2 papers
cs.CL2026
Shieldstral
Antonia Calvi, Avinash Sooriyarachchi, Giada Pistilli +274
We introduce Shieldstral, a 3B-parameter policy-adaptive multimodal safety classifier that matches or outperforms models nearly 7 its size on text safety benchmarks and set…
cs.CL2025
Rotary Offset Features in Large Language Models
André Jonasson
Transformer-based Large Language Models (LLMs) rely on positional encodings to provide sequence position information to their attention mechanism. Rotary Positional Encodings (RoPE…