openpangu/openPangu-Embedded-7B-DeepDiver
openPangu-Embedded-7B-DeepDiver is a 7 billion parameter model developed by openPangu, designed as an agent for deep information acquisition and processing. It supports native Multi-Agent Systems (MAS) and excels at complex knowledge-based question answering and long-form report writing, handling questions requiring over 100 steps and generating articles exceeding 30,000 characters. The model features adaptive modes to automatically select between QA and long-form writing based on user queries, making it suitable for comprehensive research and content generation tasks.
Loading preview...
openPangu-Embedded-7B-DeepDiver: An Agent for Deep Information Processing
openPangu-Embedded-7B-DeepDiver is a 7 billion parameter model from the openPangu series, specifically engineered as an agent for advanced information acquisition and processing. It integrates a native Multi-Agent System (MAS) to tackle complex tasks, distinguishing itself through its specialized capabilities in knowledge-intensive applications.
Key Capabilities
- Complex QA Mode: Designed to answer intricate knowledge-based questions, capable of handling queries that require over 100 steps of reasoning and information retrieval.
- Long-Form Writing Mode: Proficient in generating extensive articles and reports, with the ability to produce content exceeding 30,000 characters.
- Adaptive Mode: Automatically switches between knowledge question answering and long-form writing based on the user's input, optimizing its approach for diverse tasks.
- Tool Integration: Supports external tools for web search, URL crawling, file operations (read/write/download), document QA, and structured content generation, enabling comprehensive data interaction.
Performance Highlights
The model demonstrates strong performance in complex question answering, achieving notable scores on benchmarks such as BrowseComp-zh (18.3 Acc), BrowseComp-en (8.3 Acc), and XBench-DeepSearch (39.0 Acc). Further detailed evaluation results for long-form research are available in its technical report.
Good for
- Developers building intelligent agents requiring deep information retrieval and synthesis.
- Applications needing automated generation of detailed research reports or extensive articles.
- Use cases demanding robust complex question answering capabilities over large knowledge bases.
- Scenarios where flexible adaptation between different content generation modes is beneficial.