acc_best = 0.745398773006135
num_best = 326
correct_best = 243
prec_0_best = 0.8007662835249042
rec_0_best = 0.9372197309417041
f1_0_best = 0.8636363636363638
prec_1_best = 0.34782608695652173
rec_1_best = 0.26666666666666666
f1_1_best = 0.30188679245283023
prec_2_best = 0.6190476190476191
rec_2_best = 0.3561643835616438
f1_2_best = 0.45217391304347826
macro_f1_best = 0.5392323563775574
acc_MaxLen256 = 0.7147239263803681
num_MaxLen256 = 326
correct_MaxLen256 = 233
prec_0_MaxLen256 = 0.7672727272727272
rec_0_MaxLen256 = 0.9461883408071748
f1_0_MaxLen256 = 0.8473895582329316
prec_1_MaxLen256 = 0.20833333333333334
rec_1_MaxLen256 = 0.16666666666666666
f1_1_MaxLen256 = 0.1851851851851852
prec_2_MaxLen256 = 0.6296296296296297
rec_2_MaxLen256 = 0.2328767123287671
f1_2_MaxLen256 = 0.34
macro_f1_MaxLen256 = 0.4575249144727056
Best checkpoint is ./outputs/oversample/comp/roberta-large-LR1e-5-epoch10-MaxLen256/checkpoint-best, best accuracy is 0.745398773006135