Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capabilities
This addresses the problem of high pre-training costs for non-English languages, enabling more efficient cross-lingual adaptation for Japanese NLP applications, though it is incremental as it builds on existing methods.
The authors tackled adapting English-trained large language models to Japanese by extending vocabulary and continual pre-training, resulting in Swallow, which achieved superior performance on Japanese tasks, with performance increasing up to 100B tokens of training data.
Cross-lingual continual pre-training of large language models (LLMs) initially trained on English corpus allows us to leverage the vast amount of English language resources and reduce the pre-training cost. In this study, we constructed Swallow, an LLM with enhanced Japanese capability, by extending the vocabulary of Llama 2 to include Japanese characters and conducting continual pre-training on a large Japanese web corpus. Experimental results confirmed that the performance on Japanese tasks drastically improved through continual pre-training, and the performance monotonically increased with the amount of training data up to 100B tokens. Consequently, Swallow achieved superior performance compared to other LLMs that were trained from scratch in English and Japanese. An analysis of the effects of continual pre-training revealed that it was particularly effective for Japanese question answering tasks. Furthermore, to elucidate effective methodologies for cross-lingual continual pre-training from English to Japanese, we investigated the impact of vocabulary expansion and the effectiveness of incorporating parallel corpora. The results showed that the efficiency gained through vocabulary expansion had no negative impact on performance, except for the summarization task, and that the combined use of parallel corpora enhanced translation ability.