Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capabilities
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910426313261056 |
|---|---|
| author | Fujii, Kazuki Nakamura, Taishi Loem, Mengsay Iida, Hiroki Ohi, Masanari Hattori, Kakeru Shota, Hirai Mizuki, Sakae Yokota, Rio Okazaki, Naoaki |
| author_facet | Fujii, Kazuki Nakamura, Taishi Loem, Mengsay Iida, Hiroki Ohi, Masanari Hattori, Kakeru Shota, Hirai Mizuki, Sakae Yokota, Rio Okazaki, Naoaki |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_17790 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capabilities Fujii, Kazuki Nakamura, Taishi Loem, Mengsay Iida, Hiroki Ohi, Masanari Hattori, Kakeru Shota, Hirai Mizuki, Sakae Yokota, Rio Okazaki, Naoaki Computation and Language Artificial Intelligence 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. |
| title | Continual Pre-Training for Cross-Lingual LLM Adaptation: Enhancing Japanese Language Capabilities |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2404.17790 |