Search papers, labs, and topics across Lattice.
2
0
3
0
LLMs that ace code generation often fail to grasp intended program semantics, as evidenced by a stark performance decline when generating executable behavioral specifications on the new CodeSpecBench benchmark.
Rethinking IRSTD as a centroid regression problem with single-point supervision achieves competitive detection performance with significantly reduced computational cost, challenging the dominance of pixel-level segmentation approaches.
