#merge-rules

A BPE tokenizer learned these merges, in this order, from a corpus of "low", "lowest", "newer", "wider" and "new":

1. e r → er
2. n e → ne
3. ne w → new
4. l o → lo
5. lo w → low
6. new er → newer

It tokenizes a new word by splitting it into characters and applying the merges in the order they were learned. How is the unseen word "lowered" tokenized?