subword
Part of word
A subword is a unit smaller than a word but larger than a single character, used especially in computational linguistics and machine learning.
Text processing systems break words into subwords to handle unfamiliar vocabulary: "unhappiness" might become "un," "happi," and "ness." This allows language models to work with words they haven't seen before by recombining familiar pieces.
Smallest meaningful language unit
- affixWord-part modifier
- prefixRoot-modifying front element
- suffixWord-ending addition
- rootUnderground absorbing structure
stem · base · infix · bound · free · derivation · inflection · compound
A portion or segment
- partPiece of whole
- segmentDistinct part of whole
- portionDivided share given
- sectionDistinct divided part
component · element · unit · subdivision · chunk · particle · bit · division
Converting input into discrete units
- segmentationDivision into sections
- embeddingPlacing within surrounding mass
- vocabularyOne's word collection
- encodingConverting data to format
bigram · token · neural · transformer · BERT · GPT · splitting · algorithm
morpheme
Smallest meaningful language unit
root · stem · base · infix · bound · free · derivation · inflection · compound · morphology
piece
A portion or segment
section · component · element · unit · subdivision · chunk · particle · bit · division · fraction
tokenization
Converting input into discrete units
- segmentationDivision into sections
- embeddingPlacing within surrounding mass
- vocabularyOne's word collection
encoding · bigram · token · neural · transformer · BERT · GPT · splitting · algorithm · preprocessing