NexusRaven 13B

About

NexusRaven 13B is an advanced large language model developed by Nexusflow, featuring 13 billion parameters and designed primarily for function calling tasks. It is built on the Llama architecture and fine-tuned from the CodeLlama-13b-Instruct model, enhancing its capabilities in programming and data processing. The model excels with a 95% success rate in cybersecurity tool tasks, outperforming GPT-4, and can handle complex function calls, including nested and parallel ones. It supports various quantization methods, optimizing it for different hardware configurations. Notably, NexusRaven is trained on publicly available data, ensuring commercial viability, fast inference speeds, and strong generalization abilities for diverse applications.

Capabilities