to use F1 score: when classes are imbalanced and both FP and FN matter

mixed precision training does: forward in FP16, accumulate gradients in FP32

Mixed precision training: forward in FP16, accumulate gradients in FP32

Phi coefficient

Matthews correlation coefficient (MCC) measures balanced metric even with class imbalance

Precision and recall

Precision = Relevant retrieved instances / All retrieved instances

to use AUC-ROC: comparing classifiers across all thresholds

Use AUC-ROC to compare classifiers' performance across all thresholds

to use an RNN/LSTM: for sequential data where order matters (mostly replaced by transformers)

Use RNN/LSTM for sequential data where order matters (mostly replaced by transformers)

NDCG measures: ranking quality with graded relevance scores

NDCG measures ranking quality with graded relevance scores