What are the essential factors in crafting effective long context multi-hop instruction datasets? insights and best practices

Z Chen, Q Chen, L Qin, Q Guo, H Lv, Y Zou… - arxiv preprint arxiv …, 2024 - arxiv.org
Recent advancements in large language models (LLMs) with extended context windows
have significantly improved tasks such as information extraction, question answering, and …

Large Model Agents: State-of-the-Art, Cooperation Paradigms, Security and Privacy, and Future Trends

Y Wang, Y Pan, Q Zhao, Y Deng, Z Su, L Du… - arxiv preprint arxiv …, 2024 - arxiv.org
Large Model (LM) agents, powered by large foundation models such as GPT-4 and DALL-E
2, represent a significant step towards achieving Artificial General Intelligence (AGI). LM …

Text2World: Benchmarking Large Language Models for Symbolic World Model Generation

M Hu, T Chen, Y Zou, Y Lei, Q Chen, M Li… - arxiv preprint arxiv …, 2025 - arxiv.org
Recently, there has been growing interest in leveraging large language models (LLMs) to
generate symbolic world models from textual descriptions. Although LLMs have been …

A Cognitive Writing Perspective for Constrained Long-Form Text Generation

K Wan, H Mu, R Hao, H Luo, T Gu, X Chen - arxiv preprint arxiv …, 2025 - arxiv.org
Like humans, Large Language Models (LLMs) struggle to generate high-quality long-form
text that adheres to strict requirements in a single pass. This challenge is unsurprising, as …

On the Structural Memory of LLM Agents

R Zeng, J Fang, S Liu, Z Meng - arxiv preprint arxiv:2412.15266, 2024 - arxiv.org
Memory plays a pivotal role in enabling large language model~(LLM)-based agents to
engage in complex and long-term interactions, such as question answering (QA) and …

" Ghost of the past": identifying and resolving privacy leakage from LLM's memory through proactive user interaction

S Zhang, L Ye, X Yi, J Tang, B Shui, H **ng… - arxiv preprint arxiv …, 2024 - arxiv.org
Memories, encompassing past inputs in context window and retrieval-augmented
generation (RAG), frequently surface during human-LLM interactions, yet users are often …