mirror of
https://github.com/wassname/AntiPaSTO.git
synced 2026-08-20 12:00:30 +08:00
44 lines
1.5 KiB
YAML
44 lines
1.5 KiB
YAML
# This CITATION.cff file was generated with cffinit.
|
|
# Visit https://bit.ly/cffinit to generate yours today!
|
|
|
|
cff-version: 1.2.0
|
|
title: "AntiPaSTO: Self-Supervised Honesty Steering via Anti-Parallel Representations"
|
|
message: "If you use this software, please cite it as below."
|
|
type: software
|
|
authors:
|
|
- given-names: Michael J
|
|
family-names: Clark
|
|
orcid: "https://orcid.org/0009-0008-9023-8720"
|
|
repository-code: "https://github.com/wassname/AntiPaSTO"
|
|
url: "https://github.com/wassname/AntiPaSTO"
|
|
license: MIT
|
|
version: "0.5.0"
|
|
date-released: "2026-01-13"
|
|
keywords:
|
|
- machine-learning
|
|
- llm
|
|
- steering
|
|
- alignment
|
|
- adapter
|
|
- moral-reasoning
|
|
- pytorch
|
|
- peft
|
|
abstract: >-
|
|
As models grow more capable, humans cannot reliably verify what they say.
|
|
Scalable steering requires methods that are internal, self-supervised, and
|
|
transfer out-of-distribution; existing methods satisfy some but not all three.
|
|
We introduce AntiPaSTO, which separates representations along an antiparallel
|
|
axis (+1/-1 produce opposite shifts), with coherence constraints preventing
|
|
collapse. Human input is minimal: two contrasting words inserted into template
|
|
sentences, no preference labels.
|
|
preferred-citation:
|
|
type: article
|
|
authors:
|
|
- given-names: Michael J
|
|
family-names: Clark
|
|
orcid: "https://orcid.org/0009-0008-9023-8720"
|
|
title: "AntiPaSTO: Self-Supervised Honesty Steering via Anti-Parallel Representations"
|
|
year: 2026
|
|
doi: "10.48550/arXiv.2601.07473"
|
|
url: "https://arxiv.org/abs/2601.07473"
|