Search papers, labs, and topics across Lattice.
1
0
3
Latvian NLP gets a boost: a new 111M parameter model outperforms larger multilingual baselines, proving that targeted pretraining still matters for low-resource languages.
