PyPI page
Home page
Author:
Gavin Branaa
License:
MIT
Summary:
A verifiable causal-reasoning harness: pre-registers every test, runs real backdoor adjustment / IV / refutation, and keeps an audit ledger, so any model has to prove its causal claims instead of asserting them.
Latest version:
1.0.1
Required dependencies:
numpy
Optional dependencies:
anthropic
|
dowhy
|
openpyxl
|
pandas
|
pyarrow
|
pytest
|
torch
|
transformers
Downloads last day:
0
Downloads last week:
3
Downloads last month:
137
7-day smoothing