Search papers, labs, and topics across Lattice.
5
0
6
Falsification, rather than mere exposure, proves crucial in enhancing the self-repair capabilities of frozen code models, leading to a substantial increase in successful program repairs.
Semantic post-hoc operators fail to enhance accuracy in frozen small code models, but a novel recovery method boosts performance by 12 tasks on HumanEval+.
The Popperian prompt skill fails to deliver measurable improvements in code correctness beyond a simple scaffold, challenging assumptions about the efficacy of scientific reasoning in LLMs.
HJB-inspired residual selection slashes UAV path-tracking error by nearly 87%, revealing a new approach to command supervision that outperforms traditional autopilots in turbulent conditions.
LLMs can learn to safely leverage external memory for code debugging by explicitly modeling and penalizing the risk of false-positive memory injection.
