Search papers, labs, and topics across Lattice.
3
0
3
0
Machine translation benchmarks have functionally saturated, but pairing human-authored failure cases with deterministic verification rules reveals critical multimodal blind spots that automated metrics consistently miss.
Luxembourgish NLU is finally getting the attention it deserves with the launch of ltzGLUE, revealing critical insights into model performance and language capabilities.
Fine-tuning a 27B parameter model on a relatively small, low-resource language like Luxembourgish yields surprisingly strong translation performance, even generalizing to unseen languages.
