AngelinaZhai / epai-sentiment-of-tweets

1 stars 0 forks source link

Backend: Try out new loss functions #33

Open AngelinaZhai opened 1 year ago

AngelinaZhai commented 1 year ago

Evaluate and compare the model on L1, BCEwithLogits, and macro F1 score for train, val and test accuracies. My suspicion is that the current macro F1 is too coarse for a detailed accuracy evaluation.