Training text
low low low low low lower lower newest newest newest newest newest newest widest widest widest
Word to tokenize
Merge
Reset
Top pairs
“
” as tokens (
)
Learned merges