Search papers, labs, and topics across Lattice.
3
0
4
2
LLMs can debug code more effectively by reasoning about intermediate program states, not just relying on coarse pass/fail test signals.
LLMs can fix more bugs when guided by a semantic evolutionary algorithm that recombines successful partial fixes across a population of candidate repairs.
LLMs can generate more accurate and complete postconditions by engaging in iterative refinement guided by correctness feedback, rather than relying on single-pass prompting.
