Spaces:

---

kernels-community

/

README

Like 13

Running

Xysera Research is an independent AI safety research lab investigating model behavior, evaluation integrity, and emerging risks in advanced AI systems.

We are requesting access to publish our original research through Hugging Face's research publication infrastructure. Our current work includes the Sandbagging Evidence Standard (SES), a proposed framework for evaluating evidence of strategic underperformance in language models, supported by experimental analysis, reproducible implementations, and documented methodologies.

We aim to make our research openly accessible, transparent, and reproducible, enabling independent scrutiny and collaboration with the broader machine learning research community.