LongAgent: Scaling Language Models to 128k Context through Multi-Agent Collaboration
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_ | 1866916158164172800 |
|---|---|
| author | Zhao, Jun Zu, Can Xu, Hao Lu, Yi He, Wei Ding, Yiwen Gui, Tao Zhang, Qi Huang, Xuanjing |
| author_facet | Zhao, Jun Zu, Can Xu, Hao Lu, Yi He, Wei Ding, Yiwen Gui, Tao Zhang, Qi Huang, Xuanjing |
| contents | Large language models (LLMs) have demonstrated impressive performance in understanding language and executing complex reasoning tasks. However, LLMs with long context windows have been notorious for their expensive training costs and high inference latency. Even the most advanced models such as GPT-4 and Claude2 often make mistakes when processing inputs of over $100k$ tokens, a phenomenon also known as \textit{lost in the middle}. In this paper, we propose \textsc{LongAgent}, a method based on multi-agent collaboration, which scales LLMs (e.g., LLaMA) to a context of 128K and demonstrates potential superiority in long-text processing compared to GPT-4. In \textsc{LongAgent}, a leader is responsible for understanding user intent and directing team members to acquire information from documents. Due to members' hallucinations, it is non-trivial for a leader to obtain accurate information from the responses of dozens to hundreds of members. To address this, we develop an \textit{inter-member communication} mechanism to resolve response conflicts caused by hallucinations through information sharing. Our experimental results indicate that \textsc{LongAgent} offers a promising alternative for long-text processing. The agent team instantiated with LLaMA-7B achieves significant improvements in tasks such as 128k-long text retrieval, multi-hop question answering, compared to GPT-4. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_11550 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | LongAgent: Scaling Language Models to 128k Context through Multi-Agent Collaboration Zhao, Jun Zu, Can Xu, Hao Lu, Yi He, Wei Ding, Yiwen Gui, Tao Zhang, Qi Huang, Xuanjing Computation and Language Artificial Intelligence Large language models (LLMs) have demonstrated impressive performance in understanding language and executing complex reasoning tasks. However, LLMs with long context windows have been notorious for their expensive training costs and high inference latency. Even the most advanced models such as GPT-4 and Claude2 often make mistakes when processing inputs of over $100k$ tokens, a phenomenon also known as \textit{lost in the middle}. In this paper, we propose \textsc{LongAgent}, a method based on multi-agent collaboration, which scales LLMs (e.g., LLaMA) to a context of 128K and demonstrates potential superiority in long-text processing compared to GPT-4. In \textsc{LongAgent}, a leader is responsible for understanding user intent and directing team members to acquire information from documents. Due to members' hallucinations, it is non-trivial for a leader to obtain accurate information from the responses of dozens to hundreds of members. To address this, we develop an \textit{inter-member communication} mechanism to resolve response conflicts caused by hallucinations through information sharing. Our experimental results indicate that \textsc{LongAgent} offers a promising alternative for long-text processing. The agent team instantiated with LLaMA-7B achieves significant improvements in tasks such as 128k-long text retrieval, multi-hop question answering, compared to GPT-4. |
| title | LongAgent: Scaling Language Models to 128k Context through Multi-Agent Collaboration |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2402.11550 |