dreeseaw/mdlens-combined-markdown-v1
收藏资源简介：
该数据集支持`mdlens` v1 Markdown检索评估。它完全是一个Markdown问答/搜索评估数据集，并非广泛的编码代理基准，尽管Markdown问答是编码代理工作的常见部分。数据集内容包括：1,783个Markdown文件（约17.0 MB源文本）、30个锁定的难题问题、源语料库计数和字节总数的清单、聚合评估摘要以及每次运行的报告。语料库结合了三种类型的材料：精心策划的生成/场景Markdown（包含格式错误、过时笔记、复制干扰项、表格和多个关键点）、类似SciCat的科学README代理（包含Hugging Face和GitHub科学Markdown备用材料）以及来自真实仓库文档、运行手册、设计笔记和实验报告的代码库文档。30个问题中有五个是类似工作流的跨语料库分析任务，但无需代码编辑。
This dataset supports the `mdlens` v1 Markdown retrieval evaluation. It is entirely a Markdown QA/search eval. It is not a broad coding-agent benchmark, even though Markdown QA is a common part of coding-agent work. The contents include: 1,783 Markdown files (about 17.0 MB of source text), 30 locked hard questions, a manifest with source corpus counts and byte totals, aggregate eval summaries, and per-run reports. The corpus combines three fixture families: carefully curated messy generated/scene Markdown with malformed formatting, stale notes, copied distractors, tables, and multiple needles; a SciCat-style scientific README proxy with Hugging Face and GitHub scientific Markdown fallback material; and codebase documentation fixtures from real repository docs, runbooks, design notes, and experiment reports. Five of the 30 questions are workflow-like cross-corpus analysis tasks, but no question requires code edits.
