Search papers, labs, and topics across Lattice.
1
0
2
0
TUDUM reveals that fine-tuning for Turkish reasoning can enhance linguistic alignment but may compromise benchmark accuracy, challenging assumptions about model performance in multilingual settings.
