SwordBench B
AI benchmark · Vision
What is SwordBench?
SwordBench evaluates methods for steering image representations of vision models in concept removal tasks, measuring separability, orthogonality, cross-concept robustness and collateral damage.
Media mostly
“we introduce SwordBench, a benchmark for steering image representations of vision models across multiple backbones and concept removal tasks.”
arxiv.org/abs/2605.16372v1 ↗General capability
“Steering or intervening on model representations at inference time to correct predictions is essential for AI interpretability and safety”
arxiv.org/abs/2605.16372v1 ↗
Paper: arxiv.org/abs/2605.16372v1 ↗
Tier B
Established: 15 or more models scored, or curated RL directories rate it.
How much stands behind a tier: usage, who backs it, a confirmed line from its own site or paper, and a clear read on X. High with three of the four, medium with two, low otherwise.
- 18 models scored
- in curated RL directories (1 pt)
How we got to tier B
- Adoption: 18 models scored. Tier B.
- On X: unclear, none buzz No move.
- In curated RL directories: 1 point, a floor of none.
- Confirmed: a line from its own site or paper places it.
A clear view on X moves a tier one step up or down; most entries don't move. How tiers work.
Where SwordBench sits
Its domain on the map, live. Hover a logo for its name; click to open it. Open it on the full map
Next: browse all benchmarks. Entry from the RL Research daily scrape of public sources, 2026-11-14 snapshot.