Search papers, labs, and topics across Lattice.
1
0
3
8
LLM coding agents still fall short when optimizing real-world codebases, especially when balancing multiple objectives like performance and correctness, as revealed by the new FormulaCode benchmark.
