Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
1
0
3
2
LLMs can reason better without training by selectively using "soft" reasoning steps only when they're unsure, preventing overconfidence from derailing the process.