Multi-Agent Collaboration for Multilingual Code Instruction Tuning
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912229009391616 |
|---|---|
| author | Yang, Jian Zhang, Wei Yang, Jiaxi Miao, Yibo Quan, Shanghaoran Wu, Zhenhe Peng, Qiyao Yang, Liqun Liu, Tianyu Cui, Zeyu Hui, Binyuan Lin, Junyang |
| author_facet | Yang, Jian Zhang, Wei Yang, Jiaxi Miao, Yibo Quan, Shanghaoran Wu, Zhenhe Peng, Qiyao Yang, Liqun Liu, Tianyu Cui, Zeyu Hui, Binyuan Lin, Junyang |
| contents | Recent advancement in code understanding and generation demonstrates that code LLMs fine-tuned on a high-quality instruction dataset can gain powerful capabilities to address wide-ranging code-related tasks. However, most previous existing methods mainly view each programming language in isolation and ignore the knowledge transfer among different programming languages. To bridge the gap among different programming languages, we introduce a novel multi-agent collaboration framework to enhance multilingual instruction tuning for code LLMs, where multiple language-specific intelligent agent components with generation memory work together to transfer knowledge from one language to another efficiently and effectively. Specifically, we first generate the language-specific instruction data from the code snippets and then provide the generated data as the seed data for language-specific agents. Multiple language-specific agents discuss and collaborate to formulate a new instruction and its corresponding solution (A new programming language or existing programming language), To further encourage the cross-lingual transfer, each agent stores its generation history as memory and then summarizes its merits and faults. Finally, the high-quality multilingual instruction data is used to encourage knowledge transfer among different programming languages to train Qwen2.5-xCoder. Experimental results on multilingual programming benchmarks demonstrate the superior performance of Qwen2.5-xCoder in sharing common knowledge, highlighting its potential to reduce the cross-lingual gap. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_07487 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Multi-Agent Collaboration for Multilingual Code Instruction Tuning Yang, Jian Zhang, Wei Yang, Jiaxi Miao, Yibo Quan, Shanghaoran Wu, Zhenhe Peng, Qiyao Yang, Liqun Liu, Tianyu Cui, Zeyu Hui, Binyuan Lin, Junyang Computation and Language Recent advancement in code understanding and generation demonstrates that code LLMs fine-tuned on a high-quality instruction dataset can gain powerful capabilities to address wide-ranging code-related tasks. However, most previous existing methods mainly view each programming language in isolation and ignore the knowledge transfer among different programming languages. To bridge the gap among different programming languages, we introduce a novel multi-agent collaboration framework to enhance multilingual instruction tuning for code LLMs, where multiple language-specific intelligent agent components with generation memory work together to transfer knowledge from one language to another efficiently and effectively. Specifically, we first generate the language-specific instruction data from the code snippets and then provide the generated data as the seed data for language-specific agents. Multiple language-specific agents discuss and collaborate to formulate a new instruction and its corresponding solution (A new programming language or existing programming language), To further encourage the cross-lingual transfer, each agent stores its generation history as memory and then summarizes its merits and faults. Finally, the high-quality multilingual instruction data is used to encourage knowledge transfer among different programming languages to train Qwen2.5-xCoder. Experimental results on multilingual programming benchmarks demonstrate the superior performance of Qwen2.5-xCoder in sharing common knowledge, highlighting its potential to reduce the cross-lingual gap. |
| title | Multi-Agent Collaboration for Multilingual Code Instruction Tuning |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2502.07487 |