Unverified50% confidenceConceptExact time
SWE-Marathon is an end-to-end outcome-based evaluation that gives the agent a software project spanning multiple hours and observes whether it can stay on course without drifting over a long horizon.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLong-horizon coding benchmarks (such as SWE-bench) evaluate a model's ability to handle repository-level tasks, including reading codebases, locating bugs, making cross-file modifications, and passing tests.69% similarVerifiedIn multi-step agentic workflows, the final result is formed by a chain of intermediate decisions, and the same task may have multiple reasonable execution paths, making evaluation more difficult than traditional software testing.66% similarUnverified传统软件监控工具关注延迟、错误率、吞吐量等指标，但对AI智能体的语义层面质量存在盲区65% similarVerifiedMany agent frameworks are deliberately designed to isolate final evaluation data from the optimization loop, so agents can only see training and validation signals and cannot directly access the test set65% similarUnverified鸿蒙自动化测试是典型长链路任务，涉及跨领域知识融合、强依赖性和反馈循环，链路从读取需求到测试用例设计、脚本执行、输出测试报告65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1011807API
curl https://kongchang.com/api/v1/knowledge/claims/1011807MCP
get_claim(id=1011807)