Search papers, labs, and topics across Lattice.
6
0
9
0
CalibDCD reveals that post-training shifts can significantly degrade data contamination detection, but targeted calibration can restore accuracy.
SymCA achieves a remarkable balance between accuracy and interpretability in column annotation, outperforming traditional methods by over 6% in F1 scores.
ATCCaps reveals that effective call-sign recognition in ATC communications hinges not just on quantity, but on the quality and accuracy of audio-text supervision.
LLMs are now 16% more jailbreakable thanks to a novel attack strategy that evolves a structured memory of successful exploits.
LLMs can achieve state-of-the-art unsupervised multimodal entity linking by reasoning over diverse evidence types, including graph-based neighborhood information.
Neuro-symbolic memory lets multimodal agents beat purely neural memory systems by up to 12.5% on constrained reasoning tasks.
