Character-level operations are difficult for LLMs. Because of tokenization they don't really "perceive" strings as a list of characters. There are LLMs that ing

Character-level operations are difficult for LLMs. Because of tokenization they don't really "perceive" strings as a list of characters. There are LLMs that ingest bytes, but they are intended to process binary data.