我“恨”论文

通过OpenAI轻松生成论文的易读版本

我经常需要阅读一篇论文来提供相关主题的背景信息。在这种情况下，论文的技术深度可能是一个主要障碍。因此，我创建了“我恨论文”，以生成易于消化的学术研究成果。

目前支持

• arXiv论文ID
• 本地.tex文件
• 本地.md文件
• 本地.html文件（实验性）

安装

pip install i-hate-papers

示例使用

# First set your OpenAI API key
❱ export OPENAI_API_KEY=...

# Summarise a arXiv paper ID
❱ i_hate_papers 2106.09685

# Summarise a latex file
❱ i_hate_papers path/to/some-paper.tex

# Summarise a html file
❱ i_hate_papers path/to/some-paper.html

示例输出

• 示例HTML（包含使用MathJax渲染的数学公式）
• 示例Markdown

参考

❱ i_hate_papers --help
usage: i_hate_papers [-h] [--verbosity {0,1,2}] [--no-input] [--no-html] [--no-open] [--no-footer] 
                     [--no-glossary] [--detail-level {0,1,2}] [--model MODEL] INPUT

Summarise an academic paper

You must set the OPENAI_API_KEY environment variable using your OpenAi.com API key

positional arguments:
  INPUT                 arXiv paper ID (example: 1234.56789) or path to a .tex/.html/.md file

options:
  -h, --help            show this help message and exit
  --verbosity {0,1,2}   Set the logging verbosity (0 = quiet, 1 = info logging, 2 = debug logging). Default is 1
  --no-input            Don't prompt for file selection, just use the largest tex file
  --no-html             Skip HTML file generation
  --no-open             Don't open the HTML file when complete (macOS only)
  --no-footer           Don't include a footer containing metadata
  --no-glossary         Don't include a glossary
  --detail-level {0,1,2}
                        How detailed should the summary be? (0 = minimal detail, 1 = normal, 2 = more detail)
  --model MODEL         What model to use to generate the summaries

发布流程

内部使用

export VERSION=0.1.1
poetry version $VERSION
git ci -a -m "Releasing version $VERSION"
git tag "v$VERSION"
git push origin main refs/tags/v$VERSION
poetry publish --build