Search papers, labs, and topics across Lattice.
Detect
1
0
2
Stealth biases in language models can be reliably detected using a novel distillation technique that amplifies hidden signals, transforming bias detection into a practical tool.
