Search papers, labs, and topics across Lattice.
1
0
3
Guaranteeing trivially parseable LLM outputs via suffix-constrained generation not only simplifies information extraction but can also improve performance on prediction tasks.
