Search papers, labs, and topics across Lattice.
3
0
4
7
Segments with conflicting quality signals are systematically undervalued, revealing critical flaws in current evaluation metrics for multilingual tasks.
A multilingual dataset of synthetic dialogs reveals critical insights into personal information detection across diverse contexts and languages.
Instruction tuning can reduce masculine bias in decoder-only MT models, but these models still don't consistently outperform encoder-decoder architectures on gender-specific translation tasks.
