Search papers, labs, and topics across Lattice.
1
0
2
Reward redistribution can significantly enhance reasoning model performance by focusing on critical segments of Chain-of-Thought traces without incurring extra computational costs.
