| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.8760766121467434, | |
| "global_step": 110000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.97441713710709e-05, | |
| "loss": 0.0001, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.94883427421418e-05, | |
| "loss": 0.0001, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "eval_stsb_spearman": 0.779150634149958, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9232514113212696e-05, | |
| "loss": 0.0001, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.8976685484283595e-05, | |
| "loss": 0.0001, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "eval_stsb_spearman": 0.7742037737128947, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.8720856855354495e-05, | |
| "loss": 0.0001, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.8465028226425394e-05, | |
| "loss": 0.0001, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_stsb_spearman": 0.7762138392811366, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.8209199597496294e-05, | |
| "loss": 0.0001, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.795337096856719e-05, | |
| "loss": 0.0001, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "eval_stsb_spearman": 0.7749345657870104, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.769754233963809e-05, | |
| "loss": 0.0001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.7441713710708985e-05, | |
| "loss": 0.0001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_stsb_spearman": 0.7707595907216558, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.7185885081779885e-05, | |
| "loss": 0.0001, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.6930056452850784e-05, | |
| "loss": 0.0001, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_stsb_spearman": 0.7749981304786546, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 2.667422782392168e-05, | |
| "loss": 0.0001, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.641839919499258e-05, | |
| "loss": 0.0001, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_stsb_spearman": 0.7738172782882184, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.616257056606348e-05, | |
| "loss": 0.0001, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.590674193713438e-05, | |
| "loss": 0.0001, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "eval_stsb_spearman": 0.7733704832120436, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.565091330820528e-05, | |
| "loss": 0.0001, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.5395084679276174e-05, | |
| "loss": 0.0001, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "eval_stsb_spearman": 0.7775208072571205, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 2.5139256050347074e-05, | |
| "loss": 0.0001, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.4883427421417973e-05, | |
| "loss": 0.0001, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_stsb_spearman": 0.772225753495311, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.4627598792488873e-05, | |
| "loss": 0.0001, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 2.4371770163559772e-05, | |
| "loss": 0.0001, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "eval_stsb_spearman": 0.7727032901042289, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.411594153463067e-05, | |
| "loss": 0.0001, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.3860112905701568e-05, | |
| "loss": 0.0001, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_stsb_spearman": 0.7753943735719158, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 2.3604284276772467e-05, | |
| "loss": 0.0001, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.3348455647843367e-05, | |
| "loss": 0.0001, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "eval_stsb_spearman": 0.7730675807188958, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.3092627018914266e-05, | |
| "loss": 0.0001, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.2836798389985162e-05, | |
| "loss": 0.0001, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "eval_stsb_spearman": 0.7781803089941816, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.2580969761056062e-05, | |
| "loss": 0.0001, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.2325141132126958e-05, | |
| "loss": 0.0001, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_stsb_spearman": 0.7760834463511375, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 2.2069312503197857e-05, | |
| "loss": 0.0001, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.1813483874268757e-05, | |
| "loss": 0.0001, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_stsb_spearman": 0.7766114476167615, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.1557655245339653e-05, | |
| "loss": 0.0001, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.1301826616410553e-05, | |
| "loss": 0.0001, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "eval_stsb_spearman": 0.774197452942763, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.1045997987481452e-05, | |
| "loss": 0.0001, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.079016935855235e-05, | |
| "loss": 0.0001, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "eval_stsb_spearman": 0.7747344951319819, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.053434072962325e-05, | |
| "loss": 0.0001, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.0278512100694147e-05, | |
| "loss": 0.0001, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_stsb_spearman": 0.7734231292337955, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 2.0022683471765047e-05, | |
| "loss": 0.0001, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.9766854842835946e-05, | |
| "loss": 0.0001, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "eval_stsb_spearman": 0.7711219149130829, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.9511026213906845e-05, | |
| "loss": 0.0001, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9255197584977745e-05, | |
| "loss": 0.0001, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "eval_stsb_spearman": 0.7764032943131054, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.8999368956048644e-05, | |
| "loss": 0.0001, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.874354032711954e-05, | |
| "loss": 0.0001, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "eval_stsb_spearman": 0.7728760022631344, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.848771169819044e-05, | |
| "loss": 0.0001, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.823188306926134e-05, | |
| "loss": 0.0001, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "eval_stsb_spearman": 0.775568059169046, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.797605444033224e-05, | |
| "loss": 0.0001, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.772022581140314e-05, | |
| "loss": 0.0001, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "eval_stsb_spearman": 0.7716868251335787, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.746439718247403e-05, | |
| "loss": 0.0001, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.720856855354493e-05, | |
| "loss": 0.0001, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "eval_stsb_spearman": 0.7752166203205634, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.695273992461583e-05, | |
| "loss": 0.0, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.669691129568673e-05, | |
| "loss": 0.0, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "eval_stsb_spearman": 0.7722613478318026, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.644108266675763e-05, | |
| "loss": 0.0, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.6185254037828525e-05, | |
| "loss": 0.0001, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "eval_stsb_spearman": 0.7746778882608947, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.5929425408899425e-05, | |
| "loss": 0.0, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.5673596779970324e-05, | |
| "loss": 0.0, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_stsb_spearman": 0.7744804439612313, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.5417768151041224e-05, | |
| "loss": 0.0, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.5161939522112121e-05, | |
| "loss": 0.0, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_stsb_spearman": 0.7720634386643836, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.4906110893183021e-05, | |
| "loss": 0.0001, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.4650282264253919e-05, | |
| "loss": 0.0, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "eval_stsb_spearman": 0.776464481420812, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.4394453635324818e-05, | |
| "loss": 0.0, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.4138625006395718e-05, | |
| "loss": 0.0, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "eval_stsb_spearman": 0.7731371111367188, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.3882796377466614e-05, | |
| "loss": 0.0, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.3626967748537513e-05, | |
| "loss": 0.0, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_stsb_spearman": 0.7710878430175343, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.3371139119608411e-05, | |
| "loss": 0.0, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.311531049067931e-05, | |
| "loss": 0.0, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "eval_stsb_spearman": 0.7752104766273047, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.285948186175021e-05, | |
| "loss": 0.0, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.2603653232821108e-05, | |
| "loss": 0.0, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "eval_stsb_spearman": 0.7749487602603908, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.2347824603892007e-05, | |
| "loss": 0.0, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.2091995974962905e-05, | |
| "loss": 0.0, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_stsb_spearman": 0.7769257731613575, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.1836167346033804e-05, | |
| "loss": 0.0, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.1580338717104704e-05, | |
| "loss": 0.0, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "eval_stsb_spearman": 0.7756559730014233, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.13245100881756e-05, | |
| "loss": 0.0, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.10686814592465e-05, | |
| "loss": 0.0, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "eval_stsb_spearman": 0.7745848587301726, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.0812852830317397e-05, | |
| "loss": 0.0, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.0557024201388297e-05, | |
| "loss": 0.0, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "eval_stsb_spearman": 0.7721647051923735, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.0301195572459196e-05, | |
| "loss": 0.0, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.0045366943530094e-05, | |
| "loss": 0.0, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "eval_stsb_spearman": 0.7742379627015131, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.789538314600993e-06, | |
| "loss": 0.0, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.533709685671891e-06, | |
| "loss": 0.0, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "eval_stsb_spearman": 0.7729155725915814, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.27788105674279e-06, | |
| "loss": 0.0, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.02205242781369e-06, | |
| "loss": 0.0, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_stsb_spearman": 0.7731093489515521, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.766223798884586e-06, | |
| "loss": 0.0, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.510395169955486e-06, | |
| "loss": 0.0, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "eval_stsb_spearman": 0.776349877698032, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 8.254566541026384e-06, | |
| "loss": 0.0, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.998737912097283e-06, | |
| "loss": 0.0, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "eval_stsb_spearman": 0.7742966440865845, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.742909283168183e-06, | |
| "loss": 0.0, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.48708065423908e-06, | |
| "loss": 0.0, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "eval_stsb_spearman": 0.776391111019895, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.23125202530998e-06, | |
| "loss": 0.0, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.975423396380878e-06, | |
| "loss": 0.0, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_stsb_spearman": 0.7725475338449702, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.719594767451776e-06, | |
| "loss": 0.0, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.463766138522675e-06, | |
| "loss": 0.0, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "eval_stsb_spearman": 0.7748880311631935, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.207937509593574e-06, | |
| "loss": 0.0, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.952108880664473e-06, | |
| "loss": 0.0, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_stsb_spearman": 0.7736924744692966, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.696280251735371e-06, | |
| "loss": 0.0, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5.440451622806269e-06, | |
| "loss": 0.0, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_stsb_spearman": 0.7725938501752172, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5.184622993877168e-06, | |
| "loss": 0.0, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.9287943649480675e-06, | |
| "loss": 0.0, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "eval_stsb_spearman": 0.7746088551214558, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.672965736018966e-06, | |
| "loss": 0.0, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.417137107089864e-06, | |
| "loss": 0.0, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "eval_stsb_spearman": 0.7742809440881452, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.7080654239080384e-05, | |
| "loss": 0.0001, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.695273992461583e-05, | |
| "loss": 0.0026, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "eval_stsb_spearman": 0.7683574925666393, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.682482561015128e-05, | |
| "loss": 0.0023, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.669691129568673e-05, | |
| "loss": 0.002, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "eval_stsb_spearman": 0.7751801343996144, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.656899698122218e-05, | |
| "loss": 0.0021, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.644108266675763e-05, | |
| "loss": 0.002, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_stsb_spearman": 0.7706796437720312, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.631316835229308e-05, | |
| "loss": 0.002, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6185254037828525e-05, | |
| "loss": 0.0018, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "eval_stsb_spearman": 0.7756891246212327, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.6057339723363978e-05, | |
| "loss": 0.0018, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.5929425408899425e-05, | |
| "loss": 0.0017, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "eval_stsb_spearman": 0.7752558955538417, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.5801511094434878e-05, | |
| "loss": 0.002, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.5673596779970324e-05, | |
| "loss": 0.0019, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "eval_stsb_spearman": 0.780701712591289, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.5545682465505774e-05, | |
| "loss": 0.0019, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.5417768151041224e-05, | |
| "loss": 0.0018, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_stsb_spearman": 0.7873629268749875, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.5289853836576673e-05, | |
| "loss": 0.0019, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.5161939522112121e-05, | |
| "loss": 0.0017, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "eval_stsb_spearman": 0.7660054024130715, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.503402520764757e-05, | |
| "loss": 0.0019, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.4906110893183021e-05, | |
| "loss": 0.0017, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "eval_stsb_spearman": 0.7855143130229439, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.4778196578718469e-05, | |
| "loss": 0.0019, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.4650282264253919e-05, | |
| "loss": 0.0017, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "eval_stsb_spearman": 0.77861030088238, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.4522367949789368e-05, | |
| "loss": 0.0016, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.4394453635324818e-05, | |
| "loss": 0.0016, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_stsb_spearman": 0.7812534433697589, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.4266539320860268e-05, | |
| "loss": 0.0017, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.4138625006395718e-05, | |
| "loss": 0.0017, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "eval_stsb_spearman": 0.7736488598869644, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.4010710691931166e-05, | |
| "loss": 0.0018, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.3882796377466614e-05, | |
| "loss": 0.0017, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "eval_stsb_spearman": 0.7776239203767936, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.3754882063002063e-05, | |
| "loss": 0.0015, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.3626967748537513e-05, | |
| "loss": 0.0017, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "eval_stsb_spearman": 0.772728744520677, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.3499053434072963e-05, | |
| "loss": 0.0017, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.3371139119608411e-05, | |
| "loss": 0.0014, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "eval_stsb_spearman": 0.7840145435380373, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.324322480514386e-05, | |
| "loss": 0.0016, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.311531049067931e-05, | |
| "loss": 0.0015, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "eval_stsb_spearman": 0.7814009954533213, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.298739617621476e-05, | |
| "loss": 0.0015, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.285948186175021e-05, | |
| "loss": 0.0015, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "eval_stsb_spearman": 0.7801342379468632, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.2731567547285658e-05, | |
| "loss": 0.0016, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.2603653232821108e-05, | |
| "loss": 0.0015, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "eval_stsb_spearman": 0.778155239600538, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.2475738918356557e-05, | |
| "loss": 0.0015, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.2347824603892007e-05, | |
| "loss": 0.0015, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "eval_stsb_spearman": 0.7727153538621762, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.2219910289427457e-05, | |
| "loss": 0.0015, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.2091995974962905e-05, | |
| "loss": 0.0015, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "eval_stsb_spearman": 0.7839607415226314, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.1964081660498355e-05, | |
| "loss": 0.0014, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.1836167346033804e-05, | |
| "loss": 0.0014, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "eval_stsb_spearman": 0.7737039597499138, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.1708253031569254e-05, | |
| "loss": 0.0015, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.1580338717104704e-05, | |
| "loss": 0.0014, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "eval_stsb_spearman": 0.7717549383604455, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.1452424402640152e-05, | |
| "loss": 0.0014, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.13245100881756e-05, | |
| "loss": 0.0015, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "eval_stsb_spearman": 0.7761051944670604, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.119659577371105e-05, | |
| "loss": 0.0014, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.10686814592465e-05, | |
| "loss": 0.0014, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "eval_stsb_spearman": 0.7814437463546053, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.094076714478195e-05, | |
| "loss": 0.0014, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.0812852830317397e-05, | |
| "loss": 0.0012, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "eval_stsb_spearman": 0.7769911614134192, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.0684938515852847e-05, | |
| "loss": 0.0013, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.0557024201388297e-05, | |
| "loss": 0.0013, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "eval_stsb_spearman": 0.7809750612307073, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.0429109886923746e-05, | |
| "loss": 0.0014, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.0301195572459196e-05, | |
| "loss": 0.0014, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "eval_stsb_spearman": 0.7867736950467644, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.0173281257994644e-05, | |
| "loss": 0.0014, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.0045366943530094e-05, | |
| "loss": 0.0012, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "eval_stsb_spearman": 0.78602360527425, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.917452629065544e-06, | |
| "loss": 0.0014, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.789538314600993e-06, | |
| "loss": 0.0013, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "eval_stsb_spearman": 0.7879566030360313, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 9.661624000136443e-06, | |
| "loss": 0.0013, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 9.533709685671891e-06, | |
| "loss": 0.0012, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "eval_stsb_spearman": 0.7851583038178344, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.405795371207341e-06, | |
| "loss": 0.0013, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.27788105674279e-06, | |
| "loss": 0.0012, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "eval_stsb_spearman": 0.7854350613534488, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 9.14996674227824e-06, | |
| "loss": 0.0013, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 9.02205242781369e-06, | |
| "loss": 0.0012, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "eval_stsb_spearman": 0.7893658570377464, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 8.894138113349137e-06, | |
| "loss": 0.0012, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.766223798884586e-06, | |
| "loss": 0.0013, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "eval_stsb_spearman": 0.7884863161927248, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.638309484420036e-06, | |
| "loss": 0.0012, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 8.510395169955486e-06, | |
| "loss": 0.0011, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "eval_stsb_spearman": 0.7832187972814217, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 8.382480855490936e-06, | |
| "loss": 0.0012, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.254566541026384e-06, | |
| "loss": 0.0012, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_stsb_spearman": 0.7780244483654368, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.126652226561833e-06, | |
| "loss": 0.0012, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 7.998737912097283e-06, | |
| "loss": 0.0011, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "eval_stsb_spearman": 0.7798499707740966, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.870823597632733e-06, | |
| "loss": 0.0011, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.742909283168183e-06, | |
| "loss": 0.0011, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "eval_stsb_spearman": 0.783190494125622, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 7.6149949687036314e-06, | |
| "loss": 0.0011, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.48708065423908e-06, | |
| "loss": 0.0012, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "eval_stsb_spearman": 0.7804977213380432, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 7.35916633977453e-06, | |
| "loss": 0.001, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.23125202530998e-06, | |
| "loss": 0.0011, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "eval_stsb_spearman": 0.7812487997617589, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 7.103337710845429e-06, | |
| "loss": 0.0012, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.975423396380878e-06, | |
| "loss": 0.001, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "eval_stsb_spearman": 0.7816751642438525, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 6.847509081916327e-06, | |
| "loss": 0.001, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 6.719594767451776e-06, | |
| "loss": 0.0011, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "eval_stsb_spearman": 0.7820030275667162, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 6.591680452987226e-06, | |
| "loss": 0.0011, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.463766138522675e-06, | |
| "loss": 0.001, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "eval_stsb_spearman": 0.7832784805352376, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 6.335851824058125e-06, | |
| "loss": 0.001, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.207937509593574e-06, | |
| "loss": 0.0011, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_stsb_spearman": 0.7833093403145232, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.080023195129023e-06, | |
| "loss": 0.001, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 5.952108880664473e-06, | |
| "loss": 0.001, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "eval_stsb_spearman": 0.7829398244489639, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 5.824194566199921e-06, | |
| "loss": 0.0009, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 5.696280251735371e-06, | |
| "loss": 0.001, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "eval_stsb_spearman": 0.7820607890459481, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 5.5683659372708205e-06, | |
| "loss": 0.001, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 5.440451622806269e-06, | |
| "loss": 0.001, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "eval_stsb_spearman": 0.7846063901039803, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 5.312537308341719e-06, | |
| "loss": 0.0009, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 5.184622993877168e-06, | |
| "loss": 0.0009, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "eval_stsb_spearman": 0.7852033791449183, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 5.056708679412618e-06, | |
| "loss": 0.001, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 4.9287943649480675e-06, | |
| "loss": 0.001, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "eval_stsb_spearman": 0.7775529189032535, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 4.800880050483516e-06, | |
| "loss": 0.0009, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 4.672965736018966e-06, | |
| "loss": 0.0009, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "eval_stsb_spearman": 0.7785107169299121, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.545051421554414e-06, | |
| "loss": 0.001, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.417137107089864e-06, | |
| "loss": 0.0009, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "eval_stsb_spearman": 0.7788225966725707, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 4.289222792625314e-06, | |
| "loss": 0.0009, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.1613084781607625e-06, | |
| "loss": 0.0009, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "eval_stsb_spearman": 0.7775520554532559, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 4.033394163696212e-06, | |
| "loss": 0.0009, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.905479849231661e-06, | |
| "loss": 0.0009, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "eval_stsb_spearman": 0.7765839741253101, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.777565534767111e-06, | |
| "loss": 0.0009, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.64965122030256e-06, | |
| "loss": 0.0009, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "eval_stsb_spearman": 0.776046284174713, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.521736905838009e-06, | |
| "loss": 0.0008, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.393822591373459e-06, | |
| "loss": 0.0008, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "eval_stsb_spearman": 0.7734033738249794, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.265908276908908e-06, | |
| "loss": 0.0009, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.1379939624443575e-06, | |
| "loss": 0.0009, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "eval_stsb_spearman": 0.7771743629229094, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.0100796479798064e-06, | |
| "loss": 0.0009, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.882165333515256e-06, | |
| "loss": 0.0008, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_stsb_spearman": 0.7762479721057005, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.7542510190507054e-06, | |
| "loss": 0.0009, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.6263367045861547e-06, | |
| "loss": 0.0008, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "eval_stsb_spearman": 0.7775759659701185, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 2.498422390121604e-06, | |
| "loss": 0.0008, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 2.370508075657053e-06, | |
| "loss": 0.0008, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "eval_stsb_spearman": 0.7744453475475996, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.2425937611925027e-06, | |
| "loss": 0.0009, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.114679446727952e-06, | |
| "loss": 0.0009, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "eval_stsb_spearman": 0.776758060348122, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.9867651322634013e-06, | |
| "loss": 0.0007, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8588508177988504e-06, | |
| "loss": 0.0007, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "eval_stsb_spearman": 0.7779735283874216, | |
| "step": 110000 | |
| } | |
| ], | |
| "max_steps": 117266, | |
| "num_train_epochs": 2, | |
| "total_flos": 0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |