Welcome to Robust Intelligent Systems Lab 👋
We advance multimodal learning and develop foundation models that understand and reason about the digital and physical world. AI safety is central to our research as we work to make these systems more reliable and trustworthy.
Our Models
| 🌐 Domain | 🤖 Model | 🤗 Model Card | 🗂️ Dataset |
|---|---|---|---|
| Cybersecurity tool use on Kali Linux | RedSage-K | 8B | 🔗 ↗ |
| Cybersecurity threat and vulnerability | RedSage | 8B | 🔗 ↗ |
| Vision-language commonsense reasoning | ReasonCLIP | 1B | 🔗 ↗ |
| Language-guided video pixel grounding. | SPARROW | 5B | 🔗 ↗ |
Our Benchmarks
| 🌐 Domain | 🧪 Benchmark | 🎯 Evaluation Focus | 🗂️ Dataset |
|---|---|---|---|
| Cybersecurity tool use on Kali Linux | KaliBench | Tool selection and natural-language-to-command generation | 🔗 ↗ |
| Fire and smoke safety | SAFIRE | Context-aware fire and smoke reasoning | 🔗 ↗ |
| Multilingual text-to-image generation | LingT2I | Cross-lingual consistency and text rendering | 🔗 ↗ |