five

AmanPriyanshu/reasoning-sft-Edge-Agent-Reasoning-WebSearch-260K

收藏
Hugging Face2026-03-10 更新2026-03-29 收录
下载链接:
https://hf-mirror.com/datasets/AmanPriyanshu/reasoning-sft-Edge-Agent-Reasoning-WebSearch-260K
下载链接
链接失效反馈
官方服务:
资源简介:
--- license: mit task_categories: - text-generation language: - en tags: - reasoning - sft - chain-of-thought - planning - agentic - self-awareness - web-search size_categories: - 100K<n<1M --- # Edge-Agent Reasoning (Converted) Converted version of [yatin-superintelligence/Edge-Agent-Reasoning-WebSearch-260K](https://huggingface.co/datasets/yatin-superintelligence/Edge-Agent-Reasoning-WebSearch-260K), reformatted for reasoning SFT training with structured think blocks. ## Format Each row has three columns: - **`input`** — list of dicts with system and user messages. System prompt instructs the model to deconstruct problems thoroughly in a `<think>` block, audit its own knowledge, identify ambiguities, and respond with "Planning completed." - **`response`** — `<think>` block containing the full 2,000-5,000 word reasoning trajectory followed by "Planning completed." - **`domain`** — industry sector (e.g. Cosmology, Mechanical Engineering, Healthcare, Finance, etc.) ## Reasoning Structure Each response contains a 5-stage reasoning trajectory: 1. **Understanding the request** — full problem decomposition with constraint identification 2. **What I believe I know — and what I'm uncertain about** — self-aware knowledge auditing with confidence levels 3. **Ambiguities in the request** — identifying missing parameters and conflicting constraints 4. **Everything I need to confirm before responding** — explicit verification checklist 5. **Web search queries** — 10-20 expert-level, keyword-dense queries for RAG retrieval ## Coverage - 200+ professional roles across 200+ industries - 39 operating system environments (macOS, Windows, Linux, Android, iPadOS, Cloud Shells) - Tasks spanning from low difficulty to impossible, safe to catastrophic risk ## License MIT — inherited from the [original dataset](https://huggingface.co/datasets/yatin-superintelligence/Edge-Agent-Reasoning-WebSearch-260K). ## Credits Original dataset by [Yatin Taneja](https://huggingface.co/yatin-superintelligence).
提供机构:
AmanPriyanshu
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作