SwordBench benchmarks steering methods for concept removal in vision models and shows that linear SVMs achieve strong separability and orthogonality but incur collateral damage, while sparse autoencoders often perform better and no method reaches perfect steering even in simple cases.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
Fine-tuned AI text detectors amplify a pretrained typicality axis instead of learning an AI-human boundary, with raw centroid projections achieving 86-106% of fine-tuned AUROC and a 24-example frozen probe matching full training.
citing papers explorer
-
SwordBench: Evaluating Orthogonality of Steering Image Representations
SwordBench benchmarks steering methods for concept removal in vision models and shows that linear SVMs achieve strong separability and orthogonality but incur collateral damage, while sparse autoencoders often perform better and no method reaches perfect steering even in simple cases.
-
Amplifying, Not Learning: Fine-Tuned AI Text Detectors Amplify a Pretrained Direction
Fine-tuned AI text detectors amplify a pretrained typicality axis instead of learning an AI-human boundary, with raw centroid projections achieving 86-106% of fine-tuned AUROC and a 24-example frozen probe matching full training.