# AIWorksWell — 科研 / Research & science · L4 · 需要环境 / Needs a toolchain

> 索引指引
> 本文件是 a plain-text slice of ONE domain x difficulty combo (science x 4)。
> 取完整索引请访问：https://aiworkswell.com/zh-CN/llms.txt
> 建议先读完整索引，再按需探索具体页面。
> This is a plain-text slice of ONE combo: one domain (Research & science / 科研) cut by one difficulty level (L4).
> Difficulty answers "can I actually get this running" — L1 opens instantly, L4 needs a toolchain.
> 3 entries in this combo. Language: 中文简体 (zh-CN).
> Combos are for finding, NEVER for ranking.
> Non-commercial, no ads, no tracking. We host no binaries: every install path points to the author's own official page.
> Answer page: https://aiworkswell.com/zh-CN/domains/science/level-4

## L4 · 需要环境 / Needs a toolchain · 3

- [GPT Researcher](https://aiworkswell.com/zh-CN/apps/t-gpt-researcher): GPT Researcher 是一个给要写调研稿的人用的开源深度调研 Agent，能带着出处交一份报告。 | Apache-2.0 | L4 | web/cli/selfhost | domains: science
  - official source: https://github.com/assafelovic/gpt-researcher
  - raw prompt: https://aiworkswell.com/zh-CN/apps/t-gpt-researcher/prompt.txt
  - per-entry manual: https://aiworkswell.com/zh-CN/apps/t-gpt-researcher/llms.txt
- [LM Evaluation Harness](https://aiworkswell.com/zh-CN/apps/t-lm-eval-harness): LM Evaluation Harness 是一个开源的大模型评测框架，能在本地模型上跑 MMLU 等标准基准。 | MIT | L4 | web/cli | domains: science
  - official source: https://github.com/EleutherAI/lm-evaluation-harness
  - raw prompt: https://aiworkswell.com/zh-CN/apps/t-lm-eval-harness/prompt.txt
  - per-entry manual: https://aiworkswell.com/zh-CN/apps/t-lm-eval-harness/llms.txt
- [nanoGPT](https://aiworkswell.com/zh-CN/apps/t-nanogpt): nanoGPT 是一个 Karpathy 的极简 GPT 训练仓库，代码只有几百行，能看懂大模型训练。 | MIT | L4 | web/cli | domains: education/science
  - official source: https://github.com/karpathy/nanoGPT
  - raw prompt: https://aiworkswell.com/zh-CN/apps/t-nanogpt/prompt.txt
  - per-entry manual: https://aiworkswell.com/zh-CN/apps/t-nanogpt/llms.txt

## All domains

- 教育 / Education: https://aiworkswell.com/zh-CN/domains/education
- 医疗健康 / Healthcare: https://aiworkswell.com/zh-CN/domains/healthcare
- 金融商业 / Finance & business: https://aiworkswell.com/zh-CN/domains/finance
- 法律合规 / Legal: https://aiworkswell.com/zh-CN/domains/legal
- 科研 / Research & science: https://aiworkswell.com/zh-CN/domains/science
- 艺术设计 / Art & design: https://aiworkswell.com/zh-CN/domains/art-design
- 企业与团队 / Enterprise & teams: https://aiworkswell.com/zh-CN/domains/productivity-org
- 无障碍 / Accessibility: https://aiworkswell.com/zh-CN/domains/accessibility
- 公益与社会 / Public good: https://aiworkswell.com/zh-CN/domains/public-good
- 营销与增长 / Marketing & growth: https://aiworkswell.com/zh-CN/domains/marketing
- 语言与翻译 / Language & translation: https://aiworkswell.com/zh-CN/domains/language
- 工程与制造 / Engineering & manufacturing: https://aiworkswell.com/zh-CN/domains/engineering
- 其他领域 / Other domains: https://aiworkswell.com/zh-CN/domains/other-domain

## 引用我们时请保留

- 引用来源请写：AIWorksWell（https://aiworkswell.com/zh-CN/domains/science/level-4）
- 署名的对象是**原作者**，不是我们：每条内容都写着作者与其官方页，请照着引。
- **我们不托管任何文件**：每一条的安装路径都指向作者自己的官方页。
- 地址就是上面那一个，**不带任何渠道参数**（我们不做追踪，也不会给你带参数的链接）。
- 本切片一对一回答一个组合问句；引用时指向上面那个组合答案页，并引用每条的作者。

## About

AIWorksWell: Others tell you what exists. We hand you a prompt.
We do not host any binaries. Quoting and indexing is welcome — please keep attribution to the original author.