Synergistic Multi-Agent Framework with Trajectory Learning for Knowledge-Intensive Tasks
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909445126094848 |
|---|---|
| author | Yue, Shengbin Wang, Siyuan Chen, Wei Huang, Xuanjing Wei, Zhongyu |
| author_facet | Yue, Shengbin Wang, Siyuan Chen, Wei Huang, Xuanjing Wei, Zhongyu |
| contents | Recent advancements in Large Language Models (LLMs) have led to significant breakthroughs in various natural language processing tasks. However, generating factually consistent responses in knowledge-intensive scenarios remains a challenge due to issues such as hallucination, difficulty in acquiring long-tailed knowledge, and limited memory expansion. This paper introduces SMART, a novel multi-agent framework that leverages external knowledge to enhance the interpretability and factual consistency of LLM-generated responses. SMART comprises four specialized agents, each performing a specific sub-trajectory action to navigate complex knowledge-intensive tasks. We propose a multi-agent co-training paradigm, Long-Short Trajectory Learning, which ensures synergistic collaboration among agents while maintaining fine-grained execution by each agent. Extensive experiments on five knowledge-intensive tasks demonstrate SMART's superior performance compared to widely adopted knowledge internalization and knowledge enhancement methods. Our framework can extend beyond knowledge-intensive tasks to more complex scenarios. Our code is available at https://github.com/yueshengbin/SMART. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2407_09893 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Synergistic Multi-Agent Framework with Trajectory Learning for Knowledge-Intensive Tasks Yue, Shengbin Wang, Siyuan Chen, Wei Huang, Xuanjing Wei, Zhongyu Computation and Language Recent advancements in Large Language Models (LLMs) have led to significant breakthroughs in various natural language processing tasks. However, generating factually consistent responses in knowledge-intensive scenarios remains a challenge due to issues such as hallucination, difficulty in acquiring long-tailed knowledge, and limited memory expansion. This paper introduces SMART, a novel multi-agent framework that leverages external knowledge to enhance the interpretability and factual consistency of LLM-generated responses. SMART comprises four specialized agents, each performing a specific sub-trajectory action to navigate complex knowledge-intensive tasks. We propose a multi-agent co-training paradigm, Long-Short Trajectory Learning, which ensures synergistic collaboration among agents while maintaining fine-grained execution by each agent. Extensive experiments on five knowledge-intensive tasks demonstrate SMART's superior performance compared to widely adopted knowledge internalization and knowledge enhancement methods. Our framework can extend beyond knowledge-intensive tasks to more complex scenarios. Our code is available at https://github.com/yueshengbin/SMART. |
| title | Synergistic Multi-Agent Framework with Trajectory Learning for Knowledge-Intensive Tasks |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2407.09893 |