Search papers, labs, and topics across Lattice.
University of Maryland, Baltimore County
2
0
6
LLMs can learn to internalize and apply complex business rules during chain-of-thought reasoning, outperforming in-context learning while using significantly shorter prompts.
Current cross-lingual models can disastrously invert sentiment when translating between English and Bengali, with one model flipping positive intent to negative nearly 30% of the time.
