A benchmark dataset for evaluating long-term memory capabilities in conversational AI systems. It contains multi-turn group dialogues spanning approximately 250 days per topic, organized by date and chat group. The dataset was created by EverMind-AI and last updated on 2026-02-25.
Use Cases
- Benchmarking long-term memory performance in AI agents based on multi-turn dialogues.
- Training models to track conversation context and participant roles over extended time periods.
- Evaluating AI's ability to recall and reference past information from dialogues spanning ~250 days.
Strengths
- Dialogues span approximately 250 days per topic, providing a long temporal context.
- Data is organized by date and chat group, suggesting a structured temporal sequence.
- Includes at least five distinct topics, indicated by topic identifiers 01-05.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Source
- EverMind-AI
- Time Range
- Dialogues span ~250 days per topic.
- Freshness
- Last updated 2026-02-25 02:46:12; freshness should be verified.