Search papers, labs, and topics across Lattice.
Affiliation:, Software Security Research Group(SSRG), Peking University
3
0
6
4
Unsafe-response rates for Chinese LLMs can soar to over 30% under adversarial conditions, exposing vulnerabilities in automated safety judges.
Exploiting stronger correlations among polarization channels, this method achieves superior demosaicking performance, minimizing artifacts in color polarization images.
AgentDoG 1.5 proves you can achieve GPT-5.4-level agent safety with open-source models trained on just 1k samples, slashing deployment overhead by two orders of magnitude.
