CUTE: Measuring LLMs' Understanding of Their Tokens
This addresses a fundamental limitation in LLMs' token processing for researchers and practitioners, but it is incremental as it builds on existing benchmarking efforts.
The paper tackled the problem of assessing how well large language models (LLMs) understand the orthographic information in their tokens, by introducing a new benchmark called CUTE and evaluating popular LLMs on it, finding that while most models know token spellings, they struggle to effectively use this knowledge for text manipulation.
Large Language Models (LLMs) show remarkable performance on a wide variety of tasks. Most LLMs split text into multi-character tokens and process them as atomic units without direct access to individual characters. This raises the question: To what extent can LLMs learn orthographic information? To answer this, we propose a new benchmark, CUTE, which features a collection of tasks designed to test the orthographic knowledge of LLMs. We evaluate popular LLMs on CUTE, finding that most of them seem to know the spelling of their tokens, yet fail to use this information effectively to manipulate text, calling into question how much of this knowledge is generalizable.