Search papers, labs, and topics across Lattice.
1
0
3
5
LLMs struggle to extract and reason over complex text structures in scientific domains, but explicitly prompting them to build these structures first yields significant performance gains.
