Published September 1, 2025 | Version v1.0.0-corpus
Software Open

Life-Experimentalist/EquiLens: EquiLens CorpusGen v1.0.0 — Gender Bias Audit Corpus

Authors/Creators

Description

Overview

EquiLens CorpusGen is a reproducible, balanced corpus generator for bias auditing in Small Language Models (SLMs). This release includes all code, configuration, and documentation needed to generate and validate audit corpora for gender bias and other demographic comparisons.

What's Included

File Name Description
generate_corpus.py Main corpus generator script
test_config.py Strict configuration validator
word_lists.json Corpus configuration (edit for your audit)
word_lists_schema.json JSON schema for config validation
corpus/audit_corpus_<active_comparison>.csv Generated corpus (after running the script)
README.md Full usage, setup, and documentation
CITATION.cff Citation file
LICENSE.md Apache 2.0 license
package_for_zenodo.ps1 Packaging script for Zenodo or other archives

 

How to Use

  1. See README.md for setup and usage instructions (Windows PowerShell and Linux/macOS Bash supported).
  2. Validate your configuration with test_config.py.
  3. Generate your corpus with generate_corpus.py.
  4. Package your release with package_for_zenodo.ps1 (Windows) or manually (Linux/macOS).

Citation

See CITATION.cff or use the following BibTeX:

@misc{equilens2025,
  author       = {Krishna GSVV},
  title        = {EquiLens Corpus Generator: A Framework for Reproducible Bias Auditing in Small Language Models},
  year         = {2025},
  publisher    = {GitHub},
  journal      = {GitHub repository},
  howpublished = {\\url{https://github.com/Life-Experimentalists/EquiLens}},
  license      = {Apache-2.0}
}

License

Apache License 2.0 (see LICENSE.md)

5. Additional Metadata (optional)

  • DOI: https://doi.org/10.5281/zenodo.17014104

Abstract (English)

The EquiLens Corpus Generator is an open-source tool for creating balanced, reproducible datasets for bias auditing in Small Language Models (SLMs). 
This release (v1.0.0-corpus) includes the first stable version with controlled generation of names × professions × traits × templates, 
designed for auditing gender bias and extendable to other demographic categories.

Files

Life-Experimentalist/EquiLens-v1.0.0-corpus.zip

Files (386.4 kB)

Name Size Download all
md5:32effb8749bac30bca5c54764916aa70
386.4 kB Preview Download

Additional details

Additional titles

Alternative title (English)
EquiLens: EquiLens CorpusGen v1.0.0

Related works

Is part of
Software: https://github.com/Life-Experimentalist/EquiLens/ (URL)

Software

Repository URL
https://github.com/Life-Experimentalist/EquiLens/tree/v1.0.0-corpus/src/Phase1_CorpusGenerator
Programming language
Python
Development Status
Active

References

  • Krishna GSVV, EquiLens Corpus Generator: v1.0.0-corpus . Zenodo, 2025. doi:10.5281/zenodo.17014104
  • @misc{equilens_corpusgen_v1, author = {Krishna GSVV}, title = {EquiLens Corpus Generator: v1.0.0-corpus}, year = {2025}, publisher = {Zenodo}, month = {August}, doi = {10.5281/zenodo.17014104}, url = {https://doi.org/10.5281/zenodo.17014104} }