/wordcel

Swiss army-knife for composing LLM outputs

Primary LanguagePythonMIT LicenseMIT

😶 Wordcel

wordcel is a library of functions that provides a set of common tools for working with large language models.

Candidly, it is mostly a set of functions that I myself use on a regular basis — my own personal Swiss army knife.

Installation

You can simply pip install wordcel.

Documentation

  • LLM APIs: Wrapper functions over the most common LLM APIs.
  • RAG: Helper functions for RAG, and a minimal implementation of Anthropic's "Contextual Retrieval" method.
  • featurize: Helper functions for multithreaded inference over text columns in pandas DataFrames.
  • DAG: WordcelDAG is a flexible and extensible framework for defining and executing Directed Acyclic Graphs (DAGs) of data processing tasks, particularly involving LLMs and dataframes.

There is also a nascent CLI. wordcel --help:

Usage: wordcel [OPTIONS] COMMAND [ARGS]...

  Wordcel CLI.

Options:
  --help  Show this message and exit.

Commands:
  dag  WordcelDAG commands.