split,accuracy,precision,recall,f1,loss validation,0.9142,0.9138,0.9142,0.9134,0.2341 test,0.8874,0.8869,0.8874,0.8861,0.3102 ood_test,0.7612,0.7598,0.7612,0.7589,0.5874 hard_test,0.7023,0.7011,0.7023,0.6994,0.7213