | |
| { | |
| "epoch": 8.825026818590358, | |
| "grad_norm": 0.7708047032356262, | |
| "learning_rate": 1.178452349888377e-06, | |
| "loss": 0.015060078352689743, | |
| "step": 304384 | |
| }, | |
| { | |
| "epoch": 8.832449044678322, | |
| "grad_norm": 1.7523194551467896, | |
| "learning_rate": 1.1710301238004118e-06, | |
| "loss": 0.015041918493807316, | |
| "step": 304640 | |
| }, | |
| { | |
| "epoch": 8.839871270766286, | |
| "grad_norm": 0.9623550772666931, | |
| "learning_rate": 1.1636368907831029e-06, | |
| "loss": 0.014950048178434372, | |
| "step": 304896 | |
| }, | |
| { | |
| "epoch": 8.847293496854252, | |
| "grad_norm": 0.5514088273048401, | |
| "learning_rate": 1.156214664695138e-06, | |
| "loss": 0.015058852732181549, | |
| "step": 305152 | |
| }, | |
| { | |
| "epoch": 8.854715722942217, | |
| "grad_norm": 0.7167655229568481, | |
| "learning_rate": 1.1487924386071731e-06, | |
| "loss": 0.013770895078778267, | |
| "step": 305408 | |
| }, | |
| { | |
| "epoch": 8.862137949030181, | |
| "grad_norm": 0.8407350778579712, | |
| "learning_rate": 1.1413992055898642e-06, | |
| "loss": 0.014811371453106403, | |
| "step": 305664 | |
| }, | |
| { | |
| "epoch": 8.869560175118147, | |
| "grad_norm": 0.7476754784584045, | |
| "learning_rate": 1.133976979501899e-06, | |
| "loss": 0.014678629115223885, | |
| "step": 305920 | |
| }, | |
| { | |
| "epoch": 8.876982401206112, | |
| "grad_norm": 1.1233904361724854, | |
| "learning_rate": 1.1265547534139342e-06, | |
| "loss": 0.014845655299723148, | |
| "step": 306176 | |
| }, | |
| { | |
| "epoch": 8.884404627294076, | |
| "grad_norm": 0.7537969350814819, | |
| "learning_rate": 1.1191325273259691e-06, | |
| "loss": 0.014918416738510132, | |
| "step": 306432 | |
| }, | |
| { | |
| "epoch": 8.891826853382042, | |
| "grad_norm": 1.27903151512146, | |
| "learning_rate": 1.1117103012380043e-06, | |
| "loss": 0.014179849997162819, | |
| "step": 306688 | |
| }, | |
| { | |
| "epoch": 8.899249079470007, | |
| "grad_norm": 1.0804471969604492, | |
| "learning_rate": 1.1042880751500392e-06, | |
| "loss": 0.01441126223653555, | |
| "step": 306944 | |
| }, | |
| { | |
| "epoch": 8.906671305557971, | |
| "grad_norm": 1.2349823713302612, | |
| "learning_rate": 1.0968658490620743e-06, | |
| "loss": 0.014311261475086212, | |
| "step": 307200 | |
| }, | |
| { | |
| "epoch": 8.914093531645937, | |
| "grad_norm": 0.40392452478408813, | |
| "learning_rate": 1.0894436229741092e-06, | |
| "loss": 0.014079777523875237, | |
| "step": 307456 | |
| }, | |
| { | |
| "epoch": 8.921515757733902, | |
| "grad_norm": 1.213279128074646, | |
| "learning_rate": 1.0820213968861444e-06, | |
| "loss": 0.015064643695950508, | |
| "step": 307712 | |
| }, | |
| { | |
| "epoch": 8.928937983821866, | |
| "grad_norm": 0.8592965602874756, | |
| "learning_rate": 1.0745991707981793e-06, | |
| "loss": 0.01453450694680214, | |
| "step": 307968 | |
| }, | |
| { | |
| "epoch": 8.936360209909832, | |
| "grad_norm": 0.7901779413223267, | |
| "learning_rate": 1.0671769447102144e-06, | |
| "loss": 0.014776786789298058, | |
| "step": 308224 | |
| }, | |
| { | |
| "epoch": 8.943782435997797, | |
| "grad_norm": 3.1283018589019775, | |
| "learning_rate": 1.0597547186222493e-06, | |
| "loss": 0.014097070321440697, | |
| "step": 308480 | |
| }, | |
| { | |
| "epoch": 8.951204662085761, | |
| "grad_norm": 1.0753629207611084, | |
| "learning_rate": 1.0523324925342844e-06, | |
| "loss": 0.01509914267808199, | |
| "step": 308736 | |
| }, | |
| { | |
| "epoch": 8.958626888173727, | |
| "grad_norm": 0.6232772469520569, | |
| "learning_rate": 1.0449102664463194e-06, | |
| "loss": 0.014315241016447544, | |
| "step": 308992 | |
| }, | |
| { | |
| "epoch": 8.966049114261692, | |
| "grad_norm": 0.572321891784668, | |
| "learning_rate": 1.0375170334290104e-06, | |
| "loss": 0.014138281345367432, | |
| "step": 309248 | |
| }, | |
| { | |
| "epoch": 8.973471340349656, | |
| "grad_norm": 0.607771098613739, | |
| "learning_rate": 1.0300948073410455e-06, | |
| "loss": 0.01435596588999033, | |
| "step": 309504 | |
| }, | |
| { | |
| "epoch": 8.980893566437622, | |
| "grad_norm": 0.7871960997581482, | |
| "learning_rate": 1.0226725812530807e-06, | |
| "loss": 0.014779603108763695, | |
| "step": 309760 | |
| }, | |
| { | |
| "epoch": 8.988315792525587, | |
| "grad_norm": 0.7029366493225098, | |
| "learning_rate": 1.0152503551651156e-06, | |
| "loss": 0.013916012831032276, | |
| "step": 310016 | |
| }, | |
| { | |
| "epoch": 8.995738018613551, | |
| "grad_norm": 0.8514177203178406, | |
| "learning_rate": 1.0078281290771507e-06, | |
| "loss": 0.014952397905290127, | |
| "step": 310272 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_f1_B-DISEASE": 0.8909772930544246, | |
| "eval_f1_B-PROCEDURE": 0.8719728705366049, | |
| "eval_f1_B-SYMPTOM": 0.8780038589721102, | |
| "eval_f1_I-DISEASE": 0.9171104194293854, | |
| "eval_f1_I-PROCEDURE": 0.9087704166891273, | |
| "eval_f1_I-SYMPTOM": 0.8961731351375773, | |
| "eval_f1_O": 0.9530769216003563, | |
| "eval_f1_macro": 0.9022978450599409, | |
| "eval_f1_micro": 0.9312516665588606, | |
| "eval_loss": 0.08733360469341278, | |
| "eval_precision_B-DISEASE": 0.8953397901966571, | |
| "eval_precision_B-PROCEDURE": 0.8749866552791715, | |
| "eval_precision_B-SYMPTOM": 0.8848720555089052, | |
| "eval_precision_I-DISEASE": 0.9126851349564884, | |
| "eval_precision_I-PROCEDURE": 0.9130722069368921, | |
| "eval_precision_I-SYMPTOM": 0.9004993946731235, | |
| "eval_precision_O": 0.9538147672872851, | |
| "eval_precision_macro": 0.9050385721197889, | |
| "eval_precision_micro": 0.9325565886642287, | |
| "eval_rauc_macro": 0.9417385443677894, | |
| "eval_rauc_micro": 0.9591693280703284, | |
| "eval_recall_B-DISEASE": 0.8866571018651362, | |
| "eval_recall_B-PROCEDURE": 0.8689797757574151, | |
| "eval_recall_B-SYMPTOM": 0.8712414603367999, | |
| "eval_recall_I-DISEASE": 0.9215788262370541, | |
| "eval_recall_I-PROCEDURE": 0.904508970727101, | |
| "eval_recall_I-SYMPTOM": 0.8918882460505411, | |
| "eval_recall_O": 0.952340216586556, | |
| "eval_recall_macro": 0.8995992282229434, | |
| "eval_recall_micro": 0.9299503912933749, | |
| "eval_roc_auc_B-DISEASE": 0.9422091186090101, | |
| "eval_roc_auc_B-PROCEDURE": 0.9331456969416079, | |
| "eval_roc_auc_B-SYMPTOM": 0.9341184900717953, | |
| "eval_roc_auc_I-DISEASE": 0.9546612665805108, | |
| "eval_roc_auc_I-PROCEDURE": 0.9472188910849895, | |
| "eval_roc_auc_I-SYMPTOM": 0.9372537405360938, | |
| "eval_roc_auc_O": 0.9435626067505182, | |
| "eval_runtime": 50.3148, | |
| "eval_samples_per_second": 281.23, | |
| "eval_steps_per_second": 35.159, | |
| "step": 310419 | |
| }, | |
| { | |
| "epoch": 9.003160244701517, | |
| "grad_norm": 2.354722261428833, | |
| "learning_rate": 1.0004059029891856e-06, | |
| "loss": 0.013831463642418385, | |
| "step": 310528 | |
| }, | |
| { | |
| "epoch": 9.010582470789481, | |
| "grad_norm": 0.691369354724884, | |
| "learning_rate": 9.929836769012205e-07, | |
| "loss": 0.01279458124190569, | |
| "step": 310784 | |
| }, | |
| { | |
| "epoch": 9.018004696877446, | |
| "grad_norm": 0.43184471130371094, | |
| "learning_rate": 9.855614508132557e-07, | |
| "loss": 0.013234521262347698, | |
| "step": 311040 | |
| }, | |
| { | |
| "epoch": 9.025426922965412, | |
| "grad_norm": 0.7044182419776917, | |
| "learning_rate": 9.781392247252908e-07, | |
| "loss": 0.013758363202214241, | |
| "step": 311296 | |
| }, | |
| { | |
| "epoch": 9.032849149053376, | |
| "grad_norm": 0.7477379441261292, | |
| "learning_rate": 9.707169986373257e-07, | |
| "loss": 0.012897802516818047, | |
| "step": 311552 | |
| }, | |
| { | |
| "epoch": 9.04027137514134, | |
| "grad_norm": 1.0956915616989136, | |
| "learning_rate": 9.632947725493608e-07, | |
| "loss": 0.013335512019693851, | |
| "step": 311808 | |
| }, | |
| { | |
| "epoch": 9.047693601229307, | |
| "grad_norm": 1.9793846607208252, | |
| "learning_rate": 9.558725464613958e-07, | |
| "loss": 0.013732228428125381, | |
| "step": 312064 | |
| }, | |
| { | |
| "epoch": 9.055115827317271, | |
| "grad_norm": 0.616520881652832, | |
| "learning_rate": 9.484503203734308e-07, | |
| "loss": 0.01280524767935276, | |
| "step": 312320 | |
| }, | |
| { | |
| "epoch": 9.062538053405236, | |
| "grad_norm": 0.8736835718154907, | |
| "learning_rate": 9.410280942854659e-07, | |
| "loss": 0.014025941491127014, | |
| "step": 312576 | |
| }, | |
| { | |
| "epoch": 9.069960279493202, | |
| "grad_norm": 0.5940849781036377, | |
| "learning_rate": 9.336058681975009e-07, | |
| "loss": 0.013086741790175438, | |
| "step": 312832 | |
| }, | |
| { | |
| "epoch": 9.077382505581166, | |
| "grad_norm": 0.5494786500930786, | |
| "learning_rate": 9.261836421095359e-07, | |
| "loss": 0.012793807312846184, | |
| "step": 313088 | |
| }, | |
| { | |
| "epoch": 9.08480473166913, | |
| "grad_norm": 0.6113333106040955, | |
| "learning_rate": 9.187614160215709e-07, | |
| "loss": 0.01290515810251236, | |
| "step": 313344 | |
| }, | |
| { | |
| "epoch": 9.092226957757097, | |
| "grad_norm": 0.6217535138130188, | |
| "learning_rate": 9.11368183004262e-07, | |
| "loss": 0.012917356565594673, | |
| "step": 313600 | |
| }, | |
| { | |
| "epoch": 9.099649183845061, | |
| "grad_norm": 0.5938876271247864, | |
| "learning_rate": 9.039459569162971e-07, | |
| "loss": 0.013540408574044704, | |
| "step": 313856 | |
| }, | |
| { | |
| "epoch": 9.107071409933026, | |
| "grad_norm": 0.9717910885810852, | |
| "learning_rate": 8.965527238989882e-07, | |
| "loss": 0.013223686255514622, | |
| "step": 314112 | |
| }, | |
| { | |
| "epoch": 9.114493636020992, | |
| "grad_norm": 1.1537644863128662, | |
| "learning_rate": 8.891304978110232e-07, | |
| "loss": 0.012715338729321957, | |
| "step": 314368 | |
| }, | |
| { | |
| "epoch": 9.121915862108956, | |
| "grad_norm": 0.7134986519813538, | |
| "learning_rate": 8.817082717230582e-07, | |
| "loss": 0.013339064083993435, | |
| "step": 314624 | |
| }, | |
| { | |
| "epoch": 9.12933808819692, | |
| "grad_norm": 0.5394193530082703, | |
| "learning_rate": 8.742860456350933e-07, | |
| "loss": 0.01288061123341322, | |
| "step": 314880 | |
| }, | |
| { | |
| "epoch": 9.136760314284887, | |
| "grad_norm": 1.7707648277282715, | |
| "learning_rate": 8.668638195471284e-07, | |
| "loss": 0.012954437173902988, | |
| "step": 315136 | |
| }, | |
| { | |
| "epoch": 9.144182540372851, | |
| "grad_norm": 1.5921719074249268, | |
| "learning_rate": 8.594415934591634e-07, | |
| "loss": 0.013480898924171925, | |
| "step": 315392 | |
| }, | |
| { | |
| "epoch": 9.151604766460816, | |
| "grad_norm": 0.38697513937950134, | |
| "learning_rate": 8.520193673711983e-07, | |
| "loss": 0.01277309563010931, | |
| "step": 315648 | |
| }, | |
| { | |
| "epoch": 9.159026992548782, | |
| "grad_norm": 0.7114886045455933, | |
| "learning_rate": 8.445971412832333e-07, | |
| "loss": 0.012951008044183254, | |
| "step": 315904 | |
| }, | |
| { | |
| "epoch": 9.166449218636746, | |
| "grad_norm": 0.5460690855979919, | |
| "learning_rate": 8.371749151952684e-07, | |
| "loss": 0.012513099238276482, | |
| "step": 316160 | |
| }, | |
| { | |
| "epoch": 9.17387144472471, | |
| "grad_norm": 1.003661036491394, | |
| "learning_rate": 8.297816821779596e-07, | |
| "loss": 0.012976228259503841, | |
| "step": 316416 | |
| }, | |
| { | |
| "epoch": 9.181293670812675, | |
| "grad_norm": 1.0079479217529297, | |
| "learning_rate": 8.223594560899945e-07, | |
| "loss": 0.013462617062032223, | |
| "step": 316672 | |
| }, | |
| { | |
| "epoch": 9.188715896900641, | |
| "grad_norm": 0.7562265396118164, | |
| "learning_rate": 8.149372300020296e-07, | |
| "loss": 0.013435247354209423, | |
| "step": 316928 | |
| }, | |
| { | |
| "epoch": 9.196138122988605, | |
| "grad_norm": 1.1886365413665771, | |
| "learning_rate": 8.075150039140646e-07, | |
| "loss": 0.012629357166588306, | |
| "step": 317184 | |
| }, | |
| { | |
| "epoch": 9.20356034907657, | |
| "grad_norm": 0.8576349020004272, | |
| "learning_rate": 8.000927778260996e-07, | |
| "loss": 0.012689920142292976, | |
| "step": 317440 | |
| }, | |
| { | |
| "epoch": 9.210982575164536, | |
| "grad_norm": 0.5762125253677368, | |
| "learning_rate": 7.926705517381346e-07, | |
| "loss": 0.013234340585768223, | |
| "step": 317696 | |
| }, | |
| { | |
| "epoch": 9.2184048012525, | |
| "grad_norm": 1.1880782842636108, | |
| "learning_rate": 7.852483256501698e-07, | |
| "loss": 0.013366462662816048, | |
| "step": 317952 | |
| }, | |
| { | |
| "epoch": 9.225827027340465, | |
| "grad_norm": 0.9516034722328186, | |
| "learning_rate": 7.778260995622047e-07, | |
| "loss": 0.012884012423455715, | |
| "step": 318208 | |
| }, | |
| { | |
| "epoch": 9.233249253428431, | |
| "grad_norm": 0.629169762134552, | |
| "learning_rate": 7.704038734742397e-07, | |
| "loss": 0.012939594686031342, | |
| "step": 318464 | |
| }, | |
| { | |
| "epoch": 9.240671479516395, | |
| "grad_norm": 0.5135284066200256, | |
| "learning_rate": 7.630106404569309e-07, | |
| "loss": 0.013264158740639687, | |
| "step": 318720 | |
| }, | |
| { | |
| "epoch": 9.24809370560436, | |
| "grad_norm": 0.5182939171791077, | |
| "learning_rate": 7.555884143689658e-07, | |
| "loss": 0.013006733730435371, | |
| "step": 318976 | |
| }, | |
| { | |
| "epoch": 9.255515931692326, | |
| "grad_norm": 0.6168907284736633, | |
| "learning_rate": 7.48166188281001e-07, | |
| "loss": 0.012938130646944046, | |
| "step": 319232 | |
| }, | |
| { | |
| "epoch": 9.26293815778029, | |
| "grad_norm": 0.38301581144332886, | |
| "learning_rate": 7.407439621930359e-07, | |
| "loss": 0.0127954026684165, | |
| "step": 319488 | |
| }, | |
| { | |
| "epoch": 9.270360383868255, | |
| "grad_norm": 1.638322114944458, | |
| "learning_rate": 7.333217361050709e-07, | |
| "loss": 0.01299678348004818, | |
| "step": 319744 | |
| }, | |
| { | |
| "epoch": 9.27778260995622, | |
| "grad_norm": 0.9846634864807129, | |
| "learning_rate": 7.259285030877621e-07, | |
| "loss": 0.014220071956515312, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 9.285204836044185, | |
| "grad_norm": 0.6679787635803223, | |
| "learning_rate": 7.18506276999797e-07, | |
| "loss": 0.012765005230903625, | |
| "step": 320256 | |
| }, | |
| { | |
| "epoch": 9.29262706213215, | |
| "grad_norm": 0.8304718732833862, | |
| "learning_rate": 7.110840509118321e-07, | |
| "loss": 0.012488399632275105, | |
| "step": 320512 | |
| }, | |
| { | |
| "epoch": 9.300049288220116, | |
| "grad_norm": 0.9025726318359375, | |
| "learning_rate": 7.036618248238672e-07, | |
| "loss": 0.012031358666718006, | |
| "step": 320768 | |
| }, | |
| { | |
| "epoch": 9.30747151430808, | |
| "grad_norm": 0.5530338883399963, | |
| "learning_rate": 6.962395987359022e-07, | |
| "loss": 0.01363558042794466, | |
| "step": 321024 | |
| }, | |
| { | |
| "epoch": 9.314893740396045, | |
| "grad_norm": 0.8872376680374146, | |
| "learning_rate": 6.888173726479372e-07, | |
| "loss": 0.012918893247842789, | |
| "step": 321280 | |
| }, | |
| { | |
| "epoch": 9.32231596648401, | |
| "grad_norm": 1.322966456413269, | |
| "learning_rate": 6.813951465599722e-07, | |
| "loss": 0.012260723859071732, | |
| "step": 321536 | |
| }, | |
| { | |
| "epoch": 9.329738192571975, | |
| "grad_norm": 0.671420693397522, | |
| "learning_rate": 6.739729204720071e-07, | |
| "loss": 0.01272868923842907, | |
| "step": 321792 | |
| }, | |
| { | |
| "epoch": 9.33716041865994, | |
| "grad_norm": 0.7592810988426208, | |
| "learning_rate": 6.665506943840423e-07, | |
| "loss": 0.01253275666385889, | |
| "step": 322048 | |
| }, | |
| { | |
| "epoch": 9.344582644747906, | |
| "grad_norm": 0.4805082380771637, | |
| "learning_rate": 6.591284682960773e-07, | |
| "loss": 0.012881642207503319, | |
| "step": 322304 | |
| }, | |
| { | |
| "epoch": 9.35200487083587, | |
| "grad_norm": 0.6440457105636597, | |
| "learning_rate": 6.517062422081123e-07, | |
| "loss": 0.01271052472293377, | |
| "step": 322560 | |
| }, | |
| { | |
| "epoch": 9.359427096923834, | |
| "grad_norm": 1.0848995447158813, | |
| "learning_rate": 6.442840161201473e-07, | |
| "loss": 0.013099320232868195, | |
| "step": 322816 | |
| }, | |
| { | |
| "epoch": 9.3668493230118, | |
| "grad_norm": 0.7066841125488281, | |
| "learning_rate": 6.368617900321824e-07, | |
| "loss": 0.01212881039828062, | |
| "step": 323072 | |
| }, | |
| { | |
| "epoch": 9.374271549099765, | |
| "grad_norm": 0.3485105037689209, | |
| "learning_rate": 6.294395639442173e-07, | |
| "loss": 0.012444918043911457, | |
| "step": 323328 | |
| }, | |
| { | |
| "epoch": 9.38169377518773, | |
| "grad_norm": 0.562164306640625, | |
| "learning_rate": 6.220173378562524e-07, | |
| "loss": 0.013291002251207829, | |
| "step": 323584 | |
| }, | |
| { | |
| "epoch": 9.389116001275696, | |
| "grad_norm": 0.7132983207702637, | |
| "learning_rate": 6.145951117682874e-07, | |
| "loss": 0.01332166325300932, | |
| "step": 323840 | |
| }, | |
| { | |
| "epoch": 9.39653822736366, | |
| "grad_norm": 0.46629589796066284, | |
| "learning_rate": 6.071728856803225e-07, | |
| "loss": 0.012883233837783337, | |
| "step": 324096 | |
| }, | |
| { | |
| "epoch": 9.403960453451624, | |
| "grad_norm": 1.0050158500671387, | |
| "learning_rate": 5.997506595923575e-07, | |
| "loss": 0.012202881276607513, | |
| "step": 324352 | |
| }, | |
| { | |
| "epoch": 9.41138267953959, | |
| "grad_norm": 0.6948258876800537, | |
| "learning_rate": 5.923284335043925e-07, | |
| "loss": 0.012272646650671959, | |
| "step": 324608 | |
| }, | |
| { | |
| "epoch": 9.418804905627555, | |
| "grad_norm": 0.7558926343917847, | |
| "learning_rate": 5.849062074164275e-07, | |
| "loss": 0.01273445412516594, | |
| "step": 324864 | |
| }, | |
| { | |
| "epoch": 9.42622713171552, | |
| "grad_norm": 0.852893054485321, | |
| "learning_rate": 5.774839813284625e-07, | |
| "loss": 0.01295941136777401, | |
| "step": 325120 | |
| }, | |
| { | |
| "epoch": 9.433649357803485, | |
| "grad_norm": 1.063214659690857, | |
| "learning_rate": 5.700617552404976e-07, | |
| "loss": 0.012278242968022823, | |
| "step": 325376 | |
| }, | |
| { | |
| "epoch": 9.44107158389145, | |
| "grad_norm": 1.389140009880066, | |
| "learning_rate": 5.626685222231887e-07, | |
| "loss": 0.012448016554117203, | |
| "step": 325632 | |
| }, | |
| { | |
| "epoch": 9.448493809979414, | |
| "grad_norm": 0.7348615527153015, | |
| "learning_rate": 5.552462961352237e-07, | |
| "loss": 0.012524059973657131, | |
| "step": 325888 | |
| }, | |
| { | |
| "epoch": 9.45591603606738, | |
| "grad_norm": 1.8809510469436646, | |
| "learning_rate": 5.478240700472588e-07, | |
| "loss": 0.012544876895844936, | |
| "step": 326144 | |
| }, | |
| { | |
| "epoch": 9.463338262155345, | |
| "grad_norm": 0.5809705257415771, | |
| "learning_rate": 5.404018439592938e-07, | |
| "loss": 0.01242187898606062, | |
| "step": 326400 | |
| }, | |
| { | |
| "epoch": 9.47076048824331, | |
| "grad_norm": 0.6680272221565247, | |
| "learning_rate": 5.329796178713288e-07, | |
| "loss": 0.012716073542833328, | |
| "step": 326656 | |
| }, | |
| { | |
| "epoch": 9.478182714331275, | |
| "grad_norm": 0.4519018530845642, | |
| "learning_rate": 5.255573917833638e-07, | |
| "loss": 0.012645130045711994, | |
| "step": 326912 | |
| }, | |
| { | |
| "epoch": 9.48560494041924, | |
| "grad_norm": 1.3270643949508667, | |
| "learning_rate": 5.181351656953988e-07, | |
| "loss": 0.012105450965464115, | |
| "step": 327168 | |
| }, | |
| { | |
| "epoch": 9.493027166507204, | |
| "grad_norm": 0.8089259266853333, | |
| "learning_rate": 5.107129396074339e-07, | |
| "loss": 0.012163015082478523, | |
| "step": 327424 | |
| }, | |
| { | |
| "epoch": 9.50044939259517, | |
| "grad_norm": 1.107285976409912, | |
| "learning_rate": 5.032907135194689e-07, | |
| "loss": 0.01267937757074833, | |
| "step": 327680 | |
| }, | |
| { | |
| "epoch": 9.507871618683135, | |
| "grad_norm": 0.49829861521720886, | |
| "learning_rate": 4.958684874315039e-07, | |
| "loss": 0.011714424006640911, | |
| "step": 327936 | |
| }, | |
| { | |
| "epoch": 9.515293844771099, | |
| "grad_norm": 0.42109233140945435, | |
| "learning_rate": 4.884462613435389e-07, | |
| "loss": 0.012568791396915913, | |
| "step": 328192 | |
| }, | |
| { | |
| "epoch": 9.522716070859065, | |
| "grad_norm": 0.863654375076294, | |
| "learning_rate": 4.81024035255574e-07, | |
| "loss": 0.012434298172593117, | |
| "step": 328448 | |
| }, | |
| { | |
| "epoch": 9.53013829694703, | |
| "grad_norm": 0.41061705350875854, | |
| "learning_rate": 4.736018091676089e-07, | |
| "loss": 0.012782089412212372, | |
| "step": 328704 | |
| }, | |
| { | |
| "epoch": 9.537560523034994, | |
| "grad_norm": 1.0521044731140137, | |
| "learning_rate": 4.66179583079644e-07, | |
| "loss": 0.012506213039159775, | |
| "step": 328960 | |
| }, | |
| { | |
| "epoch": 9.54498274912296, | |
| "grad_norm": 0.551642119884491, | |
| "learning_rate": 4.58757356991679e-07, | |
| "loss": 0.012626152485609055, | |
| "step": 329216 | |
| }, | |
| { | |
| "epoch": 9.552404975210925, | |
| "grad_norm": 0.7666914463043213, | |
| "learning_rate": 4.5133513090371404e-07, | |
| "loss": 0.012316416949033737, | |
| "step": 329472 | |
| }, | |
| { | |
| "epoch": 9.559827201298889, | |
| "grad_norm": 0.7405276298522949, | |
| "learning_rate": 4.4391290481574906e-07, | |
| "loss": 0.012117513455450535, | |
| "step": 329728 | |
| }, | |
| { | |
| "epoch": 9.567249427386855, | |
| "grad_norm": 0.8547769784927368, | |
| "learning_rate": 4.364906787277841e-07, | |
| "loss": 0.011824346147477627, | |
| "step": 329984 | |
| }, | |
| { | |
| "epoch": 9.57467165347482, | |
| "grad_norm": 0.9016408920288086, | |
| "learning_rate": 4.2906845263981916e-07, | |
| "loss": 0.012442250736057758, | |
| "step": 330240 | |
| }, | |
| { | |
| "epoch": 9.582093879562784, | |
| "grad_norm": 0.332159161567688, | |
| "learning_rate": 4.2164622655185413e-07, | |
| "loss": 0.01256286446005106, | |
| "step": 330496 | |
| }, | |
| { | |
| "epoch": 9.58951610565075, | |
| "grad_norm": 0.82328200340271, | |
| "learning_rate": 4.1425299353454523e-07, | |
| "loss": 0.012896089814603329, | |
| "step": 330752 | |
| }, | |
| { | |
| "epoch": 9.596938331738714, | |
| "grad_norm": 0.48921993374824524, | |
| "learning_rate": 4.068597605172364e-07, | |
| "loss": 0.012673907913267612, | |
| "step": 331008 | |
| }, | |
| { | |
| "epoch": 9.604360557826679, | |
| "grad_norm": 0.9077366590499878, | |
| "learning_rate": 3.994375344292714e-07, | |
| "loss": 0.012012193910777569, | |
| "step": 331264 | |
| }, | |
| { | |
| "epoch": 9.611782783914645, | |
| "grad_norm": 1.329208254814148, | |
| "learning_rate": 3.920153083413065e-07, | |
| "loss": 0.01216165255755186, | |
| "step": 331520 | |
| }, | |
| { | |
| "epoch": 9.61920501000261, | |
| "grad_norm": 0.8970664143562317, | |
| "learning_rate": 3.8459308225334145e-07, | |
| "loss": 0.01230388879776001, | |
| "step": 331776 | |
| }, | |
| { | |
| "epoch": 9.626627236090574, | |
| "grad_norm": 1.2703585624694824, | |
| "learning_rate": 3.771708561653765e-07, | |
| "loss": 0.011930800974369049, | |
| "step": 332032 | |
| }, | |
| { | |
| "epoch": 9.63404946217854, | |
| "grad_norm": 1.3918522596359253, | |
| "learning_rate": 3.6974863007741155e-07, | |
| "loss": 0.01243145763874054, | |
| "step": 332288 | |
| }, | |
| { | |
| "epoch": 9.641471688266504, | |
| "grad_norm": 0.5909280776977539, | |
| "learning_rate": 3.623264039894465e-07, | |
| "loss": 0.012650659307837486, | |
| "step": 332544 | |
| }, | |
| { | |
| "epoch": 9.648893914354469, | |
| "grad_norm": 0.6482845544815063, | |
| "learning_rate": 3.5490417790148154e-07, | |
| "loss": 0.012349645607173443, | |
| "step": 332800 | |
| }, | |
| { | |
| "epoch": 9.656316140442435, | |
| "grad_norm": 1.1534124612808228, | |
| "learning_rate": 3.474819518135166e-07, | |
| "loss": 0.011956226080656052, | |
| "step": 333056 | |
| }, | |
| { | |
| "epoch": 9.6637383665304, | |
| "grad_norm": 0.71855229139328, | |
| "learning_rate": 3.4005972572555164e-07, | |
| "loss": 0.012277196161448956, | |
| "step": 333312 | |
| }, | |
| { | |
| "epoch": 9.671160592618364, | |
| "grad_norm": 0.5995551347732544, | |
| "learning_rate": 3.326374996375866e-07, | |
| "loss": 0.012190536595880985, | |
| "step": 333568 | |
| }, | |
| { | |
| "epoch": 9.67858281870633, | |
| "grad_norm": 0.7513915300369263, | |
| "learning_rate": 3.252152735496217e-07, | |
| "loss": 0.012531750835478306, | |
| "step": 333824 | |
| }, | |
| { | |
| "epoch": 9.686005044794294, | |
| "grad_norm": 0.9372618198394775, | |
| "learning_rate": 3.177930474616567e-07, | |
| "loss": 0.012361818924546242, | |
| "step": 334080 | |
| }, | |
| { | |
| "epoch": 9.693427270882259, | |
| "grad_norm": 1.2047150135040283, | |
| "learning_rate": 3.103708213736917e-07, | |
| "loss": 0.012158993631601334, | |
| "step": 334336 | |
| }, | |
| { | |
| "epoch": 9.700849496970225, | |
| "grad_norm": 1.2447707653045654, | |
| "learning_rate": 3.0297758835638283e-07, | |
| "loss": 0.013520949520170689, | |
| "step": 334592 | |
| }, | |
| { | |
| "epoch": 9.70827172305819, | |
| "grad_norm": 0.4660310447216034, | |
| "learning_rate": 2.9555536226841785e-07, | |
| "loss": 0.012339101172983646, | |
| "step": 334848 | |
| }, | |
| { | |
| "epoch": 9.715693949146154, | |
| "grad_norm": 0.5139338970184326, | |
| "learning_rate": 2.881331361804529e-07, | |
| "loss": 0.012180372141301632, | |
| "step": 335104 | |
| }, | |
| { | |
| "epoch": 9.72311617523412, | |
| "grad_norm": 0.3752732574939728, | |
| "learning_rate": 2.807109100924879e-07, | |
| "loss": 0.012703219428658485, | |
| "step": 335360 | |
| }, | |
| { | |
| "epoch": 9.730538401322084, | |
| "grad_norm": 0.5550410747528076, | |
| "learning_rate": 2.7328868400452297e-07, | |
| "loss": 0.01223105937242508, | |
| "step": 335616 | |
| }, | |
| { | |
| "epoch": 9.737960627410049, | |
| "grad_norm": 0.6406332850456238, | |
| "learning_rate": 2.65866457916558e-07, | |
| "loss": 0.012319369241595268, | |
| "step": 335872 | |
| }, | |
| { | |
| "epoch": 9.745382853498015, | |
| "grad_norm": 0.7530149817466736, | |
| "learning_rate": 2.5844423182859296e-07, | |
| "loss": 0.012388728559017181, | |
| "step": 336128 | |
| }, | |
| { | |
| "epoch": 9.752805079585979, | |
| "grad_norm": 0.6517982482910156, | |
| "learning_rate": 2.5102200574062803e-07, | |
| "loss": 0.01227741688489914, | |
| "step": 336384 | |
| }, | |
| { | |
| "epoch": 9.760227305673943, | |
| "grad_norm": 0.887261152267456, | |
| "learning_rate": 2.4359977965266306e-07, | |
| "loss": 0.012427708134055138, | |
| "step": 336640 | |
| }, | |
| { | |
| "epoch": 9.76764953176191, | |
| "grad_norm": 0.7006745338439941, | |
| "learning_rate": 2.3617755356469805e-07, | |
| "loss": 0.012144133448600769, | |
| "step": 336896 | |
| }, | |
| { | |
| "epoch": 9.775071757849874, | |
| "grad_norm": 0.7525506615638733, | |
| "learning_rate": 2.2875532747673307e-07, | |
| "loss": 0.012310302816331387, | |
| "step": 337152 | |
| }, | |
| { | |
| "epoch": 9.782493983937838, | |
| "grad_norm": 1.1072790622711182, | |
| "learning_rate": 2.2133310138876812e-07, | |
| "loss": 0.0122316163033247, | |
| "step": 337408 | |
| }, | |
| { | |
| "epoch": 9.789916210025805, | |
| "grad_norm": 0.6953659057617188, | |
| "learning_rate": 2.1391087530080312e-07, | |
| "loss": 0.011760184541344643, | |
| "step": 337664 | |
| }, | |
| { | |
| "epoch": 9.797338436113769, | |
| "grad_norm": 0.9446242451667786, | |
| "learning_rate": 2.0648864921283814e-07, | |
| "loss": 0.011799545958638191, | |
| "step": 337920 | |
| }, | |
| { | |
| "epoch": 9.804760662201733, | |
| "grad_norm": 0.519162118434906, | |
| "learning_rate": 1.990664231248732e-07, | |
| "loss": 0.012246874161064625, | |
| "step": 338176 | |
| }, | |
| { | |
| "epoch": 9.8121828882897, | |
| "grad_norm": 0.3854491412639618, | |
| "learning_rate": 1.916441970369082e-07, | |
| "loss": 0.012221163138747215, | |
| "step": 338432 | |
| }, | |
| { | |
| "epoch": 9.819605114377664, | |
| "grad_norm": 0.31908509135246277, | |
| "learning_rate": 1.842219709489432e-07, | |
| "loss": 0.012250985950231552, | |
| "step": 338688 | |
| }, | |
| { | |
| "epoch": 9.827027340465628, | |
| "grad_norm": 0.8602272868156433, | |
| "learning_rate": 1.7679974486097825e-07, | |
| "loss": 0.01207782980054617, | |
| "step": 338944 | |
| }, | |
| { | |
| "epoch": 9.834449566553594, | |
| "grad_norm": 0.39700064063072205, | |
| "learning_rate": 1.6940651184366938e-07, | |
| "loss": 0.01222145278006792, | |
| "step": 339200 | |
| }, | |
| { | |
| "epoch": 9.841871792641559, | |
| "grad_norm": 0.7912943959236145, | |
| "learning_rate": 1.619842857557044e-07, | |
| "loss": 0.011850698851048946, | |
| "step": 339456 | |
| }, | |
| { | |
| "epoch": 9.849294018729523, | |
| "grad_norm": 0.6283043026924133, | |
| "learning_rate": 1.5456205966773943e-07, | |
| "loss": 0.012278848327696323, | |
| "step": 339712 | |
| }, | |
| { | |
| "epoch": 9.85671624481749, | |
| "grad_norm": 0.5299221873283386, | |
| "learning_rate": 1.4713983357977445e-07, | |
| "loss": 0.011622827500104904, | |
| "step": 339968 | |
| }, | |
| { | |
| "epoch": 9.864138470905454, | |
| "grad_norm": 0.4546976089477539, | |
| "learning_rate": 1.3971760749180947e-07, | |
| "loss": 0.012113522738218307, | |
| "step": 340224 | |
| }, | |
| { | |
| "epoch": 9.871560696993418, | |
| "grad_norm": 0.610438883304596, | |
| "learning_rate": 1.322953814038445e-07, | |
| "loss": 0.012094656936824322, | |
| "step": 340480 | |
| }, | |
| { | |
| "epoch": 9.878982923081384, | |
| "grad_norm": 0.4057309031486511, | |
| "learning_rate": 1.248731553158795e-07, | |
| "loss": 0.011664223857223988, | |
| "step": 340736 | |
| }, | |
| { | |
| "epoch": 9.886405149169349, | |
| "grad_norm": 0.661514163017273, | |
| "learning_rate": 1.1745092922791453e-07, | |
| "loss": 0.011828630231320858, | |
| "step": 340992 | |
| }, | |
| { | |
| "epoch": 9.893827375257313, | |
| "grad_norm": 0.5697357058525085, | |
| "learning_rate": 1.1005769621060566e-07, | |
| "loss": 0.012373203411698341, | |
| "step": 341248 | |
| }, | |
| { | |
| "epoch": 9.901249601345278, | |
| "grad_norm": 0.5505850911140442, | |
| "learning_rate": 1.026354701226407e-07, | |
| "loss": 0.011953489854931831, | |
| "step": 341504 | |
| }, | |
| { | |
| "epoch": 9.908671827433244, | |
| "grad_norm": 0.7869247794151306, | |
| "learning_rate": 9.521324403467572e-08, | |
| "loss": 0.012051274068653584, | |
| "step": 341760 | |
| }, | |
| { | |
| "epoch": 9.916094053521208, | |
| "grad_norm": 1.0268815755844116, | |
| "learning_rate": 8.779101794671074e-08, | |
| "loss": 0.011977662332355976, | |
| "step": 342016 | |
| }, | |
| { | |
| "epoch": 9.923516279609174, | |
| "grad_norm": 0.612642228603363, | |
| "learning_rate": 8.036879185874576e-08, | |
| "loss": 0.01191222108900547, | |
| "step": 342272 | |
| }, | |
| { | |
| "epoch": 9.930938505697139, | |
| "grad_norm": 0.8545002341270447, | |
| "learning_rate": 7.294656577078079e-08, | |
| "loss": 0.012285839766263962, | |
| "step": 342528 | |
| }, | |
| { | |
| "epoch": 9.938360731785103, | |
| "grad_norm": 0.6520818471908569, | |
| "learning_rate": 6.552433968281581e-08, | |
| "loss": 0.012015718966722488, | |
| "step": 342784 | |
| }, | |
| { | |
| "epoch": 9.945782957873067, | |
| "grad_norm": 0.964957594871521, | |
| "learning_rate": 5.8102113594850837e-08, | |
| "loss": 0.01214210782200098, | |
| "step": 343040 | |
| }, | |
| { | |
| "epoch": 9.953205183961034, | |
| "grad_norm": 0.7185612320899963, | |
| "learning_rate": 5.0679887506885865e-08, | |
| "loss": 0.012188009917736053, | |
| "step": 343296 | |
| }, | |
| { | |
| "epoch": 9.960627410048998, | |
| "grad_norm": 0.7212699055671692, | |
| "learning_rate": 4.325766141892088e-08, | |
| "loss": 0.011719409376382828, | |
| "step": 343552 | |
| }, | |
| { | |
| "epoch": 9.968049636136964, | |
| "grad_norm": 0.377102255821228, | |
| "learning_rate": 3.58354353309559e-08, | |
| "loss": 0.011643365025520325, | |
| "step": 343808 | |
| }, | |
| { | |
| "epoch": 9.975471862224929, | |
| "grad_norm": 0.5918943285942078, | |
| "learning_rate": 2.8413209242990927e-08, | |
| "loss": 0.011841170489788055, | |
| "step": 344064 | |
| }, | |
| { | |
| "epoch": 9.982894088312893, | |
| "grad_norm": 0.5792754888534546, | |
| "learning_rate": 2.099098315502595e-08, | |
| "loss": 0.012151996605098248, | |
| "step": 344320 | |
| }, | |
| { | |
| "epoch": 9.990316314400857, | |
| "grad_norm": 0.7357270121574402, | |
| "learning_rate": 1.3568757067060974e-08, | |
| "loss": 0.011309313587844372, | |
| "step": 344576 | |
| }, | |
| { | |
| "epoch": 9.997738540488823, | |
| "grad_norm": 2.9916160106658936, | |
| "learning_rate": 6.146530979095997e-09, | |
| "loss": 0.012390630319714546, | |
| "step": 344832 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_f1_B-DISEASE": 0.8931502462068137, | |
| "eval_f1_B-PROCEDURE": 0.8734556445198579, | |
| "eval_f1_B-SYMPTOM": 0.877200248498654, | |
| "eval_f1_I-DISEASE": 0.9189212789131903, | |
| "eval_f1_I-PROCEDURE": 0.9097494179734131, | |
| "eval_f1_I-SYMPTOM": 0.8976715996705076, | |
| "eval_f1_O": 0.9538437481412978, | |
| "eval_f1_macro": 0.9034274548462476, | |
| "eval_f1_micro": 0.9322616227406649, | |
| "eval_loss": 0.09216833114624023, | |
| "eval_precision_B-DISEASE": 0.8972090404568487, | |
| "eval_precision_B-PROCEDURE": 0.8736641120156992, | |
| "eval_precision_B-SYMPTOM": 0.8788573674900085, | |
| "eval_precision_I-DISEASE": 0.9174834351999707, | |
| "eval_precision_I-PROCEDURE": 0.9159587543644017, | |
| "eval_precision_I-SYMPTOM": 0.8990261924538372, | |
| "eval_precision_O": 0.9541458106219963, | |
| "eval_precision_macro": 0.9051921018003947, | |
| "eval_precision_micro": 0.9332345024077748, | |
| "eval_rauc_macro": 0.9428483050354564, | |
| "eval_rauc_micro": 0.9598937722806111, | |
| "eval_recall_B-DISEASE": 0.889128008927148, | |
| "eval_recall_B-PROCEDURE": 0.873247276486336, | |
| "eval_recall_B-SYMPTOM": 0.8755493668682824, | |
| "eval_recall_I-DISEASE": 0.9203636363636364, | |
| "eval_recall_I-PROCEDURE": 0.903623701605288, | |
| "eval_recall_I-SYMPTOM": 0.89632108276627, | |
| "eval_recall_O": 0.9535418768533188, | |
| "eval_recall_macro": 0.9016821356957543, | |
| "eval_recall_micro": 0.9312907693792016, | |
| "eval_roc_auc_B-DISEASE": 0.9434637984834352, | |
| "eval_roc_auc_B-PROCEDURE": 0.9352564893054347, | |
| "eval_roc_auc_B-SYMPTOM": 0.9361752734886775, | |
| "eval_roc_auc_I-DISEASE": 0.9544283223484076, | |
| "eval_roc_auc_I-PROCEDURE": 0.9469635618065941, | |
| "eval_roc_auc_I-SYMPTOM": 0.9392831339324041, | |
| "eval_roc_auc_O": 0.9443675558832417, | |
| "eval_runtime": 50.8034, | |
| "eval_samples_per_second": 278.524, | |
| "eval_steps_per_second": 34.82, | |
| "step": 344910 | |
| } | |
| ], | |
| "logging_steps": 256, | |
| "max_steps": 344910, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.2940770438369178e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |