Understanding the Evaluation Code

[Vincent950129]

Thanks for the nice code.

I am confused with the evaluation code in runner.py L570. In particular, why "if j==0: continue" and what is variable "temp" for? It would be nice if someone can walk me through.

       for i, label in enumerate(label_ids):
            temp_1 = []
            temp_2 = []
            for j,m in enumerate(label):
                if j == 0:
                    continue
                elif label_ids[i][j] == len(label_map):
                    y_true.append(temp_1)
                    y_pred.append(temp_2)
                    break
                else:
                    temp_1.append(label_map[label_ids[i][j]])
                    temp_2.append(label_map[logits[i][j]])

Also, I got the report file as follows, which seems incorrect (the extra "SEP]"):

          precision    recall  f1-score   support

     LOC     0.9213    0.9329    0.9270      1668
    MISC     0.7756    0.8319    0.8027       702
     ORG     0.8867    0.9001    0.8933      1661
     PER     0.9521    0.9586    0.9553      1617
    SEP]     0.0000    0.0000    0.0000         0

micro avg 0.9005 0.9180 0.9092 5648 macro avg 0.7071 0.7247 0.7157 5648 weighted avg 0.9018 0.9180 0.9098 5648

Any help would be highly appreciated. Thank you for your time.

[terminator123]

i met the same prob, have u solved it ?