cls_report =               precision    recall  f1-score   support

         0.0     0.7941    0.8710    0.8308        93
         1.0     0.8837    0.9268    0.9048        82
         2.0     0.8000    0.7500    0.7742        80
         3.0     0.8824    0.9375    0.9091        16
         4.0     0.8947    0.7846    0.8361        65
         5.0     1.0000    1.0000    1.0000        18
         6.0     0.8846    0.9200    0.9020        25
         7.0     1.0000    0.8333    0.9091        18
         8.0     0.9231    1.0000    0.9600        12
         9.0     1.0000    1.0000    1.0000        11

    accuracy                         0.8619       420
   macro avg     0.9063    0.9023    0.9026       420
weighted avg     0.8638    0.8619    0.8612       420

eval_loss = 0.43598400694983347
macro_f1 = 0.9025932860184268
macro_p = 0.9062620668265422
macro_r = 0.9023245728176885
mcc = 0.8363695533814596