Note Start by reading the 🚀Primus Paper! To the best of our knowledge, we are the 🔥 first to release datasets covering cybersecurity pretraining, IFT, and reasoning distillation. Of course, we are also the first to pretrain an LLM on a large-scale cybersecurity corpus.

trendmicro-ailab/Llama-Primus-Base

Text Generation • Updated 1 day ago • 8 • 4

Note Based on Llama-3.1-8B-Instruct, continually pretrained on 2.77B tokens of cybersecurity text, achieving a 🚀15.88% improvement in the aggregated score across multiple cybersecurity benchmarks.

trendmicro-ailab/Llama-Primus-Merged

Text Generation • Updated 1 day ago • 121 • 7

Note Instruct Model! While maintaining nearly the same instruction-following capability as Llama-3.1-8B-Instruct, achieving a 🚀14.84% improvement across multiple cybersecurity benchmarks.

trendmicro-ailab/Llama-Primus-Reasoning

Text Generation • Updated 3 days ago • 8 • 1

Note Distilled on reasoning and reflection data from o1-preview for cybersecurity tasks, achieving a 🚀10% improvement on CISSP.

Note Includes high-quality cybersecurity texts manually collected from reputable sources such as wikipedia, MITRE, cybersecurity company websites, CTI, and more.

Note Includes 2.57B tokens of cybersecurity texts filtered from FineWeb.

Note Includes approximately 1K QA pairs covering common cybersecurity business scenarios.

Note Includes reasoning and reflection data generated by o1-preview on cybersecurity tasks for distillation.