Search papers, labs, and topics across Lattice.
1
0
3
4
Forget scaling up data volume: repeating a smaller, high-quality German dataset yields superior language models compared to single-pass training on a larger, less filtered corpus.
