Baran

Text Splitter for Large Language Model datasets.

To avoid token constraints and improve the accuracy of vector search in the Large Language Model, it is necessary to divide the document. This gem supports splitting the text in the specified manner.

Install the gem and add to the application's Gemfile by executing:

$ bundle add baran

If bundler is not being used to manage dependencies, install the gem by executing:

$ gem install baran

• chunk_size: 1024 (characters)
• chunk_overlap: 64 (characters)

Splitting by the specified character.

splitter = Baran::CharacterTextSplitter.new(
    chunk_size: 1024,
    chunk_overlap: 64,
    separator: "\n\n"
)
splitter.chunks(text, metadata: { ... })
# => [{ cursor: 0, text: "...", metadata: { ... } }, ...]

Splitting by the specified characters recursively.

splitter = Baran::RecursiveCharacterTextSplitter.new(
    separators: ["\n\n", "\n", " ", ""]
)
splitter.chunks(text, metadata: { ... })
# => [{ cursor: 0, text: "...", metadata: { ... } }, ...]

Splitting by the Markdown descriptions.

splitter = Baran::MarkdownSplitter.new
splitter.chunks(markdown, metadata: { ... })
# => [{ cursor: 0, text: "...", metadata: { ... } }, ...]

Split with the following priority.

[
    "\n# ", # h1
    "\n## ", # h2
    "\n### ", # h3
    "\n#### ", # h4
    "\n##### ", # h5
    "\n###### ", # h6
    "```\n\n", # code block
    "\n\n***\n\n", # horizontal rule
    "\n\n---\n\n", # horizontal rule
    "\n\n___\n\n", # horizontal rule
    "\n\n", # new line
    "\n", # new line
    " ", # space
    "" # empty
]

After checking out the repo, run bin/setup to install dependencies. Then, run bundle exec rake to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

Bug reports and pull requests are welcome on GitHub at https://github.com/kawakamimoeki/baran. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the code of conduct.

The gem is available as open source under the terms of the MIT License.

Everyone interacting in the Baran project's codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.