Metadata-Version: 2.3
Name: wse
Version: 0.0.1
Summary: Weight-Space Engineering (WSE): a reproducible framework for programmatic and agent-driven neural network weight editing. Name placeholder; the first release comes with the open-source release.
Author: Xuguang Ai
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Science/Research
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.11
Description-Content-Type: text/markdown

# Weight-Space Engineering (WSE)

**A reproducible framework for programmatic and agent-driven neural network weight editing.**

This release is a placeholder for the project's name and contains no functionality. The first functional release will be published together with the open-source release of the code.

WSE turns weight-space interventions into reproducible, auditable experiments:

- it inspects a model's weights and internal activations;
- it applies targeted edits as exact patches with before-and-after tensor hashes;
- it evaluates how the model's behavior changes;
- it runs agent-driven experiments under explicit budgets, with every action audited.
