Towards Human Understanding of Paraphrase Types in Large Language Models
This work addresses the limited interpretability in paraphrase evaluation for language models, offering a granular dataset to improve specific linguistic capabilities, though it is incremental in scope.
The study assessed ChatGPT's ability to generate paraphrases across ten atomic types, revealing it excels at simple changes like additions but struggles with complex structures such as subordination, using a new dataset of 800 annotations.
Paraphrases represent a human's intuitive ability to understand expressions presented in various different ways. Current paraphrase evaluations of language models primarily use binary approaches, offering limited interpretability of specific text changes. Atomic paraphrase types (APT) decompose paraphrases into different linguistic changes and offer a granular view of the flexibility in linguistic expression (e.g., a shift in syntax or vocabulary used). In this study, we assess the human preferences towards ChatGPT in generating English paraphrases with ten APTs and five prompting techniques. We introduce APTY (Atomic Paraphrase TYpes), a dataset of 800 sentence-level and word-level annotations by 15 annotators. The dataset also provides a human preference ranking of paraphrases with different types that can be used to fine-tune models with RLHF and DPO methods. Our results reveal that ChatGPT and a DPO-trained LLama 7B model can generate simple APTs, such as additions and deletions, but struggle with complex structures (e.g., subordination changes). This study contributes to understanding which aspects of paraphrasing language models have already succeeded at understanding and what remains elusive. In addition, we show how our curated datasets can be used to develop language models with specific linguistic capabilities.