Chinese contentWork Agents：系统级框架解析智能体如何可靠完成数字工作
一篇给 Agent 开发者的系统化梳理，把工作智能体拆成模型加 harness 两层，还给了五种工作形态和五大能力族的分析框架，读个大纲就能对照自己项目查漏补缺。
#Agent评估
5 stories · 0 in 7 days · 2 in 30 days
Stories, products and papers tagged "Agent评估", newest first.
Oct 27
Oct 21
SFResearch Study Tests Fragility of Memory-Based Self-Improving Agents
The takeaway is that agent self-improvement may be more fragile than reported scores suggest, because evaluation conditions differ sharply from production task streams. Teams building memory-based agents should run multi-seed evaluations, randomise task order, and enrich memory construction with explicit rubrics and feedback. Code and data are released openly, so practitioners can replicate the findings directly.
Oct 6
Sep 23
Clay evaluates AI agents on quality, throughput and cost
The framework shifts agent evaluation from single accuracy scores to a three-way trade-off, which changes how teams should size models and routing for production workloads. Builders deploying agents in cost-sensitive markets can use LangSmith evals to run the same quality-versus-cost benchmarking on their own pipelines before committing to a provider.
Jul 23
Chinese contentLangChain 推出 agent 评估技能库，支持 Codex 和 Claude Code
想测你的 agent 好不好用？LangChain 这个新技能库让你直接用 Codex 或 Claude Code 跑 eval 任务，还能自动检查验证器准不准。