word error rate (WER) measures: edit distance between predicted and reference transcriptions

BLEU vs ROUGE: BLEU measures precision of n-grams, ROUGE measures recall

BLEU measures precision of n-grams, ROUGE measures recall

the vocabulary size matters: larger vocab = shorter sequences but more parameters

Larger vocab reduces sequence length, increasing model complexity and parameters

Brier score

Brier score measures mean squared error of probability predictions

WordPiece tokenization does: similar to BPE but uses likelihood instead of frequency

WordPiece tokenization splits words into subwords based on token likelihood rather than frequency

log-loss / cross-entropy loss penalizes: confident wrong predictions more heavily

Log-loss penalizes confident incorrect predictions more heavily

to normalize features: when features have different scales and you use distance-based methods

Normalize features when they have different scales for distance-based methods