Building Instruction-Tuning Datasets from Human-Written Instructions with Open-Weight Large Language Models
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866909736294678528 |
|---|---|
| author | Ma, Youmi Mizuki, Sakae Fujii, Kazuki Nakamura, Taishi Ohi, Masanari Shimada, Hinari Shiotani, Taihei Saito, Koshiro Maeda, Koki Hattori, Kakeru Okamoto, Takumi Ishida, Shigeki Yokota, Rio Takamura, Hiroya Okazaki, Naoaki |
| author_facet | Ma, Youmi Mizuki, Sakae Fujii, Kazuki Nakamura, Taishi Ohi, Masanari Shimada, Hinari Shiotani, Taihei Saito, Koshiro Maeda, Koki Hattori, Kakeru Okamoto, Takumi Ishida, Shigeki Yokota, Rio Takamura, Hiroya Okazaki, Naoaki |
| contents | Instruction tuning is crucial for enabling Large Language Models (LLMs) to solve real-world tasks. Prior work has shown the effectiveness of instruction-tuning data synthesized solely from LLMs, raising a fundamental question: Do we still need human-originated signals for instruction tuning? This work answers the question affirmatively: we build state-of-the-art instruction-tuning datasets sourced from human-written instructions, by simply pairing them with LLM-generated responses. LLMs fine-tuned on our datasets consistently outperform those fine-tuned on existing ones. Our data construction approach can be easily adapted to other languages; we build datasets for Japanese and confirm that LLMs tuned with our data reach state-of-the-art performance. Analyses suggest that instruction-tuning in a new language allows LLMs to follow instructions, while the tuned models exhibit a notable lack of culture-specific knowledge in that language. The datasets and fine-tuned models will be publicly available. Our datasets, synthesized with open-weight LLMs, are openly distributed under permissive licenses, allowing for diverse use cases. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_23714 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Building Instruction-Tuning Datasets from Human-Written Instructions with Open-Weight Large Language Models Ma, Youmi Mizuki, Sakae Fujii, Kazuki Nakamura, Taishi Ohi, Masanari Shimada, Hinari Shiotani, Taihei Saito, Koshiro Maeda, Koki Hattori, Kakeru Okamoto, Takumi Ishida, Shigeki Yokota, Rio Takamura, Hiroya Okazaki, Naoaki Computation and Language Instruction tuning is crucial for enabling Large Language Models (LLMs) to solve real-world tasks. Prior work has shown the effectiveness of instruction-tuning data synthesized solely from LLMs, raising a fundamental question: Do we still need human-originated signals for instruction tuning? This work answers the question affirmatively: we build state-of-the-art instruction-tuning datasets sourced from human-written instructions, by simply pairing them with LLM-generated responses. LLMs fine-tuned on our datasets consistently outperform those fine-tuned on existing ones. Our data construction approach can be easily adapted to other languages; we build datasets for Japanese and confirm that LLMs tuned with our data reach state-of-the-art performance. Analyses suggest that instruction-tuning in a new language allows LLMs to follow instructions, while the tuned models exhibit a notable lack of culture-specific knowledge in that language. The datasets and fine-tuned models will be publicly available. Our datasets, synthesized with open-weight LLMs, are openly distributed under permissive licenses, allowing for diverse use cases. |
| title | Building Instruction-Tuning Datasets from Human-Written Instructions with Open-Weight Large Language Models |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2503.23714 |