Search papers, labs, and topics across Lattice.
3
0
3
3
Frontier model performance is now within reach for a wider array of institutions, thanks to a novel Continual Learning approach that minimizes forgetting while maximizing capability.
Frontier LLMs struggle with contract scrubbing, achieving only modest recall rates despite their prowess in general benchmarks.
No current LLM can accurately identify missing legal information in user queries, with all evaluated models struggling to balance responses to both deficient and complete questions.
