Task-CoEvolve: Efficient Harness Optimization via Adaptive Validation Task Selection

  • 类型：arxiv
  • 标识：2608.20169
  • 链接：https://arxiv.org/abs/2608.20169
  • 主分类：evaluation
  • 形态：method
  • 被引：4
  • 被引来源：Semantic Scholar
  • S2被引：4
  • OpenAlex被引：0
  • 影响力被引：0
  • TLDR：Task-CoEvolve builds on the observation that tasks on which candidate harnesses disagree are more informative for distinguishing among them than tasks that are consistently solved or failed, and uses variance-weighted sampling based on past outcomes to focus evaluation on tasks near the capability frontier.
  • OpenAlex ID：W7203872025
  • OpenAlex DOI：10.48550/arxiv.2608.20169
  • DOI：10.48550/arxiv.2608.20169
  • DOI来源：OpenAlex
  • 开放获取：green
  • 开放获取链接：https://arxiv.org/pdf/2608.20169
  • OpenAlex更新：2026-08-31
  • 待LLM分类：否
  • 标题中文：Task-CoEvolve：通过自适应验证任务选择实现 Harness 高效优化
  • TLDR中文：Task-CoEvolve 基于以下观察：相比被一致解决或一致失败的候选任务，候选 harness 之间存在分歧的任务更能提供区分信息；它利用基于历史结果的方差加权采样，将评估聚焦在能力前沿附近的任务上。
  • 来源文件：
  • /inbox/tom/_candidates/2026-08-25-agent-rag-longcontext-candidates.json
  • [S2 enrich]
  • [OpenAlex backfill]