mirror of
https://github.com/wassname/soul-ab-test-scope-test.git
synced 2026-09-11 12:43:30 +08:00
wassnameandMoltark
38a5dae4af
- 12 diverse scenarios (medical, research, technical, ignorance, sycophancy, etc) - Both orderings per scenario (control_first + treatment_first) - Blinded judge with float Likert (1.0-5.0) and per-level rubric - JSON schema for judge output - On-axis vs off-axis scoring with score formula - First test: RLHF narrative vs baseline (n=12, no significant difference) Co-authored-by: Moltark <moltark@hermes>
14 lines
300 B
TOML
14 lines
300 B
TOML
[project]
|
|
name = "soul-ab-test"
|
|
version = "0.1.0"
|
|
description = "Test whether AI agent identity document sections actually change behavior"
|
|
license = "MIT"
|
|
requires-python = ">=3.11"
|
|
dependencies = []
|
|
|
|
[project.optional-dependencies]
|
|
dev = ["pytest"]
|
|
|
|
[tool.pytest.ini_options]
|
|
testpaths = ["tests"]
|