Safety Research
Popular repositories Loading
-
persona_vectors
persona_vectors PublicPersona Vectors: Monitoring and Controlling Character Traits in Language Models
-
-
-
assistant-axis
assistant-axis PublicThe Assistant Axis is a direction in activation space that captures how "Assistant-like" a model's behavior is. Models can drift away from the Assistant during conversations—sometimes toward bizarr…
-
safety-tooling
safety-tooling PublicInference API for many LLMs and other useful tools for empirical research
Repositories
- thimble Public
- subliminal-jlens Public
- toy-subliminal-opd Public
- impossiblebench Public
Official Inspect Implementation for "ImpossibleBench: Measuring LLMs' Propensity of Exploiting Test Cases"
- auditing-agents Public
- jpp_lens Public
Code for blog post `J++ Lens: Jacobian Filtering Enables More Faithful Workspace Lenses`
- bird Public
BIRD: a benchmark for iterative reward design (code for "A Bird's-Eye View of Iterative Reward Design")
- failure-disclosure Public
Code and data for the paper "When Do Models Admit They Are Wrong? Failure Disclosure Is Unstable Under Reinforcement Learning"
- red-teaming-auto-mode Public
Most used topics
Loading…