Search papers, labs, and topics across Lattice.
1
0
3
A new 30B open-weight LLM trained on 34 European languages achieves state-of-the-art performance on low-resource languages with significantly less compute, proving that clever training beats brute force.
