SWE-Gate Benchmark for Software Agents
SWE-Gate addresses critical evaluation gaps in coding agents, which currently only test functional correctness but ignore real-world review constraints that determine patch acceptance.
SWE-Gate addresses critical evaluation gaps in coding agents, which currently only test functional correctness but ignore real-world review constraints that determine patch acceptance.