Exploring Automatic Evaluation Methods based on a Decoder-based LLM for Text Generation
This work addresses the problem of evaluating text generation for researchers and practitioners, but it is incremental as it compares existing methods without introducing new techniques.
The paper investigated automatic evaluation methods for text generation using decoder-based large language models, comparing them to encoder-based models on machine translation and semantic textual similarity tasks in Japanese and English. Results showed that tuned decoder-based models performed poorly, focusing on surface sequences and failing to capture meaning, with in-context learning in very large models like ChatGPT struggling to identify fine-grained semantic differences.
Automatic evaluation of text generation is essential for improving the accuracy of generation tasks. In light of the current trend towards increasingly larger decoder-based language models, we investigate automatic evaluation methods based on such models for text generation. This paper compares various methods, including tuning with encoder-based models and large language models under equal conditions, on two different tasks, machine translation evaluation and semantic textual similarity, in two languages, Japanese and English. Experimental results show that compared to the tuned encoder-based models, the tuned decoder-based models perform poorly. The analysis of the causes for this suggests that the decoder-based models focus on surface word sequences and do not capture meaning. It is also revealed that in-context learning of very large decoder-based models such as ChatGPT makes it difficult to identify fine-grained semantic differences.